Thanks, Jerry

Yes, we're in touch with Ben and with others at Cobasys  (Ben Ovshinsky has
been very supportive of PHEVs, and recently contributed a "history of
PHEVs" to the CalCars website at http://www.calcars.org/history.html ), and
we do expect to explore sourcing from Cobasys as one of the most attractive
vendors.

While general discussions of vendors and evaluations of quality of products
are highly appropriate topics for postings, given the complexities of the
battery supply industry, I would suggest that contact names and
granular-level discussions about different companies' products, especially
pricing and availability, be sent privately to Ron and/or to me, rather
than posted to the list. We also very much welcome comments either publicly
or privately about Ron's battety spreadsheet, which makes clear which
batteries are most favorably under consideration.

At 11:52 PM 1/3/2005, Jerry Pohorsky wrote:

>Hi Ron,
>
>I think most of what you proposed is very reasonable.  If it is possible
>to get modules from Cobasys, I would go that route since you would have
>much fewer interconnections to worry about.
>
>By the way, do you know what type of coolant do they use for their liquid
>cooled modules?
>
>At my company we do cooling with oil on one of our systems and ethylene
>glycol on another one.  I think oil is messier to work with.
>
>I would avoid having the coolant come in contact with the Anderson
>connectors, if possible.  On the othere hand, I do have a layer of grease
>on the Anderson connectors in my EV  and they seem to have enough contact
>pressure to displace the grease for a decent metal to metal connection.
>
>Charge balancing is a good thing, to and would be easier to manage on a
>Cobasys module basis rather than having to deal with hundreds of D
>cells.   By the way, have you talked with anyone at Cobasys?
>
>I have spoken with Ben Ovshinsky, (son of the inventor of NiMH technology)
>who was involved with the ECD marketing back in the days before Texaco
>Chevron.  I still have his card somewhere.  I think he used to also teach
>classes at UC Berkeley, so he may still be in the area.
>
>Adios,
>
>Jerry

--- In priusplus@yahoogroups.com, Ron Gremban <rgremban@c...> wrote:
> I realize I haven't reported on this before.
>
> Note that my PRIUS+ has so far been far more limited than planned:
> 1.  I had only a few days of full PHEV operations before continued
> troubleshooting of the CAN bus noise precluded PHEV operation for
a while.
> 2.  The PbA batteries have half the EV range we are shooting for
and are
> so soft (high internal resistance) when at low SOC (state-of-
charge)
> that EV-only mode can be maintained only with difficulty.
> 3.  The SOC reported to the CAN bus must be tweaked by hand
whenever my
> driving mode and/or the SOC changes significantly.
> 4.  EV-only mode must be entered manually, then manually re-
entered
> whenever exited due to speed, acceleration, or low voltage.
> 5.  Maximum regenerative braking has been an all-or-nothing
proposition,
> dependent on reported SOC, and often too little or too much for
the
> battery pack.
>
> Despite these significant limitations, the experience of driving
my
> Prius as a PHEV has been wonderful!  Though I am not a
conservative
> driver, I found myself voluntarily keeping my low-speed
acceleration
> below the EV-only cutoff threshold to remain in EV-only mode as
much as
> possible.  It just feels cool to drive quietly, all-electrically,
> without the ICE running -- and after a while it is annoying to
have the
> ICE start before extra acceleration or speed is really needed.

this has the important side benifit of reducing the acceleration of
all the conventional ice cars stacked up behind you, thus saving
them and us gasoline as well.  as a long time diesel rabbit driver i
can advise you that you will start driving with your rearview mirror
to avoid being hit by the over accelerators.

I have also found your technique useful in the 02 prius, by limiting
the accel to stay in ev mode (often by using the drivable breakdown
lanes on many highways to let the impatient pass) i can stay in ev
mode to the next stop light where i catch them.
i'm not sure this is the best pr for hybrids, but it exercises the
battery more which we know from experience is overdesigned. (the car
will rust out before the battery dies)
>
> It is also cool psychologically to see instantaneous and 5-minute
> reported MPGs of anywhere from 80 to 99.9!  At this rate, I will
expect
> over 800 miles between fillups!

or perhaps the gas tank could be reduced to half the size for a cost
and weight saving or you could keep it less than half full for your
normal driving, reserving a fillup for long trips (although your
bladder will want you to stop sooner).
>
> Most days I have needed to drive less than 20 miles in a stretch.
When
> I come home between trips (as well as at night), I put the car on
charge
> -- which takes all of 30 seconds -- and it is likely to be fully
> recharged before the next trip, as recharge time is now 3 hours or
less
> (note that a full recharge is totally unnecessary for continued
driving;
> also, a battery with more capacity will take longer to fully
charge
> while providing for more EV range between charges).  I am always
> disappointed when the battery is empty and the car begins to
operate
> "merely" as an ordinary hybrid.

opportunity charging for evs has always proven useful.  overnight
charging for full recharge  has always been the way to go.  in the
long run, some solar panels will be right on the vehicle.  in our
annual american tourdesol rally

  http://www.nesea.org/transportation/tour/

, one of the better performers is a ford ranger ev with an 8 foot by
20 foot tiltable array.  during the regular year it gets an average
of 10+ miles per day directly from the sun. we would love to see
your plugin hybrid either entered or displayed at this year's event
in mid may in ny.
>
> Due to the current state of CAN-bus noise debugging, my Prius is
> currently running as an ordinary hybrid.  My experience of this is
> similar to that of going back to driving an ordinary non-hybrid
after
> driving the Prius:  it feels unsophisticated -- the ICE seems to
run too
> much, my mileage seems way too low, and I am irritated at having
to
> visit gas stations so often.
>
> +++++++++++++++++++++++++++++++++++++++++
>
>         Ron Gremban, rgremban@C... <mailto:rgremban@C...>
> Moderator & Technical Lead, PRIUS+ PHEV Conversion Group
>     http://groups.yahoo.com/group/priusplus
> <http://groups.yahoo.com/group/priusplus/>
>         http://www.priusplus.org
> +++++++++++++++++++++++++++++++++++++++++

Felix and I believe it makes sense to bifurcate our search for an NiMH
battery pack, as it is already going in two separate directions -- and
pursue both directions at once.  To this end, I have split our
preliminary NiMH spec sheet into two separate specifications.  If we end
up with two packs, we will put them in two vehicles and compare the results.

One direction is toward a custom or semi-custom pack from a company like
Cobasys or Saft that sells few if any batteries to the retail market.
This direction may produce the most optimum pack, but may take months of
working through contacts and corporate politics to get there.  At that
point, CalCars may get a substantial discount for a pack for a prototype
vehicle, but will we also be able to work out reasonable pricing for
possible future PHEV conversions?

The other direction is to design a pack around custom NiMH D-cell
modules.  I specifically mention D-cells, as good ones are in high
volume production, they are (just barely) large enough to be effective,
and all the larger high-volume cells we have found have poorer power
characteristics and higher relative costs.  By doing several important
things, I believe we can handle charge-balance, reliability, uneven
heating, and other issues of such a pack; and that we can get a pack in
relatively short time and at prices that are acceptable for now and for
possible future conversions.

I have uploaded the two sets of specs -- also pasted inline below -- to
Files/Battery info/PrelimDCellNiMHSpecs_050110rdg.txt and Files/Battery
info/PrelimCustomNiMHSpecs_050110rdg.txt.  I have also updated my
battery spreadsheet yet again, to Files/Battery
info/PriusPlusBatteries050108rdg.{xls,pdf}.

+++++++++++++++++++++++++++++++++++++++++
         Ron Gremban, rgremban@... <mailto:rgremban@...>
Moderator & Technical Lead, PRIUS+ PHEV Conversion Group
     http://groups.yahoo.com/group/priusplus
<http://groups.yahoo.com/group/priusplus/>
         http://www.priusplus.org
+++++++++++++++++++++++++++++++++++++++++

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++


California Cars Initiative PRIUS+ Project
Preliminary D-cell NiMH Battery Module Specification
Subject to Change
1/10/05 Ron Gremban

A PRIUS+ battery pack will consist of 17 modules in series, for 204V
nominal (170 cells).

Cell specifications:
===============
Amp-hr at 2C (1/2 hr) rate, minimum*:
     8 Ah

Max continuous / intermittent discharge rate (down to 20%
state-of-charge (SOC)), minimum:
     23A / 50A

Max continuous / intermittent charge rate (at up to 1.5V plus internal
resistance @ 25 deg C), minimum:
     2A / 31A

DC internal resistance, 20-100% SOC, maximum*:
     0.0045 ohm

Module specifications:
==================
Electrical configuration:
     40 cells:  10 in series x 4 in parallel (12V nominal)
     Each set of 4 parallel cells must be wired together at both ends.
     Cells wired together with spot-welded bus bars.
     Cells within each module must be matched for Ah capacity and
internal resistance
         so that active intra-module charge balancing is not required.

Physical layout:
     8 x 5 physical layout, one cell deep in the radial direction
     Physically held together and insulated for stacking without shock or
shorting hazard
     A cooling hole must be provided at each end of each gap where 3 or 4
cells abut.
     Cells must be surrounded or separated by high temperature insulation.
     Leads are pigtails from one corner of the module:
         Positive lead, 6 inches #6 wire, terminating in red Anderson
PP120 connector
         Negative lead, 3 inches #6 wire, terminating in black PP120
connector
         Positive & negative low power leads, 6 inches #18 wire,
             terminating in 2-pin, red/black Anderson PP10 connector
     Please indicate intended method of construction (e.g. end caps,
shrink wrap, etc.).

Cooling:
     (we will add our own thermal sensing devices)
     Please indicate recommendations for cooling pack of 17 modules
         at 3C RMS discharge/regen rate, 0.25C charge rate from grid
     Liquid option (much preferred):
         Use close-packing scheme
         Dimensions approx. 290 x 160 x 70 mm (11.5 x 6.125 x 2.75 inches)
         Recommend liquid medium that is non-reactive to all components
     Air option:
         Use rectangular packing scheme
         Dimensions approx. 280 x 175 x 70 mm (11 x 7 x 2.75 inches)
         Good intra-module heat conductivity is required to minimize
thermal differentials between cells

DC internal resistance, 20-100% SOC, maximum*:
     0.011 ohm

Minimum voltage during 125A discharge at 20% SOC*:
     10.6V

Max. weight:
     16 lb (7.25 kg)

Min cycle life*:
     At least 500 cycles, 80% DOD, 1000 cycles, 50% DOD
     Higher values of this parameter can make a huge improvement in
lifetime cost.

Price*, maximum:
     $450 each for 18 modules

Test module:
     One 12V module of 10 cells needed immediately for pre-purchase cell
testing.


  * The values of these parameters should be maximized (or minimized, as
appropriate), given a reasonable cost of doing so

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++


California Cars Initiative PRIUS+ Project
Preliminary Custom NiMH Battery Pack Specification
Subject to Change
1/10/05 Ron Gremban

Nominal voltage:
     202-204V (168-170 NiMH cells in series)

Optimum module size:
     For ease of assembly, trouble-shooting, and replacements, the
battery pack should be divided into approximately 10-20 equal modules.
     If the charge equalization system is not included, each module
should have 10 cells in series if possible, so that commercially
available 12V nominal equalizaton circuits can be employed.

Amp-hr at 2C (1/2 hr) rate, minimum*:
     30 (6.0 kWh) (can be via individually paralleled cells within each
module)

Max continuous / intermittent discharge rate (down to 20%
state-of-charge (SOC)), minimum:
     90A / 200A

Max continuous / intermittent charge rate (at up to 255V plus internal
resistance @ 25 deg C), minimum:
     7.5A / 125A

DC internal resistance, 20-100% SOC, maximum*:
     0.20 ohm

Minimum voltage during 125A discharge at 20% SOC*:
     180V

Charge equalization:
     Cells within each module must be carefully matched to remove need
for intra-module charge equalization
     Indicate whether or not inter-module equalization electronics is
included in the pack; if so, how it works; if not, what is recommended

Thermal management:
     Please indicate recommendations for cooling pack
         at 3C RMS discharge/regen rate, 0.25C charge rate from grid
     Liquid cooling is optimum to minimize temperature differentials
     Otherwise:
     + Modules must be laid out for good, even forced air cooling
     + Good intra-module heat conductivity is required to minimize
thermal differentials between cells
     + Per-module thermal sensing is needed for SOC monitoring, charge
control, and safety cutoff purposes

Min cycle life*:
     At least 1000 cycles, 80% DOD, 2000 cycles, 50% DOD
     Higher values of this parameter can make a huge improvement in
lifetime cost.

Max battery pack weight*:
     300 lb (136 kg)
     Note:  this and minimum 30 Ah rate force a minimum specific energy
of 44 Whr/kg at 2C

Max price*:
     $1/W-hr or less to CalCars, for 1 to 100 vehicles
         including charge balancing, cooling, and thermal management
     $1.25/kWh lifetime throughput

Projected high-volume auto manufacturer's cost in two years*:
     $0.40 or less per kilowatt-hr lifetime throughput


  * The values of these parameters should be maximized (or minimized, as
appropriate), given a reasonable cost of doing so

On that topic, Thought this news might be of interest...

As seen on the EVlist.
> I happened across a press release from Sony about li-ions coming
> out in January - check out the power density:
> http://www.sony.net/SonyInfo/News/Press/200412/04-060E/
>
> If you can't tell, I'm looking at the 26650VT cells, which have
> a maximum output of 50A (20C). At 90gm, that's twice as much as
> the G8 cells of the same capacity, but the wt/gm is better than
> li-polys. A series/parallel (2000 cells!) could max out a 1K Zilla
> in a 400# pack; it's nothing like the Bolders fueling the
> Killacycle, but perhaps useful hybridized with another pack.

Pretty good news, these new cells beat out the old power leaders,
Li-polymer!  The 18650V does 10A, similar to the higher capacity
"standard" 18650G8 which are rated for 2C(5A), 4C(10A) short peaks.
The real news is the 26650VT cell which is capable of 50A, more than
double the power by weight.  Since the voltage is higher it would
only take 1/3 the cells, 56 cells in series instead of 168. Though
50A and 2.5Ah won't cut it, you would still need to parallel cells:

3 parallel gets you enough power with 150A but still just 7.5Ah.
8 parallel gets you to the current 20Ah Pba capacity but with
400Amps available and at only 89 Lbs.  And Lithium doesn't suffer
from the Peukert effect, so you can get all 20Ah out of them.

18650- diameter 18 mm, height 65.0 mm, 90g
26650- diameter 26 mm, height 65.0 mm, 44g
C-cell diameter 22 mm, height 42.3 mm,
D-cell diameter 33 mm, height 61.5 mm,

Anyway, just one more thing to consider...

L8r
  Ryan

I have been having an email discussion with Ron Gremban about grid
charging.   So far I have been able to grid charge my prius using an
inexpensive home made charger and an auxilliary battery containing
180 3Ah NiMH sub-C cells. Felix has asked me to share the following
exchange with this group.  It will be difficult to follow because of
the numerous replys.  If you have any questions, please feel free to
ask.


Dan:
Just to let you know, I am now grid charging my Prius.  Below is an
excerpt from a message I sent to some interested parties...

I finished reconfiguring my 5 D-Packs this morning.  If I had 9 more
modules, I would have enough for a 6th pack.

Features include:
- 30 modules per D-Pack, 216V nominal (up from 29 modules, 208.8V
nominal)

Ron:
Our temporary PbA pack is also 216V nominal.  It replaces the hybrid
battery.  Are these Armando Tech. hi-rate Titanium D-cells?

Dan:
Most people assume that the D in D-Pack stands for the cell type.
That is not the case.  I am using sub-C cells in the prepackaged 7.2V
3Ah configuration.

Dan:
- All battery surfaces exposed to air (no land locked surfaces).
This will greatly improve natural convection.
- Flow thru design.  1/4" space between modules creates a direct
pathway for forced air cooling.
- 5 5/8" tall (down from 6").  This enables the D-Packs to fit in the
cargo tray while allowing the cargo tray cover to close completely.

Ron:
Indeed, this sounds like a wise design.

Dan:
SO far I have been able to charge the D-Packs up to 250V and drive
with them with no malfunction or warning lights.

Ron:
Cool!

Dan:
I have also been able to control the output of the charger by adding
light bulb resistors.  Right now I have
four 100W bulbs in parallel which limits the output to about 250VA on
the high voltage setting and 150VA on the low voltage setting.  The
transformer is rated at 300VA.  I can lower the output further by
removing one or more bulbs or by switching to a lower wattage bulb.
To increase the current I just have to add another bulb.  I have been
driving around in this 20F weather and have managed about
60MPG.  Normally I would expect about 50MPG. So it does make a
difference to grid charge.

Ron:
We're getting around 65 mpg vs. 40 mpg as a normal hybrid in much
warmer weather, around 50 deg F.  The PbA batteries already have
decreased capacity and increased voltage at these moderately low
temperatures.  In fact, regenerative braking voltages can easily
exceed 270V, causing an intermittent problem where the ICE races
and/or provides very little power until the car is put into Neutral
then returned to Drive.

Dan:
I always try to end a trip at a low SOC by using the EV botton.  That
way I can store more grid energy when I recharge.  It seems that it
takes awhile for the battery ecu to figure out what is going on after
a charge. The battery icon will show 3 bars while my measured voltage
is 240V.  Normally I would only see 215-220V with 3 bars.  During
regen, my voltage peaks at nearly 270V while before I rarely measured
above 250V.  That is probably due to the higher nominal voltage of
the new D-Packs.

Dan:
After driving for awhile, 10-15 mins, the battery icon catches up
with the measured voltage and for a long time I am showing 7 green
bars.  At this point the battery ecu now seems to recognize the
additional capacity and is able to use it.

Ron:
Interesting.

Dan:
A few times after charging I have powered on, hit the EVB and waited
to see it the battery icon would change to prove that charge was
moving from the D-Packs to the OEM battery.  So far I have not been
able to prove that this happens. Next time I will try to take some
before and after voltage readings to see if there is a difference
after 5 minutes or so.

Ron:
A way of measuring this would be to buy a "replacement" '04 Prius
Hall-effect current sensor from Toyota (around $50), and put it in
the path between your battery pack and the OEM pack.  Feed it 5V and
read the current from the middle terminal (I can send you a pinout).
100A/V; 2.5V = 0A.

Dan:
Update.... After 5-10 minutes powered up with the EV mode engaged,
the battery icon updates from 4 blue bars to 7 green bars.
Eventually the ICE kicks on and begins charging the battery.  This
seems strange to me.

Dan:
Up till now I have been charging with the D-Packs installed in the
cargo tray and with the curcuit breakers open.  It is a pain turning
them on and off at every charge.  So here is my question....

Dan:
Is it OK to grid charge with the circuit breakers closed? Since the
Power is off the solenoid that parallels the D-Packs with the OEM
battery is open.  But if I am not mistaken, the negative goes
directly to the OEM negative so they are not completely isolated.
Does anyone think it would be a bad idea to charge with the breakers
closed?

Ron:
I think either way will be fine.  This is especially true because you
are using a transformer-isolated charging arrangement.  However,
the '04 Prius' high voltage circuit is fully isolated from the
vehicle's chassis.  In fact, any leakage between any point on the
high voltage circuit and chassis ground is detected by the THS, and
causes a fatal fault.

Dan:
I do not plan on charging while powered up.  There shouldn't be a
problem with a fault.

Ron:
Roger.

Ron:
I have some other concerns about your setup, though -- mainly
connected with the fact that NiMH cells have little tolerance to
overcharging, and their end-of-charge is not as easy to determine as
it is with other chemistries.  I'm sure you have already considered,
in one way or another, the things I mention below, but I'll throw
them out just in case they are helpful.

Ron:
Two things have to do with paralleling a higher voltage battery pack
with the OEM pack.  The THS appears to use an integration of Amp-hr
in/out of the hybrid battery to keep track of state-of-charge (SOC).
I believe it uses voltage (possibly temperature-corrected open-
circuit voltage) in some way to periodically verify and/or correct
its reading of the SOC.  This is why the SOC reading eventually
catches up to your grid-charged level, and then stays there in spite
of using up some of the charge.

Ron:
My first concern is that the higher voltage pack in parallel with the
OEM pack will tend to charge the OEM pack (via equalization currents)
above the 80% SOC to which the THS otherwise limits it.  This could
significantly reduce the lifetime of the OEM battery, unless the fact
that the packs are paralleled only during driving sufficiently limits
such charge equalization.

Ron:
Secondly (and in the opposite direction), the higher voltage of your
overall pack may throw off the THS' determination of SOC.  Looking at
overall voltage, it will believe the OEM pack's SOC is higher than it
actually is, and therefore allow its actual SOC to fall below its 40%
lower limit (that occurs at 0 bars), thereby once again reducing the
OEM pack's lifetime.  At the point where the THS believes SOC is 40%,
not only will the OEM pack's SOC be below that, but your D-cell
pack's SOC will probably be below the 20% it should be kept above for
long life.

Ron:
Thirdly, because of the poor tolerance of NiMH cells to overcharging,
grid-charging NiMH packs is best done with a charger that is
sensitive to several parameters for reducing charge rate, then
stopping the charge:  temperature-compensated voltage, rate of
voltage change, and total Amp-hr of charging after a certain point.
Explanations of this process can be found in several places on the
web.  The NiMH handbook, available at www.rabbittool.com, has lots of
details.

Ron
I hope these thoughts are helpful, and are not taken as criticism of
your very cool work.  For example, the fact that you have previously
seen increased mileage by paralleling batteries to the OEM pack --
despite increased weight and without grid charging -- is very
significant.

Dan:
Yes I understand the possibility of all of those things happening.
But I am monitoring voltages and temperatures such that I am
comfortable that I would be able to see a problem.

Dan:
As to your first concern, the OEM pack is only paralleled with the
aux pack while the Prius is powered up.  The system voltage usually
drops to 238 within 30 seconds of driving.  After that the voltage
remains in the 215-238 range (no load) for the remainder of the
trip.  I believe that this is within an acceptable range for the OEM
battery. I am considering only charging to 240V, which would
completely eliminate the possibility of overcharging.

Dan:
As for your second concern, I have never measured no load voltages
below 210V.  I have measured them as low as 190 but only under heavy
load.  The voltage always returns to acceptable levels when the load
is removed.  The big assumption here is that intermittent discharge
sags to 190v and regen spikes to 270V do not damage the batteries.
If you think they do then I will have to re-evaluate the nominal
voltage level of the aux packs.

Ron:
I suspect 270V spikes on the OEM battery pack are only O.K. at low
temperatures.  This is 1.6V/cell; 1.5V/cell (252V) is the recommended
maximum charging voltage for most NiMH cells at room temperature, but
1.6V/cell is O.K. at freezing temperatures.

Ron:
On the other hand, I believe Toyota's Battery ECU makes sure that the
overall voltage doesn't exceed that which the OEM pack is designed to
handle.  Therefore, in retrospect, I think my first concern is
handled as long as the packs are paralleled only during use (as long
as a driver doesn't turn on the vehicle and just sit for a long time,
immediately after a grid charge).  And, as to my second concern, I
think the Battery ECU will continue to correctly read the OEM
battery's voltage, and therefore SOC, though this SOC will correspond
to a different SOC (as noted above) for the paralleled battery.
There will probably be some transient inaccuracies, but as long as my
first concern is handled, these should not cause the OEM battery to
exceed its normal SOC range.

Dan:
And as for your third concern, during charging, I cannot detect a
noticeable change in the battery temperature.  The charge rate is
only 2A.  I am also using a timer which will automatically shut off
the charger after a specified period of time, usually 2 hours.  I am
only charging to 80%SOC so I am not worried so much about determining
end of charge.

Ron:
You're right; this should be fine.  However, I think there is
additional opportunity.  With a smart NiMH charger, the 209V pack I
did the calculations for above can be reliably charged to 95-100%
SOC.  After sufficient driving, it should end up at 20-30% SOC,
making this setup an exceptionally simple, effective, inexpensive
PHEV conversion that does not depend on replacing Toyota's Battery
ECU with a custom battery management system!  In fact, I think we
ought to try out this configuration, too, and directly compare its
effectiveness with that of our current configuration.

Dan:
Last night I added one more module to each pack, bringing the nominal
voltage to 223.2V.  This should enable me to discharge the aux
batteries to 40%.  I charged them to 250v as usual. This morning I
did experience the ICE racing for a few seconds right when I started
up.  After that everything was fine.  I am not sure if there is a
benefit by raising the nominal voltage to 223.2.  I may go back down
to 216v.

Ron:
You may be best off going back down even further, to 209V. On page
four of http://cobasys.com/pdf/transportation/Series%201000%
20Brochure.pdf is a graph of open-circuit voltage vs. SOC for their
Series 1000 NiMH batteries (10 cells in series).  Assuming that this
curve is fairly characteristic of all NiMH cells, the SOC that the
cells, given enough time in parallel, will equalize should be
determinable by scaling and offsetting the graph.

Ron:
Though your packs are paralleled only during use, they will still try
to equalize toward the following:

At 180 (nominal 216V) vs. 168 (Toyota's 202V hybrid battery) cells in
series, if the 168-cell stack varies between Toyota's limits of 80%
and 40% SOC, the open circuit voltage will vary between 1.315V * 168
= 221V and 1.27V * 168 = 213V, the 180-cell individual cell voltages
will vary between 221V/180 = 1.23V and 213V/180 = 1.185V, and the 180-
cell SOC will vary between around 20% and around 10%.

Ron:
I believe you used to have a 174-cell (209V) stack.  It's SOC would
track the Toyota battery's SOC in an interesting and very helpful
way:  221V/174 = 1.27V => 40% SOC to 213V/174 = 1.22V => 20% SOC.
This pack seems optimum for the sort of paralleling you are doing, as
you can separately charge it to 80-100% SOC, then parallel it with
the OEM pack only during driving.  Toyota's Battery ECU, using open-
circuit voltage as a guide, would mostly discharge until reaching its
normal range of 40-80% SOC, which corresponds to 20-30% SOC for the
added batteries, thereby using up the grid-added charge in them, down
to an optimum low SOC of just over 20%.

Dan:
I will go down to the 209V configuration.  Wayne Brown came  to the
same conclusion many months ago  that 209V would be optimal.  I just
wanted to experiment a little.  You mentioned a NiMH smart charger.
Are these available for this application and how much do they cost?
I built my charger from scratch for under $100.  I seem to remember
that you bought a charger for several thousands of dollars.  I cannot
afford that.

Dan:
Thanks for taking the time to voice your concerns.  It always helps
to get a second opinion.  Hopefully I have addressed your concerns.
If you feel that my precautions are still inadequate, please let me
know.

Ron:
I believe 200+V NiMH chargers can be found for around $1000, but I
can't find one in my references.  I'll look further.

Dan:
I have been following your progress at prius+.  You have made great
progress.  If there is any way that I can help please let me know.

Ron:
Thanks a lot, and the same goes in reverse.  You have already helped,
with your reports of improved mileage despite significant extra
battery weight, and long term success with the sub-C packs.  And our
combined thinking about the paralleling may have helped again (see my
paragraph above)!

I'd like to highlight and expand upon a very intriguing possibility
brought up by Dan's success in grid charging NiMH batteries added in
parallel with the Prius' hybrid battery (previous message).  Dan's
success was unexpected, given the failure of our similar (though
different in an important way) experiments.  In retrospect, and knowing
what we now understand about Toyota's Battery ECU, I can see how this
success happened, and how it can be built upon.

In short, Dan added a slightly higher-voltage pack of NiMH cells in
parallel with Toyota's OEM hybrid battery pack.  The two battery packs
are paralleled only when the car is being driven (READY mode).  Dan then
grid-charged his added battery pack when it was disconnected from the
OEM pack, and found improved hybrid mileage due (presumably) to
depletion of the charge in the added battery pack.

Our unsuccessful paralleled battery setup, unlike Dan's, paralleled the
batteries only during discharge, causing the THS electronics to believe
that something was wrong with the battery or high power electronics,
depending upon which side of the Battery ECU's current sensor the
provisional paralleling (and thus current injection) occurred.
Amazingly, when Dan paralleled the batteries all the time, the THS did
not complain about accepting far more energy from the combined battery
than the hybrid battery could possibly provide!

This lack of complaint is probably because, though the Battery ECU keeps
track of battery state-of-charge (SOC) by integrating Amp-hours into and
out of the battery pack, it periodically resets itself when voltage
readings indicate that its knowledge of the battery's SOC is
inaccurate.  We don't yet know what sort of voltage reading is used or
when or how often the re-evaluation is triggered, but we do know that
this process occurs after the accessory battery has been disconnected,
destroying the Battery ECU's SOC memory.  Then it restarts itself by
assuming the pack's SOC is 60%, but shortly thereafter resets its SOC
reading to a more accurate number, presumably via some
temperature-compensated voltage reading (if I were designing it, my
first impulse would be to average the most recent positive and negative
zero-current-crossing voltages to get an approximate open-circuit voltage).

The clever thing about paralleling a higher-voltage battery pack of the
same chemistry is that current will flow between the two packs until
they equalize at the same open-circuit voltage.  However, because the
added pack has more cells in series, its cells will be at a lower
voltage, and thus a lower SOC, than those of the OEM pack.  If we choose
the difference between the packs carefully enough, we can end up with
the added pack's SOC at 20% when the OEM pack's SOC is at its minimum of
40%.  Conveniently, because the voltage/SOC curve for NiMH cells is
steep at the ends and flat in the middle, this means that when the OEM
pack is at its maximum of 80% SOC, the added pack's SOC will be only
around 30%!

What Dan did was parallel the packs after separately grid-charging the
added pack, let's say to 100% SOC.  If this were done without
immediately driving the car, current would flow from the added pack to
the OEM pack, eventually overcharging the OEM pack.  However, when the
car is immediately driven, two things happen:  First, current is drawn
from the parallel packs, reducing the average voltage and removing or
greatly reducing the tendency for the OEM battery to get overcharged.
Secondly, Toyota's Battery ECU, within a few minutes, notices that the
battery pack voltage is higher than expected, and resets its SOC reading
to "very high" (how high, we don't know yet).  The Battery ECU would
normally reduce its SOC reading from maximum (80%, all bars) to minimum
(40%, zero bars) by the time only 2.6 Ah (22%) of the extra charge
provided by Dan's 15 Ah added battery pack was used up.  However,
according to Dan's experience, the ECU must actually keep re-adjusting
its SOC reading based on the fact that the voltage remains high until
the added pack is discharged to the point where its open-circuit voltage
is within the OEM pack's normal range.  For the added battery pack, due
to having more cells, that occurs when its SOC is 20-30%.  Therefore,
around 75% of the added pack's charge is used up before Toyota's Battery
ECU believes the SOC to be in its normal range and stops using up the
extra charge.  Then the system keeps the added pack's SOC withing the
20-30% range while the OEM pack is kept within its normal range of
40-80% -- and the car operates as a normal hybrid (only a little more
efficiently, due to a lower internal resistance of the combined battery
packs).

The upshot of all this is that it may be possible to convert a Prius
into a PHEV by merely adding a slightly higher voltage NiMH battery
pack, some solenoids, and a charger, while not replacing any of Toyota's
electronics (specifically its Battery ECU)!  The new, higher voltage
pack could either be added to or replace the OEM pack.  An added battery
pack would need only half the power handling ability of a replacement
pack, but would need to be 65-90 lbs. lighter.  For full benefit, a
PRIUS+ using this system would need a simple CAN-bus-reading computer to
automatically enter (and possibly exit) EV-only mode whenever appropriate.

Note that this proposed configuration is dependent on characteristics of
the NiMH chemistry as well as the fact that Toyota's Battery ECU is
tuned to this chemistry.  It will probably not work with an added or
replacement pack of any other chemistry, e.g. Li-ion.  However, I have
done some checking against our existing temporary PbA pack.  Due to its
straight-line voltage/SOC curve, a different temperature coefficient
than NiMH cells, and a high internal resistance at low SOCs, the PbA
pack will work poorly in this configuration, and only over a small
temperature range.  However, with one additional battery, I believe (and
hope) it will work well enough to verify the concept.  I tried it today,
but found that I have to move the Battery ECU's current sensor to
properly do the test.  I hope to complete this operation and run the
test tomorrow (Thursday).  As soon as I manage a proper test, I will
post the results.

I see one potential problem scenario:  If an added, rather than a
replacement, battery pack is used, and, immediately after it is
grid-charged, the car is started (put in READY mode) and either left
standing for a while or immediately driven down a large hill, it is
possible to overcharge the OEM battery, maybe significantly, thereby
shortening its cycle life.  I don't yet have a solution to this
possibility other than to replace rather than add to the OEM pack.

By the way, the OEM pack has 168 cells in series for a 201.6V nominal
voltage.  The best size for the higher voltage pack is 174 cells -- e.g.
29 7.2V modules -- for a nominal 208.8V.  175 cells (17 12V modules plus
one 6V module -- or 25 8.4V modules -- for 210V nominal) is also acceptable.

Here are the most relevant passages from Dan's previous message:

  > Dan:
  > SO far I have been able to charge the D-Packs up to 250V and drive
  > with them with no malfunction or warning lights.
  >
  > I always try to end a trip at a low SOC by using the EV button.  That
  > way I can store more grid energy when I recharge.  It seems that it
  > takes awhile for the battery ecu to figure out what is going on after
  > a charge. The battery icon will show 3 bars while my measured voltage
  > is 240V.  Normally I would only see 215-220V with 3 bars.  During
  > regen, my voltage peaks at nearly 270V while before I rarely measured
  > above 250V.  That is probably due to the higher nominal voltage of
  > the new D-Packs.
  >
  > After driving for awhile, 10-15 mins, the battery icon catches up
  > with the measured voltage and for a long time I am showing 7 green
  > bars.  At this point the battery ecu now seems to recognize the
  > additional capacity and is able to use it.
  >
  > Ron:
  > <snip> I think there is
  > additional opportunity.  With a smart NiMH charger, the 209V pack I
  > did the calculations for above can be reliably charged to 95-100%
  > SOC.  After sufficient driving, it should end up at 20-30% SOC,
  > making this setup an exceptionally simple, effective, inexpensive
  > PHEV conversion that does not depend on replacing Toyota's Battery
  > ECU with a custom battery management system!  In fact, I think we
  > ought to try out this configuration, too, and directly compare its
  > effectiveness with that of our current configuration.
  >
  > <snip> I believe Toyota's Battery ECU makes sure that the
  > overall voltage doesn't exceed that which the OEM pack is designed to
  > handle.  Therefore, in retrospect, I think my first concern is
  > handled as long as the packs are paralleled only during use (as long
  > as a driver doesn't turn on the vehicle and just sit for a long time,
  > immediately after a grid charge).  And, as to my second concern, I
  > think the Battery ECU will continue to correctly read the OEM
  > battery's voltage, and therefore SOC, though this SOC will correspond
  > to a different SOC (as noted above) for the paralleled battery.
  > There will probably be some transient inaccuracies, but as long as my
  > first concern is handled, these should not cause the OEM battery to
  > exceed its normal SOC range.
  >
  > You may be best off going back down even further [from 216V], to
209V. On page
  > four of http://cobasys.com/pdf/transportation/Series%201000%
  > 20Brochure.pdf is a graph of open-circuit voltage vs. SOC for their
  > Series 1000 NiMH batteries (10 cells in series).  Assuming that this
  > curve is fairly characteristic of all NiMH cells, the SOC that the
  > cells, given enough time in parallel, will equalize should be
  > determinable by scaling and offsetting the graph.
  >
  > Though your packs are paralleled only during use, they will still try
  > to equalize toward the following:
  >
  > I believe you used to have a 174-cell (209V) stack.  It's SOC would
  > track the Toyota battery's SOC in an interesting and very helpful
  > way:  221V/174 = 1.27V => 40% SOC to 213V/174 = 1.22V => 20% SOC.
  > This pack seems optimum for the sort of paralleling you are doing, as
  > you can separately charge it to 80-100% SOC, then parallel it with
  > the OEM pack only during driving.  Toyota's Battery ECU, using open-
  > circuit voltage as a guide, would mostly discharge until reaching its
  > normal range of 40-80% SOC, which corresponds to 20-30% SOC for the
  > added batteries, thereby using up the grid-added charge in them, down
  > to an optimum low SOC of just over 20%.
  >

--

+++++++++++++++++++++++++++++++++++++++++
         Ron Gremban, rgremban@... <mailto:rgremban@...>
Moderator & Technical Lead, PRIUS+ PHEV Conversion Group
     http://groups.yahoo.com/group/priusplus
<http://groups.yahoo.com/group/priusplus/>
         http://www.priusplus.org
+++++++++++++++++++++++++++++++++++++++++

Ron,
I wanted to confirm that most of what you have said here is true.
But there are a few exceptions/corrections that I would like to make.

- I cannot yet quantify the benefit of grid charging.  I have no
comparison data at temperatures below 45F.  What I did say is that I
am fairly confident that there is a benefit and made an estimate
which was stated earlier.  I have a meter that I will start using
that will record the kwh useage of the charger so I can make a cents
per mile calculation and compare that to using gasoline only.

- In theory your statement that my aux packs will be at 20% SOC while
the OEM pack is at 40% is true.  But this is never realistically
achieveable with my setup for at least two reasons.

- First, EV mode is automatically overridden when the battery icon
reaches 2 bars.  The the ICE starts and begins charging.  So you want
to arrive home at about 3 bars.  The problem is about 50% of the time
you cannot engage EV to equalize the two packs which is part of my
charging routine.  So to ensure the ability to engage EV 100% of the
time, you have to shoot for 4 bars.  That being the case, you can
only get the OEM SOC down to 66% (I am guessing here).  So for my
216v nominal aux packs (I am at 216v not 209V because of this) the
SOC is about 47%.

- The second reason is that the aux packs bounce back significantly
after powering down.  As such, I have never been able to get them
below 223v, 55% SOC (no load) before recharging.

My recharging routine is as follows:
- Arrive home with battery icon at 3-4 bars and power down.
- Charge the aux packs to 80% SOC (charger is adjustable between 1.5
and 2.5A) with the prius powered down.
- Power up the Prius and engage EV mode.
- Allow the aux pack to charge the OEM pack, the battery icon will go
to six or seven bars.
- At some point the ICE will override EV mode and start to charge the
battery.  (I am confused by the purpose of this behavior)
- Immediately power down as soon as the ICE starts.
- Charge aux packs back to 80%

I will experiment with charging the aux packs to a higher SOC after I
get some good baseline data for 80% SOC.

And finally..... You give me too much credit.

Please be aware that the hard work of many went into the design,
construction and installation of my paralleling interface.  My
contribution has been limited to identifying and procuring suitable
batteries and the development (with technical support from others)
and construction of a low cost grid charger.

Dan



--- In priusplus@yahoogroups.com, Ron Gremban <rgremban@c...> wrote:
> I'd like to highlight and expand upon a very intriguing possibility
> brought up by Dan's success in grid charging NiMH batteries added
in
> parallel with the Prius' hybrid battery (previous message).  Dan's
> success was unexpected, given the failure of our similar (though
> different in an important way) experiments.  In retrospect, and
knowing
> what we now understand about Toyota's Battery ECU, I can see how
this
> success happened, and how it can be built upon.
>
> In short, Dan added a slightly higher-voltage pack of NiMH cells in
> parallel with Toyota's OEM hybrid battery pack.  The two battery
packs
> are paralleled only when the car is being driven (READY mode).  Dan
then
> grid-charged his added battery pack when it was disconnected from
the
> OEM pack, and found improved hybrid mileage due (presumably) to
> depletion of the charge in the added battery pack.
>
> Our unsuccessful paralleled battery setup, unlike Dan's, paralleled
the
> batteries only during discharge, causing the THS electronics to
believe
> that something was wrong with the battery or high power
electronics,
> depending upon which side of the Battery ECU's current sensor the
> provisional paralleling (and thus current injection) occurred.
> Amazingly, when Dan paralleled the batteries all the time, the THS
did
> not complain about accepting far more energy from the combined
battery
> than the hybrid battery could possibly provide!
>
> This lack of complaint is probably because, though the Battery ECU
keeps
> track of battery state-of-charge (SOC) by integrating Amp-hours
into and
> out of the battery pack, it periodically resets itself when voltage
> readings indicate that its knowledge of the battery's SOC is
> inaccurate.  We don't yet know what sort of voltage reading is used
or
> when or how often the re-evaluation is triggered, but we do know
that
> this process occurs after the accessory battery has been
disconnected,
> destroying the Battery ECU's SOC memory.  Then it restarts itself
by
> assuming the pack's SOC is 60%, but shortly thereafter resets its
SOC
> reading to a more accurate number, presumably via some
> temperature-compensated voltage reading (if I were designing it, my
> first impulse would be to average the most recent positive and
negative
> zero-current-crossing voltages to get an approximate open-circuit
voltage).
>
> The clever thing about paralleling a higher-voltage battery pack of
the
> same chemistry is that current will flow between the two packs
until
> they equalize at the same open-circuit voltage.  However, because
the
> added pack has more cells in series, its cells will be at a lower
> voltage, and thus a lower SOC, than those of the OEM pack.  If we
choose
> the difference between the packs carefully enough, we can end up
with
> the added pack's SOC at 20% when the OEM pack's SOC is at its
minimum of
> 40%.  Conveniently, because the voltage/SOC curve for NiMH cells is
> steep at the ends and flat in the middle, this means that when the
OEM
> pack is at its maximum of 80% SOC, the added pack's SOC will be
only
> around 30%!
>
> What Dan did was parallel the packs after separately grid-charging
the
> added pack, let's say to 100% SOC.  If this were done without
> immediately driving the car, current would flow from the added pack
to
> the OEM pack, eventually overcharging the OEM pack.  However, when
the
> car is immediately driven, two things happen:  First, current is
drawn
> from the parallel packs, reducing the average voltage and removing
or
> greatly reducing the tendency for the OEM battery to get
overcharged.
> Secondly, Toyota's Battery ECU, within a few minutes, notices that
the
> battery pack voltage is higher than expected, and resets its SOC
reading
> to "very high" (how high, we don't know yet).  The Battery ECU
would
> normally reduce its SOC reading from maximum (80%, all bars) to
minimum
> (40%, zero bars) by the time only 2.6 Ah (22%) of the extra charge
> provided by Dan's 15 Ah added battery pack was used up.  However,
> according to Dan's experience, the ECU must actually keep re-
adjusting
> its SOC reading based on the fact that the voltage remains high
until
> the added pack is discharged to the point where its open-circuit
voltage
> is within the OEM pack's normal range.  For the added battery pack,
due
> to having more cells, that occurs when its SOC is 20-30%.
Therefore,
> around 75% of the added pack's charge is used up before Toyota's
Battery
> ECU believes the SOC to be in its normal range and stops using up
the
> extra charge.  Then the system keeps the added pack's SOC withing
the
> 20-30% range while the OEM pack is kept within its normal range of
> 40-80% -- and the car operates as a normal hybrid (only a little
more
> efficiently, due to a lower internal resistance of the combined
battery
> packs).
>
> The upshot of all this is that it may be possible to convert a
Prius
> into a PHEV by merely adding a slightly higher voltage NiMH battery
> pack, some solenoids, and a charger, while not replacing any of
Toyota's
> electronics (specifically its Battery ECU)!  The new, higher
voltage
> pack could either be added to or replace the OEM pack.  An added
battery
> pack would need only half the power handling ability of a
replacement
> pack, but would need to be 65-90 lbs. lighter.  For full benefit, a
> PRIUS+ using this system would need a simple CAN-bus-reading
computer to
> automatically enter (and possibly exit) EV-only mode whenever
appropriate.
>
> Note that this proposed configuration is dependent on
characteristics of
> the NiMH chemistry as well as the fact that Toyota's Battery ECU is
> tuned to this chemistry.  It will probably not work with an added
or
> replacement pack of any other chemistry, e.g. Li-ion.  However, I
have
> done some checking against our existing temporary PbA pack.  Due to
its
> straight-line voltage/SOC curve, a different temperature
coefficient
> than NiMH cells, and a high internal resistance at low SOCs, the
PbA
> pack will work poorly in this configuration, and only over a small
> temperature range.  However, with one additional battery, I believe
(and
> hope) it will work well enough to verify the concept.  I tried it
today,
> but found that I have to move the Battery ECU's current sensor to
> properly do the test.  I hope to complete this operation and run
the
> test tomorrow (Thursday).  As soon as I manage a proper test, I
will
> post the results.
>
> I see one potential problem scenario:  If an added, rather than a
> replacement, battery pack is used, and, immediately after it is
> grid-charged, the car is started (put in READY mode) and either
left
> standing for a while or immediately driven down a large hill, it is
> possible to overcharge the OEM battery, maybe significantly,
thereby
> shortening its cycle life.  I don't yet have a solution to this
> possibility other than to replace rather than add to the OEM pack.
>
> By the way, the OEM pack has 168 cells in series for a 201.6V
nominal
> voltage.  The best size for the higher voltage pack is 174 cells --
e.g.
> 29 7.2V modules -- for a nominal 208.8V.  175 cells (17 12V modules
plus
> one 6V module -- or 25 8.4V modules -- for 210V nominal) is also
acceptable.
>
> Here are the most relevant passages from Dan's previous message:
>
>  > Dan:
>  > SO far I have been able to charge the D-Packs up to 250V and
drive
>  > with them with no malfunction or warning lights.
>  >
>  > I always try to end a trip at a low SOC by using the EV button.
That
>  > way I can store more grid energy when I recharge.  It seems that
it
>  > takes awhile for the battery ecu to figure out what is going on
after
>  > a charge. The battery icon will show 3 bars while my measured
voltage
>  > is 240V.  Normally I would only see 215-220V with 3 bars.  During
>  > regen, my voltage peaks at nearly 270V while before I rarely
measured
>  > above 250V.  That is probably due to the higher nominal voltage
of
>  > the new D-Packs.
>  >
>  > After driving for awhile, 10-15 mins, the battery icon catches up
>  > with the measured voltage and for a long time I am showing 7
green
>  > bars.  At this point the battery ecu now seems to recognize the
>  > additional capacity and is able to use it.
>  >
>  > Ron:
>  > <snip> I think there is
>  > additional opportunity.  With a smart NiMH charger, the 209V
pack I
>  > did the calculations for above can be reliably charged to 95-100%
>  > SOC.  After sufficient driving, it should end up at 20-30% SOC,
>  > making this setup an exceptionally simple, effective, inexpensive
>  > PHEV conversion that does not depend on replacing Toyota's
Battery
>  > ECU with a custom battery management system!  In fact, I think we
>  > ought to try out this configuration, too, and directly compare
its
>  > effectiveness with that of our current configuration.
>  >
>  > <snip> I believe Toyota's Battery ECU makes sure that the
>  > overall voltage doesn't exceed that which the OEM pack is
designed to
>  > handle.  Therefore, in retrospect, I think my first concern is
>  > handled as long as the packs are paralleled only during use (as
long
>  > as a driver doesn't turn on the vehicle and just sit for a long
time,
>  > immediately after a grid charge).  And, as to my second concern,
I
>  > think the Battery ECU will continue to correctly read the OEM
>  > battery's voltage, and therefore SOC, though this SOC will
correspond
>  > to a different SOC (as noted above) for the paralleled battery.
>  > There will probably be some transient inaccuracies, but as long
as my
>  > first concern is handled, these should not cause the OEM battery
to
>  > exceed its normal SOC range.
>  >
>  > You may be best off going back down even further [from 216V], to
> 209V. On page
>  > four of http://cobasys.com/pdf/transportation/Series%201000%
>  > 20Brochure.pdf is a graph of open-circuit voltage vs. SOC for
their
>  > Series 1000 NiMH batteries (10 cells in series).  Assuming that
this
>  > curve is fairly characteristic of all NiMH cells, the SOC that
the
>  > cells, given enough time in parallel, will equalize should be
>  > determinable by scaling and offsetting the graph.
>  >
>  > Though your packs are paralleled only during use, they will
still try
>  > to equalize toward the following:
>  >
>  > I believe you used to have a 174-cell (209V) stack.  It's SOC
would
>  > track the Toyota battery's SOC in an interesting and very helpful
>  > way:  221V/174 = 1.27V => 40% SOC to 213V/174 = 1.22V => 20%
SOC.
>  > This pack seems optimum for the sort of paralleling you are
doing, as
>  > you can separately charge it to 80-100% SOC, then parallel it
with
>  > the OEM pack only during driving.  Toyota's Battery ECU, using
open-
>  > circuit voltage as a guide, would mostly discharge until
reaching its
>  > normal range of 40-80% SOC, which corresponds to 20-30% SOC for
the
>  > added batteries, thereby using up the grid-added charge in them,
down
>  > to an optimum low SOC of just over 20%.
>  >
>
> --
>
> +++++++++++++++++++++++++++++++++++++++++
>         Ron Gremban, rgremban@C... <mailto:rgremban@C...>
> Moderator & Technical Lead, PRIUS+ PHEV Conversion Group
>     http://groups.yahoo.com/group/priusplus
> <http://groups.yahoo.com/group/priusplus/>
>         http://www.priusplus.org
> +++++++++++++++++++++++++++++++++++++++++

Hello crew:

Well , this is all fascinating.  I am itching to try the parallel
battery option.  Here are my thoughts.

The Toyota system must be capable of managing an older and tired battery
pack with altered polarization curve - thus it has to be adaptive to
manage battereis that fall in voltage more quickly under load, and by
extension, probably has some ability to manage batteries that seem
better than new (i.e. on hot days when chemistry is active).  In the
North here, the system can manage the SOC at -36C, so imagine how
different the IR curves are then.  So, the important thing about the
additional pack is that it injects the current on the battery side of
the sensors, so the ECU does not thik there is a current loss somewhere
- how does the ECU know if this is a sign of the battery being
exceptionally good condition, or of an extra pack being added, it
probably cannot tell the difference, or it may use the recharge
conditions.  Since we want to drain the additonal pack as much as
possible, why not put a diode in the extra pack to prevent it from
charging from the stock battery, this would cause a loss of 0.2 volts
maybe, but on recharge the car would have stock characteristics? It will
be true the combined battery has lower resistance for better
regenerative capacity, but I wonder if that will make much overall
efficiency difference,so why keep putting power back into the augmenting
battery, this will alos eliminate the temperature control issues with
the extra battery, as it will only have discharge waste heat to manage..

In the long term, I think there will need to be a programmable power
exchange interface for an existng and augmenting battery pack, there are
just too many operating conditions that require specific modes to be in
place to prevent under and overcharging if the operator is not involved
minute by minute for all daily operating possibilities, but I can see
logic in keeping stock pack (especially when you go for warranty work!!!).

On  another note.  Does the vehicle not go in EV mode at a specific
level of charge whenever the engine temperature is above a certain
level.  Can a simpler EV mode be created by modifying the engine temp
signal (from a thermister)?

Keep up interesting efforts.

Steve Lapp
Lapp Renewables Ltd.
www.lapprenewables.com
ph. 613 376 6363
fax 613 376 3940

I have begun experiments with our PbA pack paralleled with the hybrid
pack whenever the car is in READY mode, with generally positive but
imperfect results.

Note that NiMH batteries have an S-curved open-circuit-voltage (OCV) vs.
SOC curve -- flat in the middle and steep at both low and high SOC --
whereas PbA batteries have a linear OCV vs. SOC curve.  One can take
advantage of the S-shaped curve in paralleling unequal voltages of NiMH
batteries:  when the lower voltage (Toyota OEM hybrid) battery is in the
range of 40-80% SOC, centered on 60%, the (added) higher voltage
battery's per-cell voltage can equalize to a smaller range of e.g.
20-45% SOC, centered on 25%.

With PbA batteries' linear OCV/SOC curve, I have to go to a higher
voltage yet (228V nominal vs. the OEM's 202V nominal) to get the added
battery's related range to 10-45% SOC, centered on 22%.  I got this
extra-high voltage by adding a 19th 12V PbA battery in series with the
existing 18-battery (216V nominal) pack.  By the way, I don't know how
these "nominal" voltages were decided upon for various chemistries:
PbA's nominal 2V/cell occurs at around 30% SOC, while NiMH's nominal
1.2V/cell occurs at around 11% SOC.  The average NiMH cell voltage is
more like 1.28V/cell.

At any rate, with 228V of PbA batteries paralleled with Toyota's hybrid
battery, and using Toyota's Battery ECU, the reported SOC eventually
peaked at 88% before getting used up while driving.  For the first 5-10
miles, the voltage stayed very high, around 240V peaking well above
250V, and the engine kept racing -- a problem we have gotten
periodically (and nowhere near as persistently) using our EnergyCS CDU
in place of Toyota's Battery ECU.  It appears that what is really
happening is that the THS is programmed to pump current into MG1 against
the ICE's friction drag, in order to quickly dump excess battery charge
when over-high battery voltage is detected.

After the battery voltage went down sufficiently that engine racing was
no longer a problem, I was on the freeway and found that the system was
using the battery, getting around 65 mpg, until around 8 Ah was used.
By that point (after 20-30 miles on the highway), reported SOC had
settled down to around 60% and on the average, current was no longer
being drawn from the PbA battery, and mileage fell to around my normal
40 mpg.  The mileage for the whole 44 mile trip was above 50 mpg.

Next, I recharged the PbA pack, removed the extra PbA battery, and tried
again.  This time I had a trip around town.  Though battery voltage
stayed centered around a quite high 232V throughout the trip, reported
SOC never went above 65%.  When I went into EV-only mode, reported SOC
would eventually fall to 45%, kicking the car out of EV-only mode.  A
mile or so of driving  with the ICE running would move the reported SOC
back up to 50%, even though sufficient Ah were not added back to do so.
At this point EV-only was again possible.  I managed to use 3 Ah of PbA
battery charge by doing this; otherwise, the Battery ECU did not
translate the high open-circuit voltage into reporting a high SOC!  I
don't know when or how this does happen, and will need to discover more
to make this system work.

Throughout driving with the batteries paralleled, the charge/discharge
voltage swings were dramatically reduced, probably to at most half,
presumably increasing battery cycle efficiency and, hopefully, vehicle
mileage.  On the above 13 mile trip, I used 3 Ah of electricity and got
47.5 mpg (vs. high 30's otherwise for this sort of trip), so this may be
happening.  Tomorrow I plan a standard trip to Sausalito, which should
be a better test.

Steve Lapp has suggested connecting the added battery pack with a diode
so that it gets only discharged, not charged.  This is very close to the
SCR configuration we tried at the beginning.  We found the reported SOC
went down as Ah were used, quickly reaching 45%, kicking the vehicle out
of EV-only mode and causing the ICE to recharge the combined pack.  It
is possible that it didn't work because the added battery's voltage
wasn't high enough to entice the Battery ECU to temporarily increase its
reported SOC.  Though the SCR configuration would be hard to get back
to, I will try with 19 batteries and a diode.

It would actually be useful for the added battery to be rechargeable via
regen braking, but not when its SOC and voltage are too high.  Maybe,
with an added NiMH pack that has the desired SOC equalization
characteristics discussed above, a diode could be in the circuit when
recharge voltage would otherwise be too high, but shorted out with a
contactor once enough charge has been depleted.

Another configuration worth trying is a higher-voltage NiMH pack in
place of (rather than paralleling) the OEM hybrid battery.  The Battery
ECU may run it down to 20-30% SOC, thinking it is 40-80%.  Excess regen
charge voltage may be avoided due to lower internal resistance than the
OEM pack's.  Internal resistance would be higher than with both battery
packs, but weight would be 65-95 lb lower.

Earlier, before the above experiments, I had a day when PbA battery
performance got really terrible.  I thought I may have charge balance
problems, so I took the following readings of the 18 PbA batteries:
     12.46, 12.43, 12.46, 12.44, 12.44, 12.44, 12.46, 12.44, 0
     12.43, 12.42, 12.44, 12.44, 12,48, 12,45, 12,44, 12,49, 12.44
I concluded that the charge was very well balanced, but found that one
of the batteries had been shorted out against a temperature probe.  I
replaced the dead (open-circuit) battery with our spare, tightened all
connections, and moved all connecting bus bars further away from the
temperature probes.  Though I later removed the offending temperature
probe from the circuit, I don't yet have the probes working right again.

I have been noticing two significant effects of our current cool 40-50
deg F (5-11 deg C) weather:

My car's mileage appears to be very poor -- 24-36 mpg -- for the first
couple of miles of ICE operation; then it improves to its now-normal 40+
mpg range.  On the short 7 mile trips I often take, this significantly
lowers my overall HEV or PHEV mileage.

The PbA batteries appear to be significantly higher voltage and softer
(higher internal resistance) than at the 60-80 deg F temperatures we
were having when the PbA pack was first installed.  Either that or the
batteries are already wearing out.  In either case, the PbA pack is
quite limited and, for many reasons, I hope we can get the right NiMH
pack soon.

--

+++++++++++++++++++++++++++++++++++++++++
         Ron Gremban, rgremban@... <mailto:rgremban@...>
Moderator & Technical Lead, PRIUS+ PHEV Conversion Group
     http://groups.yahoo.com/group/priusplus
<http://groups.yahoo.com/group/priusplus/>
         http://www.priusplus.org
+++++++++++++++++++++++++++++++++++++++++

In writing my preliminary D-cell battery pack specification, I took Dr.
Andy Frank's suggestion to parallel cells on a cell-by-cell basis,
rather than connecting strings of cells in series and paralleling the
strings.  I presumed I was not only taking advantage of Andy's broad
experience, but following what appears to make sense anyway:
open-circuit voltage vs. SOC is determined mainly by chemistry, rather
than by cell size or manufacturing variations.  Cells in parallel should
then tend to equalize to the same SOC, and capacity variations between
sets of paralleled cells should be, on average, smaller than between
individual cells.

However, a possible supplier, and very interested and helpful person
(who I won't name until I learn whether or not he wants to be named),
has been using series/parallel strings of high-power NiMH D-cells in
power supplies.  They match the cells at high power, but find then that
self-discharge and other low-power characteristics are insufficiently
matched.  He says that the paralleled cells tend to equalize to unequal
SOCs, and some paralleled sets of cells eventually begin to look like
bad supercells.

This puts enough uncertainty on my planned method of building a
series/parallel pack that I'm concerned.  My hope has been to build
modules with sufficient intra-module consistency that charge balancing
need be done only between modules.  However, CalCars doesn't have the
time or resources to do the level of experimentation that would be
necessary to solve this dilemma experimentally; nor, if possible, do we
wish to learn the hard way via after-the-fact battery pack failures.

Therefore, I'm asking to learn from anyone's experience with cell
paralleling, especially NiMH cell series/paralleling.  If you have such
experience, please send me an email, preferably with contact information
so we can talk.

This seems like an appropriate point to also publicly thank Ron Freund,
who not only came up with what became the successful solutions to our
previous CAN bus noise problem, but also lent me his high quality
oscilloscope (which I still have) to help track down the problem.  I am
also thankful for the other helpful suggestions that several people
made, bits and pieces of which helped us converge on the solution.

Thanks once again in advance,
/ron
--

+++++++++++++++++++++++++++++++++++++++++
         Ron Gremban, rgremban@... <mailto:rgremban@...>
Moderator & Technical Lead, PRIUS+ PHEV Conversion Group
     http://groups.yahoo.com/group/priusplus
<http://groups.yahoo.com/group/priusplus/>
         http://www.priusplus.org
+++++++++++++++++++++++++++++++++++++++++

Hello Crew:

See comments inserted:

> I have been noticing two significant effects of our current cool 40-50
> deg F (5-11 deg C) weather:
>
> My car's mileage appears to be very poor -- 24-36 mpg -- for the first
> couple of miles of ICE operation; then it improves to its now-normal 40+
> mpg range.  On the short 7 mile trips I often take, this significantly
> lowers my overall HEV or PHEV mileage.

Where we are in Canada - with -30C on some mornings ( I have seen -34C
on dash display), the mileage is terrible - it just takes many BTU's to
get everything up to operating temperature - and of course all the oil
and grease viscosities are higher, even the air density is greater so
fricton and air resistance all add up to more losses!  I have seen 8
litres per 100km at these temperatures for first 10 minutes - that's 28
mpg (US gallon).  If backing up, I can tell the battery is very weak -
can hardly back up until car has run for 1/2 hour or so.  I just hate to
think what mileage non-hybrid cars get in these conditions!  I suspect
that when the car is old, it will be the morning start at these
conditions that may require NiMH battery replacement.

>
>
> The PbA batteries appear to be significantly higher voltage and softer
> (higher internal resistance) than at the 60-80 deg F temperatures we
> were having when the PbA pack was first installed.  Either that or the
> batteries are already wearing out.  In either case, the PbA pack is
> quite limited and, for many reasons, I hope we can get the right NiMH
> pack soon.
>
Batteries are at higher voltage now?? - I would have thought lower if
aging is happening.  Lead acid battery discharge characteristics can
improve over the first few dozen charge cycles if they are well
managed.  If it is getting colder - you will need to increase charge
voltage at any given amperage at a specific SOC, so yes, if it is
colder, higher voltage required for charging, but open circuit voltage
would be slightly lower.  There must be lots of stuff in the Toyota  ECU
to manage cold weather charging.  Do you need to know lead acid battery
temperature compensation with temperature?, I can dig that up.

Cheers

Steve

> --
>
> +++++++++++++++++++++++++++++++++++++++++
>         Ron Gremban, rgremban@... <mailto:rgremban@...>
> Moderator & Technical Lead, PRIUS+ PHEV Conversion Group
>     http://groups.yahoo.com/group/priusplus
> <http://groups.yahoo.com/group/priusplus/>
>         http://www.priusplus.org
> +++++++++++++++++++++++++++++++++++++++++
>
>
> ------------------------------------------------------------------------
> Yahoo! Groups Links
>
>     * To visit your group on the web, go to:
>       http://groups.yahoo.com/group/priusplus/
>
>     * To unsubscribe from this group, send an email to:
>       priusplus-unsubscribe@yahoogroups.com
>       <mailto:priusplus-unsubscribe@yahoogroups.com?subject=Unsubscribe>
>
>     * Your use of Yahoo! Groups is subject to the Yahoo! Terms of
>       Service <http://docs.yahoo.com/info/terms/>.
>
>

--
Steve Lapp
Lapp Renewables Ltd.
www.lapprenewables.com
ph. 613 376 6363
fax 613 376 3940



[Non-text portions of this message have been removed]

Steve:

Up in Canada you need to keep your batteries plugged in when not driving to
keep the batteries warmed up, and it helps to keep a block heater plugged
as well.  The average person drives his car less than 3 hours a day, the
other 21 hrs. it could be plugged in. If the battery charger is semi
intelligent it should be at a trickle charge most of the time, just enough
to keep things warm.

Prof Frank


At 07:36 PM 1/24/2005, you wrote:

>Hello Crew:
>
>See comments inserted:
>
> > I have been noticing two significant effects of our current cool 40-50
> > deg F (5-11 deg C) weather:
> >
> > My car's mileage appears to be very poor -- 24-36 mpg -- for the first
> > couple of miles of ICE operation; then it improves to its now-normal 40+
> > mpg range.  On the short 7 mile trips I often take, this significantly
> > lowers my overall HEV or PHEV mileage.
>
>Where we are in Canada - with -30C on some mornings ( I have seen -34C
>on dash display), the mileage is terrible - it just takes many BTU's to
>get everything up to operating temperature - and of course all the oil
>and grease viscosities are higher, even the air density is greater so
>fricton and air resistance all add up to more losses!  I have seen 8
>litres per 100km at these temperatures for first 10 minutes - that's 28
>mpg (US gallon).  If backing up, I can tell the battery is very weak -
>can hardly back up until car has run for 1/2 hour or so.  I just hate to
>think what mileage non-hybrid cars get in these conditions!  I suspect
>that when the car is old, it will be the morning start at these
>conditions that may require NiMH battery replacement.
>
> >
> >
> > The PbA batteries appear to be significantly higher voltage and softer
> > (higher internal resistance) than at the 60-80 deg F temperatures we
> > were having when the PbA pack was first installed.  Either that or the
> > batteries are already wearing out.  In either case, the PbA pack is
> > quite limited and, for many reasons, I hope we can get the right NiMH
> > pack soon.
> >
>Batteries are at higher voltage now?? - I would have thought lower if
>aging is happening.  Lead acid battery discharge characteristics can
>improve over the first few dozen charge cycles if they are well
>managed.  If it is getting colder - you will need to increase charge
>voltage at any given amperage at a specific SOC, so yes, if it is
>colder, higher voltage required for charging, but open circuit voltage
>would be slightly lower.  There must be lots of stuff in the Toyota  ECU
>to manage cold weather charging.  Do you need to know lead acid battery
>temperature compensation with temperature?, I can dig that up.
>
>Cheers
>
>Steve
>
> > --
> >
> > +++++++++++++++++++++++++++++++++++++++++
> >         Ron Gremban, rgremban@... <mailto:rgremban@...>
> > Moderator & Technical Lead, PRIUS+ PHEV Conversion Group
> >     http://groups.yahoo.com/group/priusplus
> > <http://groups.yahoo.com/group/priusplus/>
> >         http://www.priusplus.org
> > +++++++++++++++++++++++++++++++++++++++++
> >
> >
> > ------------------------------------------------------------------------
> > Yahoo! Groups Links
> >
> >     * To visit your group on the web, go to:
> >       http://groups.yahoo.com/group/priusplus/
> >
> >     * To unsubscribe from this group, send an email to:
> >       priusplus-unsubscribe@yahoogroups.com
> >       <mailto:priusplus-unsubscribe@yahoogroups.com?subject=Unsubscribe>
> >
> >     * Your use of Yahoo! Groups is subject to the Yahoo! Terms of
> >       Service <http://docs.yahoo.com/info/terms/>.
> >
> >
>
>--
>Steve Lapp
>Lapp Renewables Ltd.
>www.lapprenewables.com
>ph. 613 376 6363
>fax 613 376 3940
>
>
>
>[Non-text portions of this message have been removed]
>
>
>
>
>
>Yahoo! Groups Links
>
>
>
>

d0li0 wrote:

>First regarding the PbA performance as related to temperature.
>You may see a loss of up to half the normal capacity of the PbA
>pack if you are not keeping them warm with heaters, ~70-80 deg F.
>If temps are near freezing 30-40 ~50 degrees, performance will drop.
>This is normal for the PbA chemestry.
>
>
Though I'm aware of dramatically lower PbA performance as temperatures
decrease, it still surprised me at our modestly lower temperatures
(Steve Lapp reports on Prius performance at a shivering -30 deg C (-16
deg F).  PbA temperature compensation for charging is around -4 mV/degC
per cell; presumably, open-circuit voltage varies at approximately the
same rate.  Therefore, my experience of our PbA pack 's higher
open-circuit voltage as well as higher internal resistance at lower
temperatures makes sense.  I doubt that our batteries are aging; I just
threw that possibility out as a possibility, since I was surprised at
the extent of increased internal resistance with only modestly lower
temperatures.

>
>Regarding NiMH Series/Paralleling, The general rule is to not
>parallel cells, but to choose a single cell which is large enough
>for the job at hand.  But we both know that this isn't always
>possible. So, here are some things to consider...
>
>
Among high (though lower) volume production cells, there are F (13Ah)
and M (20Ah) cells as well as D (9Ah) cells.  These would require less
paralleling to get the 35Ah (30Ah @ 2C rate) that we need.  However, the
larger cells I've found have significantly poorer high-power
capabilities (e.g. 5 and 7 vs. 2.2% voltage sag/C, respectively).

>PbA's are some times paralleled in BEV's in various ways.
>One method is to use two parallel strings, another is to
>buddy pair 'cells' and string them in series, each works
>fine as long as you consider the interconnect lengths, and
>try to keep from connecting a pair of cells such that one
>has a much shorter path than the other...
>( but you aren't doing this, so I won't go any further )
>
>More importantly is the SOC curves of the various chemestries.
>PbA and Li-ion both have fairly linear curves, so you can parallel
>them and they pretty much stay at equal voltage and SOC...
>I have a pack with 12 parallel Li-ion cells and it performes
>quite well, ACP also uses 68P blocks of Li-ion cells...
>
>However this is quite different from Ni based cells which have a
>peak and then a voltage drop as they reach full charge.  This
>characteristic makes for trouble when cells are in parallel.
>
>Imagine you have 3 NiMH cells in parallel, for some reason
>cell 1 reaches full first, It's voltage peakes and then falls.
>Now the other two cells haven't reached their peak, and they
>never will because all the charge current is now going to flow
>into the fully charged cell, causing it to heat up and fail.
>
>
Very good point!  Thanks for pointing it out!  It's one of those "Why
didn't I think of that?" things.  Also, because the open-circuit voltage
vs. SOC is an S curve with little change through middle SOC levels,
paralleled cells with only slightly different voltage curves will not
equalize at the same SOC *except* at high or low SOC.  And high-SOC
equalization, as you aptly point out, doesn't work well during charging!

This, along with Jim Bohorquez' actual problematic experiences of
paralleling NiMH D-cells on a cell-by-cell basis at Mesa Power, seems to
rule out this way of building a series-parallel array of cells for our
NiMH battery pack.

It seems that the offset of per-cell charging voltage peaks with strings
of series NiMH cells would not be as serious an issue as with paralleled
cells.  Therefore, maybe one can parallel long strings of NiMH cells IF
the strings match closely enough in average capacity, which should be
more likely than individual cell matches.  Dan Kroushl and associates
have had good experience over many, many (HEV, not PHEV) Prius miles
with multiple 209V strings of sub-C cells paralleled together (and
paralleled with Toyota's hybrid battery).  This experience is far more
significant than theory, though new problems could develop now that they
are experimenting with the major operational change of adding grid-charging.

>So, I guess the point is that paralleling cells is not the ideal
>solution, but paralleling Ni cells is simple a very bad idea.
>
>Now you could parallel two strings of NiMH cells, assuming each
>one will have it's own bms to keep it in line, which they will.
>Of course a BMS isn't a requirement, but it's a very good idea
>regardless of the chemestries, Ni and Li almost require them.
>PbA can get by without a bms, but they last longer with one.
>
>
Maybe multiple full-voltage strings, handled by a single overall BMS,
but with individual sub-string charge equalization, is a good option.
It would require an extreme amount of electronics to parallel four
strings of D cells this way.  On the other hand, Dan Kroushl et al's
experience has been excellent with no sub-string charge balancing at all.

Strings of series cells, by the way, could be packaged in aluminum tubes
for heat conductivity and dissipation, as suggested by O.J. Birkestrand
at RabbitTool.

Two new possibilities:

1.  I just found, in the RabbitTool "NiMH Handbook", an LM cell --
presumably meaning a "long" M cell, as it is the same diameter and 1.6
times the length of an M cell.  It is labeled as 40.5 Ah and 700 grams
(in the same drawing, the M cell is labeled as 25.5Ah and 440 grams vs.
20Ah at 1C rate and 410 grams elsewhere).  These cells, if acceptable
and reasonably priced, are just the right size to use without
paralleling.  They may have the poor power handling of M cells, but it
may be just sufficient, and/or more than sufficient if paralleling with
Toyota's hybrid battery turns out to work.  174 of them would weigh 268
lb plus packaging.  I will check these cells out.

2.  If we do as Dan Kroushl et al have done, and parallel complete
strings of smaller cells without intra-string charge balancing, 8
strings of Suppo C cells might be interesting, as they have 30% higher
specific energy than other sizes, with good power handling.  (Note:
sub-C and D have the highest specific power, 50% higher than C, double
that of F, and triple that of M).  We would get 37.6 Ah (at 1C) for 208
lb plus packaging, and probably around 3 times the power handling
capability of a string of LM cells.  We'd have 8 times the likelihood of
a cell failure, but the failure of a single string would only reduce
capacity by 12%.  This could actually increase reliability if we provide
a way of detecting failures (e.g. by detecting unusually high
differential currents between strings), so they can be repaired before
enough have occurred to cause a serious problem.

>L8r
> Ryan
>
>ps. feel free to share this with the list if you like.
>
>
Thanks.  I am.

--

+++++++++++++++++++++++++++++++++++++++++
         Ron Gremban, rgremban@... <mailto:rgremban@...>
Moderator & Technical Lead, PRIUS+ PHEV Conversion Group
     http://groups.yahoo.com/group/priusplus
<http://groups.yahoo.com/group/priusplus/>
         http://www.priusplus.org
+++++++++++++++++++++++++++++++++++++++++

By the way, before I go on, I would like to thank Tom Driscoll for
coming over yesterday and fixing a couple of physical problems that had
developed in our PbA battery pack, which was getting close to collapse
due to too-weak angle aluminum rails that had bent over time.

What I have learned so far in paralleling higher-voltage PbA packs with
Toyota's hybrid battery -- and using Toyota's Battery ECU -- is that it
doesn't work very well, but that that is probably because the PbA cells
don't have the S-shaped voltage/SOC curve that NiMH cells have.

With a 228V PbA pack, the PbA pack stabilizes at around 40% SOC (-8Ah)
after a grid charge followed by sufficient driving.  However, above
maybe 70% PbA SOC, regen voltage is so high that the THC keeps racing
the ICE.

With a 216V PbA pack (one less 12V battery), the PbA pack is never
discharged below 70-80% SOC.

However, in both cases, battery voltage swings during acceleration and
regen braking are significantly reduced.  I don't yet have enough
experience, but it appears that this paralleling may improve mileage by
up to 6 mpg without grid-charging.  If so, it is a powerful argument for
minimizing overall battery pack internal resistance.

My conclusion is that, with a higher-voltage NiMH battery pack than
Toyota's hybrid battery, either in parallel with or in place of Toyota's
hybrid battery, it may indeed be possible to fool Toyota's Battery ECU
into doing the right things for an effective PRIUS+.  However, we will
be unable to tell for sure short of actually installing and
experimenting with an NiMH battery pack.  The one parameter available to
play with in order to make the Battery ECU do what we want is the
voltage of the new NiMH pack.  If there is a voltage that works, it will
be between 174 and 180 cells in series (209V to 216V nominal).
According to the Cobasys NiMH open-circuit-voltage vs. SOC curve I have,
174 cells is theoretically correct, but give the data from my PbA
experiments, I suspect that the Battery ECU will need a higher voltage
than this to override its Ah integration and keep resetting reported SOC
until the grid charge is depleted.

The problem is that it appears that getting this voltage exactly right
will be critical, and I know of no way to determine it for sure, except
by experimentation, meaning by installing a real NiMH pack, then
adjusting its voltage cell by cell.  On the other hand, ordering a pack
broken into equal-voltage modules to enable the addition of inter-module
charge balancing requires already knowing the voltage of the pack!  We
therefore have a chicken-and-egg problem if we want to attempt
paralleling with Toyota's hybrid battery and/or saving the Battery ECU.

The best solution I can think of is to make large matched-cell modules,
e.g. four modules of 45 cells each.  Then, to lower the cell count below
180, we remove one cell each from two of the modules, then from all
four, then another from two.  If we find a voltage that works and it is
with the same number of cells in all four modules, then we charge
balance them all against each other; if two have a higher voltage than
the other two, we charge balance two-by-two and hope that is good
enough.  A similar alternative is with six modules of 30 cells each.

Another issue:  if we increase the battery pack voltage above the native
168 cells, the engine will race if regen voltage gets too high.  Once
again, however, lower internal resistance may come to the rescue.  It
may be possible to use a voltage of up to around 180 cells if the
internal resistance of our resulting pack is significantly lower than
that of Toyota's hybrid battery alone.  Then the reduced voltage swings
during regen braking should make up for the increase in nominal voltage,
keeping peak voltage no higher than before.

Comments?

P.S.  I will add C and LM cell packs (from my previous message) to my
battery spreadsheet.

--

+++++++++++++++++++++++++++++++++++++++++
         Ron Gremban, rgremban@... <mailto:rgremban@...>
Moderator & Technical Lead, PRIUS+ PHEV Conversion Group
     http://groups.yahoo.com/group/priusplus
<http://groups.yahoo.com/group/priusplus/>
         http://www.priusplus.org
+++++++++++++++++++++++++++++++++++++++++

--- In priusplus@yahoogroups.com, Ron Gremban  wrote:
> Strings of series cells, by the way, could be packaged in
> aluminum tubes for heat conductivity and dissipation.

NO!, Aluminum is conductive, if you flex that tube you are
likely to brake through one or two! of the cells insulation!
Dead shorts inside a sealed tube spells bad news, been there.
( I used copper pipe and melted down a few cells with it, better off )
( using a cardboard, plastic, or some other NON-Conductive tubes!    )

And yes,
A Single BMS that handles multiple long NiMH strings sounds great!

L8r
  Ryan

--- In priusplus@yahoogroups.com, Ron Gremban <rgremban@c...> wrote:
> My conclusion is that, with a higher-voltage NiMH battery pack than
> Toyota's hybrid battery, either in parallel with or in place of
Toyota's
> hybrid battery, it may indeed be possible to fool Toyota's Battery
ECU
> into doing the right things for an effective PRIUS+.  However, we
will
> be unable to tell for sure short of actually installing and
> experimenting with an NiMH battery pack.  The one parameter
available to
> play with in order to make the Battery ECU do what we want is the
> voltage of the new NiMH pack.  If there is a voltage that works, it
will
> be between 174 and 180 cells in series (209V to 216V nominal).
> According to the Cobasys NiMH open-circuit-voltage vs. SOC curve I
have,
> 174 cells is theoretically correct, but give the data from my PbA
> experiments, I suspect that the Battery ECU will need a higher
voltage
> than this to override its Ah integration and keep resetting
reported SOC
> until the grid charge is depleted.

Ron,
I have been experimenting with different voltages and can confirm
your suspicion.  For about 17,000 miles I have been running with 174
cells (209V nominal).  No matter how hard I tried, I could not get
them below 60% SOC in EV mode before the ICE kicked back in to
recharge them.  Lately I have experimented with 180, 186 and 192
cells.  Because of the higher voltage, I have to inject current
intermittantly (switch the paralleling relay on and off) until I have
depleted the grid charged aux packs to a low enough level where it
would be safe to leave the system paralleled.  With 192 cells, I have
been able to depleate the SOC to 30%.

The problem is that the ECU has apparently learned about my
additional capacity and adjusted the regen to direct more energy into
my larger "energy sink".  When you remove that energy sink the regen
voltages spike to dangerous levels, 275V in my case when going down a
long hill.  The ICE races as a result.  So unless you have depleated
the aux packs to an acceptable level before approaching a long hill,
you cannot turn on the energy sink.  This has caused me to go back
down to 180 cells and the regen voltage is still higher but only
about 260V.

>
> The problem is that it appears that getting this voltage exactly
right
> will be critical, and I know of no way to determine it for sure,
except
> by experimentation, meaning by installing a real NiMH pack, then
> adjusting its voltage cell by cell.  On the other hand, ordering a
pack
> broken into equal-voltage modules to enable the addition of inter-
module
> charge balancing requires already knowing the voltage of the pack!
We
> therefore have a chicken-and-egg problem if we want to attempt
> paralleling with Toyota's hybrid battery and/or saving the Battery
ECU.

The arrangement that I am using is extremely easy to configure.  I
use the 7.2V NiMH modules with Anderson Power Pole connectors.  I can
add or remove in 6 cell increments in a matter of 5 minutes.  I could
post some pictures if you would like to see it.  At any rate, I would
agree that if you wanted to work with the Toyota ECU, you should
consider starting at 174 or 180 modules.  In my opinion, the optimal
voltage is in that range.

> Another issue:  if we increase the battery pack voltage above the
native
> 168 cells, the engine will race if regen voltage gets too high.
Once
> again, however, lower internal resistance may come to the rescue.

I can tell you from 17,000 miles of experience that with 174 cells,
my engine has only raced on a few occasions on long down hill
stretches.  This behavior was no different than with the OEM battery
only.

It
> may be possible to use a voltage of up to around 180 cells if the
> internal resistance of our resulting pack is significantly lower
than
> that of Toyota's hybrid battery alone.  Then the reduced voltage
swings
> during regen braking should make up for the increase in nominal
voltage,
> keeping peak voltage no higher than before.

Keep in mind, as I stated earlier that the ECU can learn about your
extra capacity and will increase regen output to take advantage of
it.  I have compelling evidence to prove this.  Because of this your
regen voltage will go higher, not lower.  And if your remove your
additional capacity you will see even higher voltage spikes because
you have taken away the ability to absorb the energy.  Eventually I
assume that the ECU will learn that the extra capacity is gone and
readjust.  But I have no data to support that arguement.

I hope this helps,
Dan

Ryan:

I'm uncomfortable with the D cell concept in any case especially if we
parallel strings.  A good and well thought out battery monitor system with
good temperature data is really needed.  I saw a catastrophic failure of a
5 kwhr parallel strings of Nimhd D cell batt.  Not pretty! and a lot of
explosions like gunshots and smoke!!

I think we should be looking at full batteries of the right capacity like
about 50 to 60 amphrs to form single strings.

Andy Frank

At 08:03 PM 1/25/2005, you wrote:


>--- In priusplus@yahoogroups.com, Ron Gremban  wrote:
> > Strings of series cells, by the way, could be packaged in
> > aluminum tubes for heat conductivity and dissipation.
>
>NO!, Aluminum is conductive, if you flex that tube you are
>likely to brake through one or two! of the cells insulation!
>Dead shorts inside a sealed tube spells bad news, been there.
>( I used copper pipe and melted down a few cells with it, better off )
>( using a cardboard, plastic, or some other NON-Conductive tubes!    )
>
>And yes,
>A Single BMS that handles multiple long NiMH strings sounds great!
>
>L8r
>  Ryan
>
>
>
>
>
>
>
>Yahoo! Groups Links
>
>
>
>

Thanks once again for your messages, Dan.  First, I have a question for
you, then some general discussion for everyone, including some new ideas.

Below, you talk about how far your added pack gets discharged (to 60%
SOC with 174 cells, to 30% with 192 cells).  What method are you using
to discover the SOC of your added pack?  Reliably reading SOC is easier
said than done.

The prospect that Toyota's Battery ECU learns about the capacity of the
battery pack is very interesting, and complicates the process of
learning how to fool it into doing the right things for a larger,
grid-charged battery pack.  For one thing, any system changes being
tested need to be sustained long enough for the Battery ECU to change
its tables and stabilize relative to the new configuration.  This only
adds to the uncertainty of whether or not  the ECU can be enticed into
properly responding to the discontinuity introduced by grid charging, as
it might keep resetting regen capacity to match the discharge capacity
it sees immediately after a grid charge.

Another thing I meant to mention in my previous message is that it may
be necessary to charge a higher voltage pack no higher than 80-90% SOC
to avoid the sharp voltage rise at the high-SOC end of the voltage
curve.  This is consistent with your new grid-charging experiments.  It
does, however, limit usable charge to 60% or less, depending on
attainable low SOC, which still needs to leave room for normal hybrid
operation once charge depletion is complete.

I am beginning to rethink my abandonment of my SCR-controlled battery
paralleling scheme that was our first attempted PRIUS+ configuration.
With the paralleling on the battery side of the current sensor, the
reason it didn't work was because the Battery ECU uses Ah integration to
keep track of battery pack SOC.  However, given what we are now
learning, it may be just the ticket, as a higher-voltage NiMH pack can
cause the ECU to keep resetting its understanding of SOC, and the
SCR-controlled paralleling can limit paralleling to occur (almost) only
when the vehicle is drawing current, thereby helping to avoid both
overcharging of the OEM pack (Toyota's hybrid battery) and overly high
voltage peaks.  Once the added pack has been sufficiently depleted, a
contactor can bypass the SCR, thereby reducing regen internal resistance
and allowing for extended regen, e.g. down a mountain road.

/ron

Dan Kroushl wrote:

>
> --- In priusplus@yahoogroups.com, Ron Gremban <rgremban@c...> wrote:
> > My conclusion is that, with a higher-voltage NiMH battery pack than
> > Toyota's hybrid battery, either in parallel with or in place of
> Toyota's
> > hybrid battery, it may indeed be possible to fool Toyota's Battery
> ECU
> > into doing the right things for an effective PRIUS+.  However, we
> will
> > be unable to tell for sure short of actually installing and
> > experimenting with an NiMH battery pack.  The one parameter
> available to
> > play with in order to make the Battery ECU do what we want is the
> > voltage of the new NiMH pack.  If there is a voltage that works, it
> will
> > be between 174 and 180 cells in series (209V to 216V nominal).
> > According to the Cobasys NiMH open-circuit-voltage vs. SOC curve I
> have,
> > 174 cells is theoretically correct, but give the data from my PbA
> > experiments, I suspect that the Battery ECU will need a higher
> voltage
> > than this to override its Ah integration and keep resetting
> reported SOC
> > until the grid charge is depleted.
>
> Ron,
> I have been experimenting with different voltages and can confirm
> your suspicion.  For about 17,000 miles I have been running with 174
> cells (209V nominal).  No matter how hard I tried, I could not get
> them below 60% SOC in EV mode before the ICE kicked back in to
> recharge them.  Lately I have experimented with 180, 186 and 192
> cells.  Because of the higher voltage, I have to inject current
> intermittantly (switch the paralleling relay on and off) until I have
> depleted the grid charged aux packs to a low enough level where it
> would be safe to leave the system paralleled.  With 192 cells, I have
> been able to depleate the SOC to 30%.
>
> The problem is that the ECU has apparently learned about my
> additional capacity and adjusted the regen to direct more energy into
> my larger "energy sink".  When you remove that energy sink the regen
> voltages spike to dangerous levels, 275V in my case when going down a
> long hill.  The ICE races as a result.  So unless you have depleated
> the aux packs to an acceptable level before approaching a long hill,
> you cannot turn on the energy sink.  This has caused me to go back
> down to 180 cells and the regen voltage is still higher but only
> about 260V.
>
> >
> > The problem is that it appears that getting this voltage exactly
> right
> > will be critical, and I know of no way to determine it for sure,
> except
> > by experimentation, meaning by installing a real NiMH pack, then
> > adjusting its voltage cell by cell.  On the other hand, ordering a
> pack
> > broken into equal-voltage modules to enable the addition of inter-
> module
> > charge balancing requires already knowing the voltage of the pack!
> We
> > therefore have a chicken-and-egg problem if we want to attempt
> > paralleling with Toyota's hybrid battery and/or saving the Battery
> ECU.
>
> The arrangement that I am using is extremely easy to configure.  I
> use the 7.2V NiMH modules with Anderson Power Pole connectors.  I can
> add or remove in 6 cell increments in a matter of 5 minutes.  I could
> post some pictures if you would like to see it.  At any rate, I would
> agree that if you wanted to work with the Toyota ECU, you should
> consider starting at 174 or 180 modules.  In my opinion, the optimal
> voltage is in that range.
>
> > Another issue:  if we increase the battery pack voltage above the
> native
> > 168 cells, the engine will race if regen voltage gets too high.
> Once
> > again, however, lower internal resistance may come to the rescue.
>
> I can tell you from 17,000 miles of experience that with 174 cells,
> my engine has only raced on a few occasions on long down hill
> stretches.  This behavior was no different than with the OEM battery
> only.
>
> It
> > may be possible to use a voltage of up to around 180 cells if the
> > internal resistance of our resulting pack is significantly lower
> than
> > that of Toyota's hybrid battery alone.  Then the reduced voltage
> swings
> > during regen braking should make up for the increase in nominal
> voltage,
> > keeping peak voltage no higher than before.
>
> Keep in mind, as I stated earlier that the ECU can learn about your
> extra capacity and will increase regen output to take advantage of
> it.  I have compelling evidence to prove this.  Because of this your
> regen voltage will go higher, not lower.  And if your remove your
> additional capacity you will see even higher voltage spikes because
> you have taken away the ability to absorb the energy.  Eventually I
> assume that the ECU will learn that the extra capacity is gone and
> readjust.  But I have no data to support that arguement.
>
> I hope this helps,
> Dan
>

+++++++++++++++++++++++++++++++++++++++++
         Ron Gremban, rgremban@... <mailto:rgremban@...>
Moderator & Technical Lead, PRIUS+ PHEV Conversion Group
     http://groups.yahoo.com/group/priusplus
<http://groups.yahoo.com/group/priusplus/>
         http://www.priusplus.org
+++++++++++++++++++++++++++++++++++++++++

--- In priusplus@yahoogroups.com, Ron Gremban <rgremban@c...> wrote:
> Below, you talk about how far your added pack gets discharged (to
60%
> SOC with 174 cells, to 30% with 192 cells).  What method are you
using
> to discover the SOC of your added pack?  Reliably reading SOC is
easier
> said than done.

Well that is a good question.  Maybe it is not as simple as I
thought.  Typically I will take a voltage reading 1 hour after
arriving home from work before charging.  Then I divide the reading
by the number of cells.  The assumption is that 1V is 0% and 1.44V is
100%.  So if I have 1.266v per cell, there SOC is .266/.44 = 60%.  I
realize that calling this the SOC is probably not as accurate as
calling it "Percentage of available voltage" because this is a linear
equation and the discharge characteristics of batteries are curved.

So please, tell me the correct way to reliably read SOC?

Dan

Thanks for the details.  I don't know if there *is* a "correct" way of
measuring SOC; as I said, it is easier said than done.  I think your
resting open-circuit voltage reading is excellent.  There is a graph on
page 4 of the Cobasys 1000 brochure at
http://cobasys.com/pdf/transportation/Series%201000%20Brochure.pdf that
relates voltage to SOC.  You can use it to translate voltage to SOC.
Your 1.27V reading would translate to around 40% SOC.  Your previous 30%
reading presumably was 1.132V/cell; on the graph that translates to
essentially 0% SOC!

Though this is at 35 deg C and specifically for their cells, I believe
it is pretty much characteristic of NiMH cells.  Again, I am uncertain,
but from what I have gleaned, the temperature coefficient of mid-SOC
NiMH open-circuit voltage is insignificant, though at lower temperatures
the S of the curve becomes more pronounced (the endpoint voltages lower
and higher).

I just uploaded a spreadsheet to Files/Battery
Info/NiMHPackOCVoltages.xls indicating open-circuit voltages vs. SOC for
168, 174, 175, and 180-cell battery packs, based on data from the
Cobasys graph (I used midpoints between the curves for new and old
batteries).  It is this graph that I used to determine that,
theoretically, a 174-cell pack's 25% SOC open-circuit voltage would
match that of the (168-cell) OEM pack at 60% SOC, mid-range for the Prius.

/ron

Dan Kroushl wrote:

>
> --- In priusplus@yahoogroups.com, Ron Gremban <rgremban@c...> wrote:
> > Below, you talk about how far your added pack gets discharged (to
> 60%
> > SOC with 174 cells, to 30% with 192 cells).  What method are you
> using
> > to discover the SOC of your added pack?  Reliably reading SOC is
> easier
> > said than done.
>
> Well that is a good question.  Maybe it is not as simple as I
> thought.  Typically I will take a voltage reading 1 hour after
> arriving home from work before charging.  Then I divide the reading
> by the number of cells.  The assumption is that 1V is 0% and 1.44V is
> 100%.  So if I have 1.266v per cell, there SOC is .266/.44 = 60%.  I
> realize that calling this the SOC is probably not as accurate as
> calling it "Percentage of available voltage" because this is a linear
> equation and the discharge characteristics of batteries are curved.
>
> So please, tell me the correct way to reliably read SOC?
>
> Dan
>

+++++++++++++++++++++++++++++++++++++++++
         Ron Gremban, rgremban@... <mailto:rgremban@...>
Moderator & Technical Lead, PRIUS+ PHEV Conversion Group
     http://groups.yahoo.com/group/priusplus
<http://groups.yahoo.com/group/priusplus/>
         http://www.priusplus.org
+++++++++++++++++++++++++++++++++++++++++

--- In priusplus@yahoogroups.com, Ron Gremban <rgremban@c...> wrote:
> Thanks for the details.  I don't know if there *is* a "correct" way
of
> measuring SOC; as I said, it is easier said than done.  I think
your
> resting open-circuit voltage reading is excellent.  There is a
graph on
> page 4 of the Cobasys 1000 brochure at
> http://cobasys.com/pdf/transportation/Series%201000%20Brochure.pdf
that
> relates voltage to SOC.  You can use it to translate voltage to
SOC.
> Your 1.27V reading would translate to around 40% SOC.  Your
previous 30%
> reading presumably was 1.132V/cell; on the graph that translates to
> essentially 0% SOC!
>
> Though this is at 35 deg C and specifically for their cells, I
believe
> it is pretty much characteristic of NiMH cells.  Again, I am
uncertain,
> but from what I have gleaned, the temperature coefficient of mid-
SOC
> NiMH open-circuit voltage is insignificant, though at lower
temperatures
> the S of the curve becomes more pronounced (the endpoint voltages
lower
> and higher).
>
> I just uploaded a spreadsheet to Files/Battery
> Info/NiMHPackOCVoltages.xls indicating open-circuit voltages vs.
SOC for
> 168, 174, 175, and 180-cell battery packs, based on data from the
> Cobasys graph (I used midpoints between the curves for new and old
> batteries).  It is this graph that I used to determine that,
> theoretically, a 174-cell pack's 25% SOC open-circuit voltage would
> match that of the (168-cell) OEM pack at 60% SOC, mid-range for the
Prius.
>
> /ron

Excellent Info Ron!  This also shows that changing back down to 180
cells was the right thing to do.  I started this morning at 250V of
100%.  I just checked the voltage and they are at 234V or 70%.  I
will shoot for 228V, 40% by the time I get home.

Dan

At the moment, because we have so far been unable to find cells of the
right size (30-40 Ah) available on the retail market and for other
reasons described below, it looks like our best chance to quickly get
our first reasonably safe and effective NiMH PRIUS+ battery pack is to
order a set of only slightly custom 900-9Dx36 modules from RabbitTool.
This module consists of four parallel strings of 8 D cells, in four
electrically-insulated aluminum tubes for effective, even air cooling.

So far, among our potential suppliers, only RabbitTool appears to have
had good experience with NiMH paralleling, done at the module level
(each module consisting of several strings in parallel) -- and even they
have only dozens of modules in use.  Though such modules are still
sub-optimum, a shorted cell can blow up only one module, and, on the
other hand, cell-to-cell capacity variations for charging should be more
or less averaged out in each string.

It appears that it is dangerous to parallel long strings of cells (see
below and Andy Frank's posting of 1/26).  Also, paralleling NiMH cells
at the cell level gives very poor results, as not all cells are likely
to get fully charged, and others will overheat, due voltage peaking
before full charge.  See Ryan's  (d0li0's) quote in my posting of 1/25;
Jim B at Mesa Power has had this problem, too:

  > More importantly is the SOC curves of the various chemestries.
  > PbA and Li-ion both have fairly linear curves, so you can parallel
  > them and they pretty much stay at equal voltage and SOC...
  > I have a pack with 12 parallel Li-ion cells and it performes
  > quite well, ACP also uses 68P blocks of Li-ion cells...
  >
  > However this is quite different from Ni based cells which have a
  > peak and then a voltage drop as they reach full charge.  This
  > characteristic makes for trouble when cells are in parallel.
  >
  > Imagine you have 3 NiMH cells in parallel, for some reason
  > cell 1 reaches full first, It's voltage peakes and then falls.
  > Now the other two cells haven't reached their peak, and they
  > never will because all the charge current is now going to flow
  > into the fully charged cell, causing it to heat up and fail.

Two options are left:  finding cells of the right size and paralleling
cells module by module.

So far, we have found only one right-sized cell on the retail market,
the LM (presumably meaning Long M) cell, 40Ah, available from
RabbitTool.  However, it has very marginal high-power capabilities and
may not even be available yet.  Our other option is finding something
available only to OEM customers and/or not in volume production.  We
have one likelihood we are following up on from Electro Energy.  We are
also developing contacts at Cobasys and Saft, but expect this to take a
while.  We therefore cannot depend on quickly getting right-sized cells.

More about NiMH series and parallel strings:

Last night I had a conversation with Andy Frank.  The following is a
synopsis about the catastrophic failure he reported (correct me, Andy,
if I am misrepresenting you):

Another university's car had four 300V strings of D cells wired in
parallel.  What happened is that one cell in one string shorted out,
thereby lowering the voltage of the string, causing current to flow in
from the other strings, overheating the bad cell and adjacent cells,
etc, until cells were exploding all over the place, threatening the
vehicle and everyone near it.  It appears that this is an everpresent
danger of parallel strings of cells, independent of chemistry.

Andy also indicated that they are having good results with series
strings of NiMH modules, where each module consists of well-matched
cells.  During use, the voltage of each module is monitored to ensure
that no module is going bad and to end discharge when the worst module's
voltage becomes marginal.  During charge, the temperature of each module
is likewise monitored, and charging is halted before any module gets
above 35-40 deg C.  No other charge balancing mechanism is used, as it
is tricky to design such a mechanism for NiMH cells due to their S
voltage/SOC curve and voltage peak before full charge.

Also, I got an email message today indicating the following (I have not
yet received an O.K. to name the author, so am for now posting this
anonymously):

>I'm not a member of the Prius+ Yahoo group, but I do lurk there. I'm
>an electronic tech of 30 years and I have some experience charging NiMH.
>
>IMO, charging two parallel strings is tricky, charging 5 stings would
>be unwieldy. Current "hogging" due to the neg delta V charge
>characteristic
>becomes a serious problem with charge currents >0.1C (of one cell). I
>don't think BMS at the cell level is practical at the cell level for
>clusters of parallel cells. You'd have to monitor either temperature
>or current of each cell to detect imbalance.
>
>My own experience parallel charging two strings of cells: They can be
>charged agressively (C) up to 1.35V per (room temp) cell with very
>little heating and no voltage foldback. This will fill the cells to
>60-75% SOC. Above this level charge at 0.05C (one cell) to avoid
>current hogging and overheating even if it does. Full charge in about
>8 hours.
>
--

+++++++++++++++++++++++++++++++++++++++++
         Ron Gremban, rgremban@... <mailto:rgremban@...>
Moderator & Technical Lead
PRIUS+ PHEV Conversion Group
     http://groups.yahoo.com/group/priusplus
<http://groups.yahoo.com/group/priusplus/>
         http://www.priusplus.org
+++++++++++++++++++++++++++++++++++++++++

Ron,

I must be missing something.

Can't you just put a voltage comparator on each cell? A signal that any cell
is so overvoltage that is likely to catastrophically fail could trip a
breaker, or relay or..., and remove this string from the charging circuit.

You might want this to abort the charging cycle. Seems a better option than
having a cascading failure with "cells were exploding all over the place,
threatening the vehicle and everyone near it".

-Leonard


----- Original Message -----
From: "Ron Gremban" <rgremban@...>
To: <priusplus@yahoogroups.com>
Sent: Friday, February 04, 2005 10:24 PM
Subject: [priusplus] Current battery strategy, comments please


>
> At the moment, because we have so far been unable to find cells of the
> right size (30-40 Ah) available on the retail market and for other
> reasons described below, it looks like our best chance to quickly get
> our first reasonably safe and effective NiMH PRIUS+ battery pack is to
> order a set of only slightly custom 900-9Dx36 modules from RabbitTool.
> This module consists of four parallel strings of 8 D cells, in four
> electrically-insulated aluminum tubes for effective, even air cooling.
>
> So far, among our potential suppliers, only RabbitTool appears to have
> had good experience with NiMH paralleling, done at the module level
> (each module consisting of several strings in parallel) -- and even they
> have only dozens of modules in use.  Though such modules are still
> sub-optimum, a shorted cell can blow up only one module, and, on the
> other hand, cell-to-cell capacity variations for charging should be more
> or less averaged out in each string.
>
> It appears that it is dangerous to parallel long strings of cells (see
> below and Andy Frank's posting of 1/26).  Also, paralleling NiMH cells
> at the cell level gives very poor results, as not all cells are likely
> to get fully charged, and others will overheat, due voltage peaking
> before full charge.  See Ryan's  (d0li0's) quote in my posting of 1/25;
> Jim B at Mesa Power has had this problem, too:
>
> > More importantly is the SOC curves of the various chemestries.
> > PbA and Li-ion both have fairly linear curves, so you can parallel
> > them and they pretty much stay at equal voltage and SOC...
> > I have a pack with 12 parallel Li-ion cells and it performes
> > quite well, ACP also uses 68P blocks of Li-ion cells...
> >
> > However this is quite different from Ni based cells which have a
> > peak and then a voltage drop as they reach full charge.  This
> > characteristic makes for trouble when cells are in parallel.
> >
> > Imagine you have 3 NiMH cells in parallel, for some reason
> > cell 1 reaches full first, It's voltage peakes and then falls.
> > Now the other two cells haven't reached their peak, and they
> > never will because all the charge current is now going to flow
> > into the fully charged cell, causing it to heat up and fail.
>
> Two options are left:  finding cells of the right size and paralleling
> cells module by module.
>
> So far, we have found only one right-sized cell on the retail market,
> the LM (presumably meaning Long M) cell, 40Ah, available from
> RabbitTool.  However, it has very marginal high-power capabilities and
> may not even be available yet.  Our other option is finding something
> available only to OEM customers and/or not in volume production.  We
> have one likelihood we are following up on from Electro Energy.  We are
> also developing contacts at Cobasys and Saft, but expect this to take a
> while.  We therefore cannot depend on quickly getting right-sized cells.
>
> More about NiMH series and parallel strings:
>
> Last night I had a conversation with Andy Frank.  The following is a
> synopsis about the catastrophic failure he reported (correct me, Andy,
> if I am misrepresenting you):
>
> Another university's car had four 300V strings of D cells wired in
> parallel.  What happened is that one cell in one string shorted out,
> thereby lowering the voltage of the string, causing current to flow in
> from the other strings, overheating the bad cell and adjacent cells,
> etc, until cells were exploding all over the place, threatening the
> vehicle and everyone near it.  It appears that this is an everpresent
> danger of parallel strings of cells, independent of chemistry.
>
> Andy also indicated that they are having good results with series
> strings of NiMH modules, where each module consists of well-matched
> cells.  During use, the voltage of each module is monitored to ensure
> that no module is going bad and to end discharge when the worst module's
> voltage becomes marginal.  During charge, the temperature of each module
> is likewise monitored, and charging is halted before any module gets
> above 35-40 deg C.  No other charge balancing mechanism is used, as it
> is tricky to design such a mechanism for NiMH cells due to their S
> voltage/SOC curve and voltage peak before full charge.
>
> Also, I got an email message today indicating the following (I have not
> yet received an O.K. to name the author, so am for now posting this
> anonymously):
>
>>I'm not a member of the Prius+ Yahoo group, but I do lurk there. I'm
>>an electronic tech of 30 years and I have some experience charging NiMH.
>>
>>IMO, charging two parallel strings is tricky, charging 5 stings would
>>be unwieldy. Current "hogging" due to the neg delta V charge
>>characteristic
>>becomes a serious problem with charge currents >0.1C (of one cell). I
>>don't think BMS at the cell level is practical at the cell level for
>>clusters of parallel cells. You'd have to monitor either temperature
>>or current of each cell to detect imbalance.
>>
>>My own experience parallel charging two strings of cells: They can be
>>charged agressively (C) up to 1.35V per (room temp) cell with very
>>little heating and no voltage foldback. This will fill the cells to
>>60-75% SOC. Above this level charge at 0.05C (one cell) to avoid
>>current hogging and overheating even if it does. Full charge in about
>>8 hours.
>>
> --
>
> +++++++++++++++++++++++++++++++++++++++++
>        Ron Gremban, rgremban@... <mailto:rgremban@...>
> Moderator & Technical Lead
> PRIUS+ PHEV Conversion Group
>    http://groups.yahoo.com/group/priusplus
> <http://groups.yahoo.com/group/priusplus/>
>        http://www.priusplus.org
> +++++++++++++++++++++++++++++++++++++++++
>
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
>
>

I forgot the following:

I reconfigured my battery spreadsheet to include two modules using
paralleling, both by RabbitTool, and both using insulated metal tubes to
promote air cooling with as good as possible temperature consistency.
The module with four parallel strings of D cells has excellent
high-power capabilities; the one with two parallel strings of M cells
has marginal power handling capabilities but fewer cells in parallel.  I
also show a module with a single string of M cells (too little capacity
and even worse power handling), and one with a single string of LM
cells.  The spreadsheet is at Files/Battery
Info/PriusPlusBatteries050204rdg.{pdf,xls}, and will soon be mirrored at
www.priusplus.com.

Also, Leonard Tramiel wrote:

> Ron,
>
> I must be missing something.
>
> Can't you just put a voltage comparator on each cell? A signal that
> any cell
> is so overvoltage that is likely to catastrophically fail could trip a
> breaker, or relay or..., and remove this string from the charging circuit.
>
> You might want this to abort the charging cycle. Seems a better option
> than
> having a cascading failure with "cells were exploding all over the place,
> threatening the vehicle and everyone near it".
>
> -Leonard

With nearly 700 cells in a D-cell pack, this is quite unwieldly.  Even a
single NiMH series string contains around 170 cells.  Each would require
separate, isolated electronics, as well as wiring to the electronics
board(s) -- and there are high relative voltages involved.  This level
of monitoring and control IS REQUIRED by Li-ion packs, and is a major
difficulty and expense, even though Li-ion cells can be paralleled at
the cell level and, with their higher voltage, a PRIUS+ requires only
around 56 in series.

+++++++++++++++++++++++++++++++++++++++++
     Ron Gremban, rgremban@... <mailto:rgremban@...>
         Moderator & Technical Lead
         PRIUS+ PHEV Conversion Group
             http://groups.yahoo.com/group/priusplus
<http://groups.yahoo.com/group/priusplus/>
             http://www.priusplus.org
+++++++++++++++++++++++++++++++++++++++++

--- In priusplus@yahoogroups.com, Ron Gremban <rgremban@c...> wrote:
> >My own experience parallel charging two strings of cells: They can
be
> >charged agressively (C) up to 1.35V per (room temp) cell with very
> >little heating and no voltage foldback. This will fill the cells to
> >60-75% SOC. Above this level charge at 0.05C (one cell) to avoid
> >current hogging and overheating even if it does. Full charge in
about
> >8 hours.

Ron,
I find this very interesting.  Specifically the reference to SOC.  In
the spreadsheet that you sent me, a 1.35V cell is at 90% SOC.  My
method of determining SOC puts 1.35V at 80% SOC.  Now a third person
has the opinion that 1.35V is in the 60-75% SOC range.  I would like
to hear more from this un-named source.  He seems to have some
valuable insight that we all could use.


In any event, based on the comment above, I should not have much
problem.  I am charging each parallel string at about 0.55A (0.18C)
each up to 243V, which happens to be 1.35V per cell.

Dan

Ron:  Your comments look good to me.

I think you all recognize the problem with Ni cells.

Our approach of monitoring temperature and voltage at least on a 12 volt
module is one way to avoid disaster!


Andy


At 10:24 PM 2/4/2005, you wrote:

>At the moment, because we have so far been unable to find cells of the
>right size (30-40 Ah) available on the retail market and for other
>reasons described below, it looks like our best chance to quickly get
>our first reasonably safe and effective NiMH PRIUS+ battery pack is to
>order a set of only slightly custom 900-9Dx36 modules from RabbitTool.
>This module consists of four parallel strings of 8 D cells, in four
>electrically-insulated aluminum tubes for effective, even air cooling.
>
>So far, among our potential suppliers, only RabbitTool appears to have
>had good experience with NiMH paralleling, done at the module level
>(each module consisting of several strings in parallel) -- and even they
>have only dozens of modules in use.  Though such modules are still
>sub-optimum, a shorted cell can blow up only one module, and, on the
>other hand, cell-to-cell capacity variations for charging should be more
>or less averaged out in each string.
>
>It appears that it is dangerous to parallel long strings of cells (see
>below and Andy Frank's posting of 1/26).  Also, paralleling NiMH cells
>at the cell level gives very poor results, as not all cells are likely
>to get fully charged, and others will overheat, due voltage peaking
>before full charge.  See Ryan's  (d0li0's) quote in my posting of 1/25;
>Jim B at Mesa Power has had this problem, too:
>
>  > More importantly is the SOC curves of the various chemestries.
>  > PbA and Li-ion both have fairly linear curves, so you can parallel
>  > them and they pretty much stay at equal voltage and SOC...
>  > I have a pack with 12 parallel Li-ion cells and it performes
>  > quite well, ACP also uses 68P blocks of Li-ion cells...
>  >
>  > However this is quite different from Ni based cells which have a
>  > peak and then a voltage drop as they reach full charge.  This
>  > characteristic makes for trouble when cells are in parallel.
>  >
>  > Imagine you have 3 NiMH cells in parallel, for some reason
>  > cell 1 reaches full first, It's voltage peakes and then falls.
>  > Now the other two cells haven't reached their peak, and they
>  > never will because all the charge current is now going to flow
>  > into the fully charged cell, causing it to heat up and fail.
>
>Two options are left:  finding cells of the right size and paralleling
>cells module by module.
>
>So far, we have found only one right-sized cell on the retail market,
>the LM (presumably meaning Long M) cell, 40Ah, available from
>RabbitTool.  However, it has very marginal high-power capabilities and
>may not even be available yet.  Our other option is finding something
>available only to OEM customers and/or not in volume production.  We
>have one likelihood we are following up on from Electro Energy.  We are
>also developing contacts at Cobasys and Saft, but expect this to take a
>while.  We therefore cannot depend on quickly getting right-sized cells.
>
>More about NiMH series and parallel strings:
>
>Last night I had a conversation with Andy Frank.  The following is a
>synopsis about the catastrophic failure he reported (correct me, Andy,
>if I am misrepresenting you):
>
>Another university's car had four 300V strings of D cells wired in
>parallel.  What happened is that one cell in one string shorted out,
>thereby lowering the voltage of the string, causing current to flow in
>from the other strings, overheating the bad cell and adjacent cells,
>etc, until cells were exploding all over the place, threatening the
>vehicle and everyone near it.  It appears that this is an everpresent
>danger of parallel strings of cells, independent of chemistry.
>
>Andy also indicated that they are having good results with series
>strings of NiMH modules, where each module consists of well-matched
>cells.  During use, the voltage of each module is monitored to ensure
>that no module is going bad and to end discharge when the worst module's
>voltage becomes marginal.  During charge, the temperature of each module
>is likewise monitored, and charging is halted before any module gets
>above 35-40 deg C.  No other charge balancing mechanism is used, as it
>is tricky to design such a mechanism for NiMH cells due to their S
>voltage/SOC curve and voltage peak before full charge.
>
>Also, I got an email message today indicating the following (I have not
>yet received an O.K. to name the author, so am for now posting this
>anonymously):
>
> >I'm not a member of the Prius+ Yahoo group, but I do lurk there. I'm
> >an electronic tech of 30 years and I have some experience charging NiMH.
> >
> >IMO, charging two parallel strings is tricky, charging 5 stings would
> >be unwieldy. Current "hogging" due to the neg delta V charge
> >characteristic
> >becomes a serious problem with charge currents >0.1C (of one cell). I
> >don't think BMS at the cell level is practical at the cell level for
> >clusters of parallel cells. You'd have to monitor either temperature
> >or current of each cell to detect imbalance.
> >
> >My own experience parallel charging two strings of cells: They can be
> >charged agressively (C) up to 1.35V per (room temp) cell with very
> >little heating and no voltage foldback. This will fill the cells to
> >60-75% SOC. Above this level charge at 0.05C (one cell) to avoid
> >current hogging and overheating even if it does. Full charge in about
> >8 hours.
> >
>--
>
>+++++++++++++++++++++++++++++++++++++++++
>         Ron Gremban, rgremban@... <mailto:rgremban@...>
>Moderator & Technical Lead
>PRIUS+ PHEV Conversion Group
>     http://groups.yahoo.com/group/priusplus
><http://groups.yahoo.com/group/priusplus/>
>         http://www.priusplus.org
>+++++++++++++++++++++++++++++++++++++++++
>
>
>
>
>
>Yahoo! Groups Links
>
>
>
>

--- In priusplus@yahoogroups.com, "Dan Kroushl" <danielbkroushl@e...>
wrote:
>
> --- In priusplus@yahoogroups.com, Ron Gremban <rgremban@c...> wrote:
> > >My own experience parallel charging two strings of cells: They can
> be
> > >charged agressively (C) up to 1.35V per (room temp) cell with very
> > >little heating and no voltage foldback. This will fill the cells to
> > >60-75% SOC. Above this level charge at 0.05C (one cell) to avoid
> > >current hogging and overheating even if it does. Full charge in
> about
> > >8 hours.
>
> Ron,
> I find this very interesting.  Specifically the reference to SOC.  In
> the spreadsheet that you sent me, a 1.35V cell is at 90% SOC.  My
> method of determining SOC puts 1.35V at 80% SOC.  Now a third person
> has the opinion that 1.35V is in the 60-75% SOC range.  I would like
> to hear more from this un-named source.  He seems to have some
> valuable insight that we all could use.

I'm the un-named source who wrote to Ron earlier. Perhaps the various
figures for SOC vs cell voltage are due partly to my not specifying
the conditions the cell voltage is measured. My 1.35V is observed
while the cell is being charged at 0.05C, more on that later. I agree
a resting cell at 1.35V is 80-90% full. I've been testing a charging
profile as follows: constant voltage of 1.35V per cell, current
limited to C. As the cell fills the current falls to 0.05C, then
charging continues with constant current at 0.05C. The figures of
1.35V and 60-75% SOC correspond to the point where the charger
switches from CV to CC. If charging stops at this point the cell
voltage drops below 1.35V.

To get this charge profile I simply connect a constant voltage source
in parallel with a constant current source, with the requirement that
the CV source can tolerate its output voltage rising higher than its
set point. As the cell charges the CV source current drops and the CC
source takes over. I've had good results with single strings as well
as parallel strings.

I also have an idea for the Prius+:

Four (electrically) separate strings of D cells. A four channel grid
charger. Charge the strings separately. Use power MOSFET's or IGBT's
to connect the strings in parallel when driving. Lock out the parallel
connection if there's too much variance in string voltages (due to a
bad cell). Fuse each string to avoid a cross current disaster. Set up
the control circuitry so only the original Prius pack gets charged
from regen braking. This could simply be a diode between the orig pack
and the second pack. Somehow trick the Prius SOC tracking.

Tim

All:

Tim has a good idea of a solution to the problem.  I'm not sure I like the
idea of paralleling the original Prius pack and a lot of strings of small
cells.

Using fuses and IGBT's to do the paralleling is a good idea but the proper
feedback sensing is needed for control. I think it illustrates the
complexity of the problem which disappears with the proper cell sizes.   In
other words, a custom cell size that has the right power and energy density
for the objectives is the right way to go.  Adapting production batteries
designed for other applications to our application, always leads to
complexities that are unnecessary, if the right batteries are specified in
the first place!

The Plug-In Hybrid vehicles we have built, have always been done with
batteries that meet or specifications.  Unfortunately sometimes it has cost
us a lot of money but the results are generally good.  For Prius Plus I
suggest we keep looking for a manufacturer who may be willing to work with
us or we find an appropriate manufactured battery.  The Valence battery
looks good to me!  They have a 40amphr 12 volt module that looks about
right.  They are Lithium technology but at 90 whrs/kg and appropriate power
density it looks very attractive!!

Check them out!!

Andy

At 06:17 PM 2/5/2005, you wrote:


>--- In priusplus@yahoogroups.com, "Dan Kroushl" <danielbkroushl@e...>
>wrote:
> >
> > --- In priusplus@yahoogroups.com, Ron Gremban <rgremban@c...> wrote:
> > > >My own experience parallel charging two strings of cells: They can
> > be
> > > >charged agressively (C) up to 1.35V per (room temp) cell with very
> > > >little heating and no voltage foldback. This will fill the cells to
> > > >60-75% SOC. Above this level charge at 0.05C (one cell) to avoid
> > > >current hogging and overheating even if it does. Full charge in
> > about
> > > >8 hours.
> >
> > Ron,
> > I find this very interesting.  Specifically the reference to SOC.  In
> > the spreadsheet that you sent me, a 1.35V cell is at 90% SOC.  My
> > method of determining SOC puts 1.35V at 80% SOC.  Now a third person
> > has the opinion that 1.35V is in the 60-75% SOC range.  I would like
> > to hear more from this un-named source.  He seems to have some
> > valuable insight that we all could use.
>
>I'm the un-named source who wrote to Ron earlier. Perhaps the various
>figures for SOC vs cell voltage are due partly to my not specifying
>the conditions the cell voltage is measured. My 1.35V is observed
>while the cell is being charged at 0.05C, more on that later. I agree
>a resting cell at 1.35V is 80-90% full. I've been testing a charging
>profile as follows: constant voltage of 1.35V per cell, current
>limited to C. As the cell fills the current falls to 0.05C, then
>charging continues with constant current at 0.05C. The figures of
>1.35V and 60-75% SOC correspond to the point where the charger
>switches from CV to CC. If charging stops at this point the cell
>voltage drops below 1.35V.
>
>To get this charge profile I simply connect a constant voltage source
>in parallel with a constant current source, with the requirement that
>the CV source can tolerate its output voltage rising higher than its
>set point. As the cell charges the CV source current drops and the CC
>source takes over. I've had good results with single strings as well
>as parallel strings.
>
>I also have an idea for the Prius+:
>
>Four (electrically) separate strings of D cells. A four channel grid
>charger. Charge the strings separately. Use power MOSFET's or IGBT's
>to connect the strings in parallel when driving. Lock out the parallel
>connection if there's too much variance in string voltages (due to a
>bad cell). Fuse each string to avoid a cross current disaster. Set up
>the control circuitry so only the original Prius pack gets charged
>from regen braking. This could simply be a diode between the orig pack
>and the second pack. Somehow trick the Prius SOC tracking.
>
>Tim
>
>
>
>
>
>
>
>Yahoo! Groups Links
>
>
>
>

Ron,

May I ask if you had any luck talking to Harding Energy? They seem
to have 20Ah cells also. They are not that powerful as D cells -
about 5C only (100A continuous discharge), but that may be fine in
some configurations, as I understand.

http://hardingenergy.com/pdfs/nimh/HS-M20000.pdf

Good luck.

The best solution I've heard yet for Parallel NiMH is here:
http://autos.groups.yahoo.com/group/priusplus/message/302
- Seperate Isolated charging of series strings.
- Parallel only durring discharge
- Each string fused for catastrophic "current dumping" failures

I like the Valence solution even more, they appear to be the only
"off the shelf" Li-ion solution with built in BMS management.
But they sure aren't the cheapest solution.

Have you considered Wet Ni-Cd cells, such as SAFT or bb600 Aircraft
cells? I know these have been used in various BEV projects.  These
cells seem to be very reliable, longlived, and parallelable.  You
can sometimes find them used from surplus sources...
http://autos.groups.yahoo.com/group/ev-list-archive/message/32700
http://autos.groups.yahoo.com/group/ev-list-archive/message/32209

L8r
  Ryan