Archives

The dreaded recalibrations that many Insight owners have to live with until the code of death, may in theory be fixable.
My theory has always been that one subpack in the series string runs out of charge before the others, which the voltage monitoring taps pick up, and the BCM then forces a recalibration, to get away from the cell reversal that would come soon after. The problem is that a series string that must charge in series will never rebalance the low subpack, especially since the SOC is limited to 80%.
With a lead acid series string, one can just keep charging to 100%, and let it float, eventually the weak battery will charge to 100% and become equal to the others.

If we could locate the low capacity/ or low SOC subpack within the recalibrating pack, and charge only that subpack, bringing the SOC higher than the rest, we could potentially rebalance the pack, and it could take months to drift back into the recalibrating zone, or it could even permanently fix the problem.
The whole pack must be pulled out to do the rebalancing and test, but a permanent connection to the weaker subpack could be brought to the top of the pack for subsequent rebalancing withour pack removal if nesessary
The subpacks end connections are easily accessed without opening the pack, by removing the relay/current monitoring end assembly. And the plastic cover on the other end. A simple voltmeter test if the pack is at a stable temperature should give a relative SOC measurement, and the charge can be applied at the same point.
I will do some experiments this weekend to try and determine if this actually works. A friend with an Insight that recalibrates reliably at > 60% SOC has volunteered his pack for the test.
Any opinions ?
;)

This is interesting. Several hybrid owners have had to have their packs replaced. I always think to myself "all the batteries couldn't have gone bad at once". I bet in many cases it was only one or two cells of the pack. This makes me wonder why there is no attempt at the dealer to find the offending cells (very easy with just a voltmeter) and replace them? Aren't the service techs supposed to be trained in this? This is stuff an electronics amateur could handle.

I think the batteries are matched by capacity or something someone once said "cap matched" but I'm only guessing its capacity. So I don't think you can just throw in a few new cells.

So make the battery manufacturer use a binning technique. They probably do that already to make pack assembly easier.

Hi Sledge:

___When you walk into your local Best Buy, do you see the NiMH C cel’s binned per CAP? Unfortunately, we do not :( When Sanyo/Panasonic builds a pack for an OEM, they are making a fair share of bucks for the package and can afford to cap match each and every cel. For us commoners? We rank anywhere from the level of dirt to the dirt under it ;)

___Good Luck

___Wayne

What I'm trying to say is that this sort of repair could easily be handled by the dealer. Maybe a properly binned battery would have to be ordered, but they could do it easily through Honda corporate.

Hi Sledge:

___That would be nice except we would be trying to cap match a single cel or two to a used pack w/ less cap then anything off the factory floor. I think this is why Honda chooses to replace the entire pack assembly when they do an Insight pack replacement after the dreaded IMA code of death?

___Mike might be onto something here …

___Good Luck

___Wayne

I think you guys may be missing my point.
The cells in the pack may all be good, just unbalanced, compared to the rest of the pack and with no balancing hardware in place, the low subpack/cell can never get recharged to the level of the other cells, since the higher SOC subpacks will reach the 80% SOC point and stop the charging, before the low SOC one can ever catch up. It can not catch up, therefore the repeated recals at about the same SOC. Highwaters repeatable recal at >4 bars from the top, is a good example.

An anology:
Think of it as 20 glasses of water. all filled and emptied at the same rate. If one starts with 20% less than all the others, and the filling stops when one gets full, the glass with 20% less will always run out first if they are emptied at the same rate. The only way to correct the imbalance is to fill the one that is low to the same level as the rest.

Charge up the low subpack/cell, and possibily the recals will go away??
At this point only a theory based on the facts we have to work with.
Things that can cause an imbalance?
Temperature differences within the pack can change a cells ability to acquire a charge under the same current as a different temperature cell at a different part of the pack.
Lower starting charge when the pack was new.
Assist at low temps where the cells cannot output full current, with some subpacks heating faster than the others.
High temperatures causing reduced output from one section of the hot pack, while the others are not hotenough to be limited.

My point is that for whatever reason the imbalance occurs, weither it be lower capacity to start with (not likely since this would rapidly accumulate with the series charging to the point where the low capacity cell would continue to loose ground and fail quickly)
or some other occasional condition, there is no known mechanism in place in the car that can get the low SOC subpack/cell rebalanced with the rest.
An active subpack balancer or an occasional recharge of only the low SOC subpack could in theory fix it?
There is of course many other mechanisms of failure that could come into play including a bad cell or low capacity cell, and thats where the code of death takes over?
It is worth the several hours of tinkering to try out my theory. If I am correct, we may be able to come up with a relitively simple rebalancing technique that will restore reasonable performance to a pack that is acting up but not enopugh to set the code?
Honda like Toyota is concerned with liability, and asking the techs to open and rebalance a pack would expose them to much more liability than just swaping the pack and doing the repair at some other location with highly trained people doing the rebuilding/rebalancing.

We may know more by Sunday evening.;)

Hi Mike:

___Keep us informed … The Insight crowd in particular would love to hear of this fix. Imagine never having to go through a recal event when you least expect it ever again! I know of at least one or two or three Insight’ers that would love to see the fix implemented ASAP!

___Good Luck

___Wayne

I think some type of balancing would work well....

Unless I am missing something ... what you would need to do is just check the voltages of the sub-packs and when one of them drops bellow the others you supply a charge to that one to bring it back up to where the others are.

A stock pack might be able to be modified so that a small wire can stay attached to the subpacks while it is back in possition in the car.... you don't need much of a wire for this as the balancing charge shouldn't need to give more than a few milliamps a few hundered milliamps at most.... so a little circuit that goes along and checks the voltages and looks for the lowest one... aplies a small balancing charge just to that sub-pack for a little bit and then checks the sub-packs again to find the lowest one.

a more acurate method would be to put a current and voltage sensor on each subpack to keep track of the amps in vs amps out plus the current voltage... but the straight voltage would be much much simplier and should do 80 to 90% as good of a job as the other.

but I am no battery expert so if anyone is ... by all means speak up.

Mike, I agree with your assessment and you've done a great job of explaining it. I assume you'll be injecting a constant current while monitoring the subpack voltage - what sort of milliamps/amps were you thinking about for this equalizing/balancing current and will you be monitoring the subpack temperature as well? Best of luck to you with this experiment.

It would have been nice if the packs were designed with per-CELL
monitoring/balancing capability. I think laptop packs *try* to
do this -- they're often a series/parallel arrangement of Li
cells and they monitor each junction to make sure no cell group
gets under, what is it, 2.4V where damage occurs... but I don't
think that even there they try to do any smart charging/balancing.
But it's a well-known approach. Some guy was making little
balancing widgets for PbA strings in EVs and storage for home PV
systems -- very simple, a circuit would sit across each 6V or 12V
module and if its float-charge got above a certain point, it
would start blinking an LED -- each blink was a certain quantity
of charge bleed-off through a dropping resistor or something.
It would also tend to send additional charging current to the
*other* batteries around it, so eventually the pack would balance
without having to overcharge any given cell group.
.
Now, this is sort of inefficient since that energy has to go
somewhere, but it seems to me that a more intelligent switching
circuit could decide, per cell or per small module, to send
charge current either to its own cell or send it *around* that
module to the next one. And that split could be PWMed based on
cell SOC, not just an either-or situation.
.
There are a lot of clever things that battery builders could be
doing, and I think at this point the technology has gotten cheap
enough and available enough that really fine-grained "smart cell"
control is not only doable but should become routine.
.
_H*

Hi All:

___These Li-Ion 12V’s are/will be expensive (I thought I remember reading close to $1K) but they are doing some internal cell balancing.

Saphion Li-Ion - U-Charge XP (http://www.valence.com/pdffiles/U-Charge%20XP%20DS%20Jan06.pdf)

___Probably pretty easy with just 3 or 4 Spiral wound cel’s to monitor :(

___Mike, the std. Insight - NiMH pack does check the balance across a stack of the D-Cell’s but not an individual cel, and this is where the Recal flag comes from, right? Can a system you are considering start an equalization like charge on that individual stack before the recal threshold is hit? It would not be perfect (not going after the individual cel culprit) but it would use the Insights NiMH OEM monitoring components and all you would need is the ability to over volt/charge that individual string (somehow) before the recal flag was touched?

___Good Luck

___Wayne

I'm not a techy, but why not just drain all the cells to they start out emptied and then just charge the whole pack at once? Won't this get all the charges equal?

Hi Tom:

___Although a Recal doesn’t drain the pack, it acts like you had a drained pack fresh from the factory. The SoC meter drops to 0 and both your acceleration and FE capabilities are severely hampered during the recharge. The BCM knows the individual cel packs are within band but it costs you. I saw one for the first time in Justin’s and it was not pretty.

___Good Luck

___Wayne

WOW lots of feedback since last night.
I will start with Tom's suggestion, why not start at zero, then they will all be equal.
Very good suggestion, except for one thing, we cannot totally discharge a subpack without reverse charging the weakest cell in the string, and since we only have access to the ends of the subpacks, we will not be able to watch for that or do anything about it. Yes we could pull the subpack, and tap in at each cell, but that is dangerous and deep surgery, with the potential for damaging the temperature probes.
Hobbits suggestions are in sync with what I have been thinking about a cell level controller, and as Ian has found, the Tzero guys are actually doing cell level control of the LI-Ions. In their pack.
It is naive to expect a series string of 6 cells of any chemistry to stay balanced over a 10-20 year period if they always get charged and discharged in series.
Microcontrollers that cost 30 cents, can read analog voltages measure temperatrure and output PWM control signals, as well as communicate with other Microcontrollers.
Each cell in our packs cost many times that price.
Each cell with a smart cell controller, that communicates with all the other cells like the Tzero guys did it, is the way to do this. It is expensive, and adds a lot of complexity, but it really is the only way putting loads batteries in series and keeping the pack at peak capacity with all cells getting individual care and feeding.
Joes
I have several lab power supplies with constant current. I will use that for the charger.
I have been playing with the Prius subpacks that I have, to get a feeling for the best way of determining relative SOC. It seems that if all the subpacks are at a stable temperature, with no load having been applied for an hour or more, that voltage differences can show SOC differences.
I will not try to charge the weak subpacks to full, as that would skew the balance the other way. I will charge at say 2A, for 2 minutes. Stop, let things sit for an hour and compare again. The idea being to get the voltage of the suspect subpack over the average subpack by say 10%.
I know that we could still have a cell within the subpack that is weak, but to go any deeper will make this impractical, so lets do the experiment tomorrow, and wait a few days for testing to confirm or deny that this can work.

The test has been completed. We started by pulling the battery relay and current system from the side of the rolled insight pack. We measured the 20 subpack voltages.
7.0, 6.69, 6.44, 6.19, 6.70, 6.84, 7.11, 6.84, 6.82, 6.76, 6.80, 6.93, 7.04, 7.22, 7.13, 6.93 6.71, 6.79, 7.38.
The pack has been sitting idle for an indefinate time, at least 3 months, so the variation subpack to subpack was not suprising.
Next we pulled the pack from our test Insight which had just recaled in my driveway at about 50% SOC, we opened that one and again took the voltage measurements.
7.41, 7.40, 7.39, 7.42, 7.41, 7.43, 7.40, 7.40, 7.41, 7.41, 7.42, 7.41, 7.44, 7.43, 7.42, 7.42, 7.41, 7.41, 7.44, 7.43.
We were really suprised that the pack that had just had a recal with virtually no time to recharge was so closely matched. The total voltage was 148.3V, and the recal had happened at about 50-60% SOC. From my experience the 148V was right in the correct range for the 50% SOC. 100% is about 169V
Next we put the pack from the rolled Insight into the test Insight.
When we started the car, it started with the 12V starter, and the 12V battery light came on.The IMA pack was no bars, so the pack was pretty badly depleted.
A check of the 12V battery showed that the 12V system was down to 11.5V and dropping, and the DC/DC was off due to low HV pack voltage.
The car was charging at about 20-25A, and soon the HV battery pack showed a bar, and began to charge. When we got to 2-3 bars, the dc/dc kicked in, and the 12V system light went out.
While I was disassembling the recalibrated pack to take the above voltage readings, the test pilot went for a test ride to see what shape the battery pack was in.
He saw some strange behavior, first regen became limited before the SOC was over 50%, he tried to fully charge it, but could not get there. When he used assist, the assist would drop out at about 50%, but the battery gauge did not do a recal, just had no assist. After several tries to cycle the pack, he gave up. We pulled the pack, again opened the rolled insight pack to see what the voltages looked like.
7.72, 7.69, 7.69, 7.71, 7.70, 7.71, 7.74, 7.73, 7.73, 7.70, 7.73, 7.71, 7.73, 7.76, 7.71, 7.71, 7.69, 7.70, 7.71, 7.81.
More variation than the one from the car, but not as much as I would have suspected.
We decided that The slight voltage difference between the subpacks on the original pack did not seem large enough to warrant trying to charge the weakest subpack, so we just put it back in.

What did we learn?
We still do not know the true cause of a recal.
The 12V system will not charge if the 144V pack is too low.(already knew that)
Even a pack that has recals can have an open circuit voltage variation subpack to subpack that is only a few millivolts.
I suspect that under load or under charge, the variation subpack to subpack would be greater, so I need to do an under load test. I will just have to do what I have been planing to do for several years, and build an opto isolated real time voltage monitor that can log the voltages while driving.
The rate of voltage drop under load may be part of the recal determining system, so until we can do a real time voltage monitor, we are still in the dark.
Sorry guys I had hoped for more meaningfull and conclusive results.
The recal mystery remains.:(

Great Stuff ... :)

Since as far as I know the internal resistance of the NiMH cells changes as it cycles and ages... I would also agree that the voltage number variation is probably best shown while under load.... also remember that the internal resistances also change slightly with SoC and with cell temperature..... so not only under load .... but make sure to compare this when the battery pack is at various temperatures and SoC.... but this is all probably stuff you were already planning on.