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

[IamIan]

Originally Posted by eq1

any reason to discharge/charge manually versus auto?

Not speaking for him ... I'll just say.

Nothing is 100% ... The control logic for the Super Brain 989 is not 100% reliable either.

I have occasionally had it stop charging early long before a battery is actually charged ... although that only happens rarely , it does exist ... this can be easily noticed and corrected if working manually for each step... from my experience I'd say less than 10% of the charge cycles will have this issue ... and it does no harm to the battery, it is just not a full charge cycle.

Another rare issue I've had with the less than 100% reliability of the Super Brain 989 is on discharge ... on very rare occasions when discharging multiple cells in series it can fail to quickly detect when a single cell reaches 0V and goes into voltage reversal ... This is even more rare, I'd say less than 1% chance ... but it can potentially do damage to the cell it puts into voltage reversal... attached at the bottom is a graph of one such event I caught while logging the voltage with a different device ... the one cell reversal is clearly apparent... and the graph ends where the Super Brain 989 ended the discharge ... it was a NiMH battery pack of 10 series D cells that had been sitting for a while and one of them was out of SoC Balance with the others.

The last issue I have with the less than 100% reliability of the Super Brain 989 is on power supply mode ... don't ever depend on a Super Brain 989 in power supply mode to sustain regulated voltage or current ... although this does not effect battery cycling.

But manual supervision of each step adds a lot of time to a project that already will eat many many hours as is without doing so... and no device is 100% reliable ... that's just how it is ... of course there are other competing devices ... but there is no published 3rd party reliability tests of them that I know of in order to get a good properly quantified evaluation to know if item B is any better or worse than item A.

[MetroMPG]

Originally Posted by eq1

Any updates on how well your rejuvenated pack has been performing?

I'm not currently driving the car, and won't be until later this spring. By then the pack will be a mess again from uneven self-discharge! (Actually, I will have the grid charger hooked up and used periodically by the beginning of March).

Looks like you switched to autocycling at some point in your process: any reason to discharge/charge manually versus auto?

IamIan's response sums it up well.

I was thinking about adding my own experiences with the process to this thread, once I have them: any objections to that?

Not at all - the more the merrier!

[user removed]

Combine the availability of used sticks with the ability to get the good ones back to close to OE specs and you have the opportunity to purchase a used hybrid with a bad battery pack, for a really cheap price and fix it yourself as well as maintain the pack at a decent level of performance.

One of the most relevant threads in this forum and a real opportunity to extend the life of many hybrids.

In areas where there are a lot of military deployments and the cars sit for months without being used their will be a lot of pack imbalances that are diagnosed as failures, when they can be rejuvenated for a relatively low cost.

regards
Mech

[eq1]

Thanks all, very helpful. Currently I'm debating whether to dig in to the pack or drive some more to get a better sense of how the pack is performing as-is. I haven't driven it much overall, and just last night and today I've had some unusual, but perhaps positive IMA/battery behavior.

It quickly discharged with little assist yesterday, but then, it took quite a long time to charge it up - about 35 miles yesterday to bring it back up to about 15 bars, and then another 10 or so today to bring it up to 20 bars. Yesterday I was only getting engine-charging and coasting regen; today I was able to get both coasting and braking regen, with high charge rates (according to the gauge cluster charge gauge). Hit 19 bars in maybe 10 miles, 20 in another 5-10, and then by the end of my 60 mile trip I was able to top it off so that it would take no more charge. That's where I want it to be when I take the pack apart for cycling...

But, I'm thinking that now it actually might have some juice in it and assist will work better, longer. I'd never seen that charging behavior before. I'll rejuvenate the pack regardless, but maybe it's worth waiting and driving more so I know the true performance of the pack before I do the rejuvenating... I'm getting impatient though and want to get going on it... plus, I don't have any external, more accurate gauge to quantify the behavior, anyway...

[MetroMPG]

Ha - nice avatar.

Aggressively cycling an old pack -- in the car, as you are -- can help bring back "lost" capacity. You can see this in the results of cycling my sticks - discharge capacity increased over ~4-6 cyles.

But!

It won't help much if the pack is also out of balance. A grid charger would, though.

[IamIan]

Originally Posted by eq1

I don't have any external, more accurate gauge to quantify the behavior, anyway...

You have a couple options for that if you want.

Peter makes a OBDII device that among other more conventional OBDII Stuff ... will read and display the Gen-1 Insight's IMA battery pack voltage, amps , Watts, and counts Ah and Wh in both Assist and Regen... you'd have to touch base with him about the schedule for his next production run... easy to use just plug into OBDII port.

Or you could tap off the IMA pack for voltage and use the OEM current sensor for current readings... if you wanted to put something more basic together yourself.

Or wait to test each of the 20 subpacks individually like was just done here.

[eq1]

I'm planning on getting one of the OBDIIC&Cs, but they won't be going out till March or so. I'm probably just gonna drive around a bit tomorrow, experiment with varying levels of assist, leave the battery topped-up when I'm done driving, and then take the pack apart and cycle the sticks... It would've been nice to have some data to better characterize my battery as-is. But I'm too impatient, I don't want to wait...

[eq1]

I'm just about finished with my stick cycling regime... I've mostly been posting this and that question and info over at insight central, in this or that thread. But I still plan to post some 'results' here...

I've got 1 1/2 sticks to go and I'm trying to resolve a couple things before I put everything back together. One question involves measuring/calculating internal resistance. Basically, after 4 cycles on each stick I went back to the lowest capacity sticks, 5 of them, and did another cycle, with slightly different settings. And then I decided to do another cycle for all of them at these different settings.

So, when I discharged this 5th time I was more careful about documenting initial voltage and voltage under a 3.5 amp load - at 1 minute for each stick. From what I've read I can calculate the sticks' internal resistances by subtracting the loaded voltage from the initial voltage and dividing the result by the current, 3.5A. For example: 8 volts initial, 7.5 volts under 3.5A load at 1 minute, divided by 3.5, which produces 0.143...

My question is, what are the units for that 0.143? 0.143 ohms? I've looked all over and tried to figure it out, but I haven't found anything definitive. I have a Panasonic technical sheet that lists 2 milliohms as the 'internal impedance' of a single cell (12 milliohms for 6 cells I presume), but the conditions are a little different, such as a 'current' of 1000hz. I don't understand that... Overall, my sticks are ranging from 0.13 to 0.17... 12 milliohms would be 0.012 I believe, so something seems amiss in my calculations...

Thoughts?

[IamIan]

Originally Posted by eq1

So, when I discharged this 5th time I was more careful about documenting initial voltage and voltage under a 3.5 amp load - at 1 minute for each stick. From what I've read I can calculate the sticks' internal resistances by subtracting the loaded voltage from the initial voltage and dividing the result by the current, 3.5A. For example: 8 volts initial, 7.5 volts under 3.5A load at 1 minute, divided by 3.5, which produces 0.143...

Yes.
Just keep in mind that 1 minute delay is not needed... just the time it takes the discharge current to stabilize to a given value... usually a second or so... just be as consistent in you testing as you reasonably can.

Originally Posted by eq1

My question is, what are the units for that 0.143? 0.143 ohms?

Ohms law V=IR is the ratio
where V is in Volts, I is Amps, R is Ohms.

Ohms law is just a ratio relationship ... meaning it obeys rules of algebra ... so I can divide both sides by 1000 to get milliVolts, milliAmps, milliOhms.... etc... and the ratio holds true... you just have to have all terms in the same scale for the ratio to work... meaning if you use milliVolts the ratio only holds true if you also use milliAmps and milliOhms... even if you convert to a different scale after using the ratio.

Originally Posted by eq1

I've looked all over and tried to figure it out, but I haven't found anything definitive. I have a Panasonic technical sheet that lists 2 milliohms as the 'internal impedance' of a single cell (12 milliohms for 6 cells I presume),

Plus any influence of the connections between the cells.

Originally Posted by eq1

but the conditions are a little different, such as a 'current' of 1000hz. I don't understand that.

That is referring to an AC test of resistance not the DC test.

The DC test makes a few assumptions:

• Insignificant capacitance.
• Insignificant change in SoC during the time taken to perform the test.
• Significant Electro-Chemical capacity to power the test.

When these assumptions are reasonably in question the more accurate AC test is used... The AC test does not require these assumptions... which is why battery manufactures often use the AC test method instead of the DC test method.

However ... sense the DC test method is significantly cheaper and easier ... if you can establish those 3 assumptions... and you can be reasonably consistent in your testing methods ... then the DC method can give reasonably accurate results for less effort and money.

[eq1]

Excellent, thank you...

So, using my example, if it were 8000 mV, 7500 mV, and 3500 mA, the result would be N mOhms? The result is 0.143, the same as before...

Also, I asked this question over at Insight Central, but haven't gotten an answer: Do you know the pair-ordering of the subpack pairs as monitored by the 10 voltage taps? The OEM numbering of the sticks, as seen from the circuit breaker side, is:

-bottom row, left to right, 1-7
-middle row, right to left, 8-14
-top row, left to right, 15-20.

I want to make sure my relatively weaker sticks aren't paired up...