The battery fires at ECU sense connector thread

When I see four, now six, of these posted in just the last couple years on our aging Gen 1s, I wonder if it merits our attention.

1. Russell Walker, Brisbane, April 2013 with pics. (For a while I had forgotten about this one.)
2. lovemy02prius, near Detroit, September 2014, with pic.
3. strawbrad, customer vehicle, Minnesota, October 2014, post #3 below.
   
4. andrewedwards89, February 2015, no pic.
5. greasemonkey007, Arkansas, March 2015, with pics.
6. Herb Gayheart, Kentucky, January 2016, with connector and ECU pics. Story suggests warning signs treated rather cavalierly for some time before incident.

I wonder if this is starting to be a thing (if four PriusChat threads from a population of ~ 52,000 Gen 1s could be enough to constitute a thing).

A couple possibly relevant NHTSA complaints could be 10141958 and 10111792. (I'm not sure if that link stays valid indefinitely; if it stops working, just plug those numbers into the search - choose "ID Number" at the bottom.) I was happy to see there aren't more ... they are vastly outnumbered by complaints about the steering rack.

PC members who are in the repair biz: how often, if ever, are you seeing batteries come in with fire damage around the sense harness/ECU connector?

Battery rebuilders (@hybriddriveguy ?) ... are you seeing cores come back this way?

-Chap

Edit: include strawbrad's in list

 

On the specimen table today is one new, never-installed sense wire/bus bar assembly.

The orange connector where the excitement happens is over at the right. Between the first and last skinny sense wire entering it can be a voltage difference near 318 V at full charge, but owing to the way Toyota has chosen the order of sense wires to pin positions and the geometry of the connector, the voltage gradient between any two pins (disregarding all others*) is never more than 10 volts per mm by my calculations. Details on the connector and pin order were posted earlier here. The fields at the connector are not as strong as they'd be just before the connector where all the wires are mashed together in a skinny bundle.

In new condition, between all possible pairs of wires (190 of 'em), the insulation resistance tests offscale >999 MΩ measured at 500 V.

There's surely nothing useful to learn by measuring one that's already burned up (it's already dead, Jim), but it might be interesting to get megger readings on some that have been removed from long service without burning up. Anybody?

-Chap

* yes, my calculation of gradients between pairs of pins "disregarding all others" is pretty much total handwaving, and it would be better to model the field of all 20 pins together and find the steepest gradient, and no, I haven't done that.

 

Brad,

Thanks for the report. I've edited it into the list at the top of this thread, but I'm unsure on the details. Are you a shop and this was a car a customer brought in? Or was it something you had picked up at an auction, etc., and had no history on?

-Chap

 

Just added Herb Gayheart's incident, Kentucky, January 2016. Story is interesting because of apparent problem signs treated a bit cavalierly by Herb and also by the person he had just bought the car from, going back evidently some time, maybe years....

-Chap

 

What I would love to see on this thread would be observations (either instrumental or physical/visual) of older busbar assemblies or ECUs that people have taken out when rebuilding older batteries, to try to determine their condition when they are aged and well-used but before any incidence of fire.

Ones that have already gone boom it's kind of too late to learn much from, and my instrumental check in this thread above only proves that a brand new busbar out of the box is really good, which we probably all could have guessed. It's the condition of old-but-unburned ones that could start to tell us something.

-Chap

 

I've got a spare right front fender I keep around for just that purpose, myself.

My dream for this thread is to try to have mostly technical info on it, though.

-Chap

 

The particular conflagrations this thread is about always seem to be starting near here or here, which I don't think would be where (Rex? Dave?) Taylor was doing the nickel.

These last pics kind of make me wonder whether the origin was in the connector or on the PCB...

-Chap

 

All those questions are exactly what I started this thread to try to find answers to. The best way anybody could contribute toward finding the answers would be, if you are doing any work on an old Gen 1 battery or battery ECU, please report in detail on the condition of the old bus assembly, ECU connector and circuit board, etc. With instruments if possible.

Upthread I contributed megger readings on a new-in-box bus assembly (all very good, as you would expect). We don't have any yet from old, removed-from-service normally (i.e. not after a fire) assemblies. Those that have already been through a fire of course are too crispy to tell us much.

Personally, I'd lean toward being pretty conservative about applying any kind of improvised treatment here that Toyota didn't. Certainly I have a new bus assembly on hand and my intention is that whenever I have reason to be in the battery, that new one is going on ... the old one's 15 years old now.

If you had reason to be in your battery and wanted to be cautious, but you don't have a megger, you might just preventively replace the bus assembly and send the old one to someone who does have a megger, so it can be studied and add to the results here. It would help to post details of the car, age, where driven, etc.

That doesn't address the condition of the ECU circuit board or connector, which I'd also be likely to examine with a highly critical eye.

Here's an interesting passage I stumbled on in a page from Maxim Integrated (happens to be about their voltage regulator products, but anyway):

It might be worth looking for a way to check that; comparative megger readings at the sense connector of a new ECU and a sampling of old (but unburned) ones could help. Perhaps there's a wash chemical that would work to gently remove such a gunk layer, but you'd want it to be one that would not damage or change the properties of the factory-applied conformal coating on the board. There's a nonzero potential for well-intended 'improvements' to make matters worse.

So in general there's a necessary balance between the value of going in and doing something and the value of letting well enough alone if the battery is not exhibiting any problems. Iatrogenic problems from working on the battery get reported here sometimes too.

One absolute recommendation I would make is never ignore a P3009 if your car reports it. Usually the fire reports don't mention whether the car was reporting current leakage in advance or for how long, but clearly current leakage does not always stay harmless and small.

-Chap

In the déjà vu department, the PriusChat record reveals that way back in 2008, within two months of buying my car and before any chance to look at how the bus assembly was put together, or hearing about any fires, I was wondering whether the bus assembly included limiting resistors. It seems like the simplest thing in the world, since they should only be feeding high-impedance analog-to-digital converters in the ECU anyway, and if each wire were built with, say, a 1 MΩ 1/8W resistor out at the module terminal end, pretty much nothing bad could happen.

I still don't get why they didn't....

I'm not sure whether it would work well, or not, to design a retrofit bus assembly incorporating that feature. The sticky bit would be that the ECU seems to have a smaller number of analog-to-digital converters, taking the voltage readings from sample-and-hold circuits that sort of make the rounds of the modules. Depending on how much time the sample/hold spends sampling each module, adding a large resistance in the sense wire might cause the sample to read low.

-Chap

 

If I understand correctly, Dorman's (formerly ReInvolt's) rebuilds come with an ECU, as a matter of course. At least, that was the case when it was ReInvolt.

I was sure I remembered someone (it was pEEf) doing an ECU teardown. It was for a Gen 2; shows the photoMOS switches used to select sample voltages from the sense connector. Later in that same thread, vertex offered a partially worked-out schematic of the sampling arrangement. Later, pEEf mentions getting a look at a Gen 1, finding it to be a similar but higher-component-count design.

I wish vertex had had more time to work out that schematic. It might indicate the effective input impedance, therefore how significantly readings would be in error if resistors were added to the bus assembly. That would also depend on how long each photoMOS switch closes to take its voltage sample. I still think somebody scoped that out once, but I haven't been able to find it again yet.

-Chap