Analyzing, refurbishing/repairing a dead hybrid battery pack (2000 Honda Insight)

[MetroMPG]

Originally Posted by Formula413

I was told that once the voltage stops increasing, that's when you stop charging, usually around 170 volts.

In talking to someone with quite a bit of experience with NiMH chemistry, I'd say you might want to see if you can borrow that charger again.

The cell isn't charged until its voltage peaks, and then declines slightly. So watching for the voltage of the pack to peak may fail to balance all the cells.

[one method of charging is to monitor] the change of voltage with respect to time and stop when this becomes zero, but this runs the risk of premature cutoffs.[15]

Nickel Metal Hydride batteries - Wikipedia

With a relatively low current 350 mA grid charger, we would need to wait a long time for the last cell in the pack to peak & decline in order to bring things into balance if they've drifted out of whack.

[IamIan]

Originally Posted by MetroMPG

I had another specific suggestion that it may be advantageous to remove the sticks for testing specifically to keep tabs on which cell/s is/are getting particularly hot during charging. Might be a useful data point to have.

Kind of ...

It can be useful yes ... but not by itself ... To be useful you would have to combine the temperature per cell data with other data for it to be useful.

Going down to the individual cell level can be beneficial ... but it takes significantly longer ... when 20 sticks can easily take a week or so to test ... going down to the individual cell level could easily end up taking 6x longer ... potentially useful ... but a significant increase in invested time... generally speaking ... my own two bits is that if you are not willing to put in the additional time to quantify the important variables ( Capacity Wh or Ah, Resistance Ohms , Self-Discharge Rate ) for each individual cell ... if you aren't going that far than it isn't worth going past the 6 cell subpack stick as a unit level... but that's just my personal preference.

Individual cells out of a 6 cell subpack stick can get hotter than other cells in the same stick because of more than one reason.... such as:

• Differences in internal resistance ( this would be useful information to have ... if you can quantify it and rule out the influence of other factors. )... otherwise identical cells with more resistance will generate more heat... and thus get hotter over the same period of time.
• Differences in SoC ... even with everything else identical including internal resistance ... if cell A is at 80% SoC and cell C is at 60% SoC ... cell A reaches 100% SoC first ... it will be converting all the charging current into heat while the cell C catches up charging up it's 20% SoC difference.
• Differences in Capacity even with everthing else identical ... first cell to 100% SoC ... converts any additional charging current into heat , while the other cells catch up.
• Differences in heat dissipation ... even if everything else is identical ... the two cells at the ends of the subpack stick have more exposed surface area with which to dissipate their heat to the ambient ... meaning even if everything else were identical , due to the increase exposed surface area of the end two cells , we would expect them to be slightly cooler in temperature.

Unfortunately everything else will not be identical ... you can have any one of or all of those 4 influences happening simultaneously ... SoC difference effects influencing Capacity differences effects influencing resistance differences effects influencing differences in heat loss area effects... all you will see is a temperature ... without additional data you don't have a way of knowing what is the cause... thus without being able to narrow down the cause of the difference in heat it is useless information by itself.

Originally Posted by MetroMPG

Question (pardon my noob-ness):

Nothing to pardon ... It is just a hobby / interest for me ... I see no good reason not to share some of what I've learned with others... So feel free to ask anything you like.

Originally Posted by MetroMPG

Good idea. But since voltage begins falling as soon as the discharge begins, at what point do I record V to get a best estimate? 1 second after starting to discharge? 10?

The easiest is as soon as the rate of current stops rampping up and has stabilized at whatever number of amps you are using... this is usually a very short period of time.

Technically there will be some other influences ... due to a non-ideal battery ... but I wouldn't worry about it.

The purpose of the testing it not to specifically determine the resistance to be accurate within +/- 0.000001 mOhms or whatever ... It is just getting a reasonable estimate ... and as long as your testing methods are consistent than you can compare your results to the results from any other test you have done the same way ... and the errors rates should be nearly identical and for all practical purposes cancel out.

Different test methods will give different results for the same item ... so keep it as consistent as you can , within reason.

- - - - - -

Glad to hear things are going well with the testing.
Looks like more than a 2.8 Ah SoC difference had built up for whatever reason ... from the BCM point of view that is effectively about a ~40% loss of usable battery capacity.

As you get more testing data in , it will be interesting to see how much actual discharge capacity Ah difference there is between the 20 subpack sticks ... from the strongest to the weakest ... I strongly suspect they will be much much small of a difference in actual usable discharge capacity than the 40% caused by the ~2.8Ah SoC difference.

Good stuff

[IamIan]

Originally Posted by MetroMPG

With a relatively low current 350 mA grid charger, we would need to wait a long time for the last cell in the pack to peak & decline in order to bring things into balance if they've drifted out of whack.

FYI
The slower the rate of current the smaller the -dV ... eventually it gets so small that it would be in the range of signal noise sized change in voltage ... also keep in mind the -dV of one cell can be disguised when one is looking at the total sum voltage of all 120 cells in series ... at small enough currents for NiMH ... the alternative often used is to take advantage of NiMH's superior abuse tolerance , and just continue to slowly overcharge everything until the voltage levels off... like you indicated ... that can take a considerable amount of time to 100% finish up ... but some topping charge will result in some better SoC balance ... even is not long enough to be 100% topped off.

[MetroMPG]

Originally Posted by Old Mechanic

Was there a pattern to the state of charge relative to the location in the series of sticks?
As if they were all in a straight line from positive to negative connection.
regards
Mech

I've borrowed an idea from Eli at Insight central, and am posting my pack analysis in a Google docs spreadsheet.

(Eli documented the analysis of a spare pack he got from a wrecked car that had been sitting in a junkyard for 1.5 years. Here's his spreadsheet: https://docs.google.com/spreadsheet/...CIeXn5cD#gid=0)

Old Mechanic, FYI my sticks are numbered 1-20 and make a "straight line" series connection. #1 is the negative end, #20 positive. Here are my results, which I'm updating as I go:

• MetroMPG's battery pack analysis - 2000 Honda Insight (Google Docs spreadsheet)

.

[MetroMPG]

How long will a complete pack analysis take?

About 9 days, going around the clock:

• 7 A charge including 20 minute rest takes ~80 minutes (@ ~6500 mAh capacity)
• 10 A discharge including 20 minute rest takes ~40 min
• So, one complete cycle (rest + discharge + rest + charge) takes ~2 hours (120 minutes). (Of course not all cycles will take 2 hours, particularly the initial ones where capacity is lower.)
• If I auto cycle each stick 5 times, that's ~10 hours per stick. 10 hours x 20 sticks = 200 hours. That's 8.3 days.
• Factor in the final self-discharge top-up and other fiddling around, and let's round that up to 9 full days.

A note on the SuperBrain 989's "auto cycle" feature:

If cycling is stopped prematurely (ie. after 4 cycles of a defined 5 cycle run), you lose the saved charge & discharge mAh data from the previous cycles. (Ask me how I know this. )

[MetroMPG]

What does the "peak detection" setting mean on the Super Brain?

Before starting this project, I had done little to no reading about NiMH batteries, so I was puzzled by the "peak detection" parameter needed for setting up a charge cycle.

I'm posting this for others who may find themselves in the same boat.

It boils down to the way NiMH batteries behave as they reach a full charge: their voltage peaks and then declines slightly. Until that happens the charger says, "nope! Not full yet," and keeps on going. (This behaviour is more obvious at higher charging currents, such as the ones being used to cycle the Insight sticks.)

You can see it in the NiMH charge profile, from Wiki:



source: Nickel

So with the Super Brain setting at "5 mV per cell peak detection" for charging the 6 cell subpack, it waits until it detects a voltage drop of 30 millivolts or more (30 = 6 cells times 5 mV) from the highest peak seen so far before concluding the charge cycle.

You can sometimes see this if you watch the voltage display near the end of the charge cycle. It will rise slowly & steadily up to, say, 9.14 Volts, then flicker back and forth to 9.13 V a little bit before ending the cycle.

I'm learning... I'm learning...

[MetroMPG]

Speeding up the process: watching initial subpack charging for "false peaks"

(The following was explained to me by a remote control airplane flyer who has quite a bit of experience with NiMH batteries.)

So now we know that the charger is watching for voltage to peak and then drop slightly to determine when the sticks are full.

But NiMH cells which have not been regularly used (like a hybrid pack that has been sitting unused for some time following the setting of error codes) may also give "false peaks". That means voltage irregularities trigger the charger to stop when there is actually more capacity available to continue charging the battery.

Packs that sit for extended periods of time may false peak when they are first used again. This is normal and will happen with all NiMH cells. You can re-peak the pack and use it. The pack will be back to normal after 2-3 charge cycles.

Source: Chargers

So we can potentially speed up our pack analysis knowing it's likely that subpacks which have been sitting for weeks/months may trigger the charger to shut off too soon during the initial charge cycles.

It's easy to test for "false peaking": wait 30 seconds after the charger stops. Then re-initiate charging. If the charger stops again in short order, it was a "real" peak. If it keeps charging, then you've just passed a "false peak" and may be able to add a significant amount of charge.

So far, I've detected false peaks in 3 of 5 sticks. In those 3 cases, I was able to add ~28-42% in mAh to the sticks:

Initial charge capacity:

• Stick #2, 4247 mAh: 3001 + 1246 (42% added after "false peak")
• Stick #4, 5159 mAh: 4129 + 1030 (25% added)
• Stick #11, 4505 mAh: 3523 + 982 (28% added)

Watching for false peaks could potentially speed up analysis (that's the theory, anyway - comments??), if it reduces the total number of cycles needed for each stick's capacity to stabilize.

Even reducing the analysis by 10 cycles would save ~20 hours.

Or I could have just bought another charger and run two, then sold one afterward.

[user removed]

Thanks again Metro for the information. Your efforts and the linked information with Eli's efforts have convinced me to seriously consider another Insight, this time a manual transmission.

I'll let you know how it pans out in a little while.

Is anyone aware of a source for sticks in case you find one that won't wake up? It seems from yours and other efforts that they seem to wake up nicely.

regards
Mech

[IamIan]

Originally Posted by Old Mechanic

Is anyone aware of a source for sticks in case you find one that won't wake up? It seems from yours and other efforts that they seem to wake up nicely.

At that point ... you have a few different options / directions ... depending on your specifics and desires.

Option #1 > Junk Yard Gen-1 Civic Hybrids
Gen-1 Honda Civic Hybrids actually use the same subpacks ... and the subpacks themselves are interchangeable with Gen-1 Insights... about ~4 years ago when I went out to test this theory of mine I got two civic IMA battery packs from a junk yard from totaled civics ... $250 each pack x2 = $500 for both ... the junk yard owner told me he knew nothing about the "hybrid stuff" , and was just going to take them to get recycled... after doing a series of tests on them using methods similar to what is being done here , I found out that all 40 subpacks were fine ... nothing was wrong with them at all.

Option #2 > Junk Yard Batteries collected and already tested for you:
Sense that time years ago ... Ron at Hybrid-Battery-Repair.com took this concept and turned it into an entire business ... You can pay him for individual subpacks he has already tested and weeded out of the questionable ones if you like ... How many of what tested quality he has fluctuates regularly.

Option #3 > Replace the Subpack with New Alternative:
Due to the issues of timely acquiring reasonably priced junk yard candidates and costs of testing and sorting the used batteries Ron then sourced his 'BetterBattery" as a new battery alternative ... as a drop in replacement ... It is not the same battery as used in the OEM ... but Ron has tested it , and he at least claims it is just as good or better than the OEM.

Option #4 > Depending on specific issue with the current pack ... it might still be usable if paralleled with a 2nd battery ... thus reducing the strain on each of the two battery packs and simultaneously increasing the total battery capacity.

Option #5> Depending on the specific issue ... regular grid charging can extend the service life of some battery packs.

Option #6 > Replace all the OEM IMA NiMH battery subpack sticks with something like Ron's 'BetterBattery'.

Option #7 > Replace the whole OEM IMA NiMH battery pack with something like the 20Ah A123 pouch cells Peter did ... they actually still fit in a modified OEM battery case... will also require a new BMS system like the one he uses or something similar.

Option #8 > Pay Honda for a Brand new OEM NiMH battery pack.

Option #9 > Use a DC-DC Keep alive mod to be able to run indefinitely without the IMA battery.

[IamIan]

Originally Posted by MetroMPG

So with the Super Brain setting at "5 mV per cell peak detection" for charging the 6 cell subpack, it waits until it detects a voltage drop of 30 millivolts or more (30 = 6 cells times 5 mV) from the highest peak seen so far before concluding the charge cycle.

Shown bellow are some of the common end of charge detection methods different NiMH chargers use.

I suspect the Super Brain also uses some of these other methods as fail safes ... for instance it has an optional temperature sensor could potentially be used for the dT ( change in Temperature ) , and the dT/dt ( Change in Temperature per unit time ) , and the TCO ( Peak Temperature Cut Off ) ... I suspect it also has a maximum peak voltage based on selected chemistry and selected number of cells ... but sense it tracks voltage and current over time ... it could potentially be doing any of those methods ... including the CV@CC ( Constant Voltage @ Constant Current ) , at a CC the voltage will eventually level off ... if +dV or -dV are too small to detect the CV@CC can still be detected.

A side note ... a less common method deployed in the IC4 NiMH batteries ... is the change in internal pressure detection ... as non-charging chemical reactions being to dominate the reactions the internal pressure also increases ... the IC4 NiMH cells used this with a internal resettable pressure switch as a fail safe ... and the Prius NiMH cells don't detect the pressure change ... but the battery pack is designed to reenforce the plastic NiMH cells cases ... which if not thus reenforced would swell outward with the increasing internal pressures ... so they reenforce the cells from the outside ... while Panasonic in the Honda design instead uses a stronger metal cell casing to contain the changes in internal pressure.

- - - - - -

Unfortunately ... all of these methods have a good chance to do some amount of overcharging ... the up curve in the dV bump going up and then back down ... is a symptom of the dominate chemical reaction in the cell changing from a charging one to a over charging one ... the up swing in the temperature slope is also a symptom of the dominate chemical reaction in the cell changing from a charging one to a over charging one ... so unfortunately all the commonly used end of charge detection methods for NiMH look for things that only show up when the dominate chemical reaction is no longer a charging one... but NiMH is a abuse tolerant battery chemistry ... so speed of charging is usually a greater concern than battery cell service life , which is why these common charging methods are common.