A message arrived to the inbox one day, from a fellow I'd seen on
the Prius forums a few years back.
Over the years I had sort of lost touch with active participation
there, but he and some others had been tracking a strange issue that
seemed worth specifically reaching out to me on.
Evidently some first-generation battery ECUs were having nasty corrosion
problems with the block-voltage monitor wiring, sometimes leading to small
fires and cooked computers, and a similar issue had apparently also begun
showing up in a few early second-generation cars as well.
He pointed me to the
Priuschat thread
on the subject, which was getting actively updated with incident reports
as they came in.
The posts there from "Texas hybrid batteries" and "strawbrad" are of
particular interest, showing how mysterious green crystalline corrosion
appears in otherwise seemingly fine monitoring connectors and eventually
leads to shorts, diagnostic codes and even a few meltdowns.
And always seeming to begin in a particular place in the connector.
This was obviously worthy of concern, as my '04 was right in the age group that could have the issues, so it was time to revisit the innards of my battery pack to check up on things. While in there I took the opportunity to make some other inspections and improvements as well. |
With the ECU removed and its cover popped off, the pins into the monitor board can be inspected. No hint of anything going on here either, around pin 22 or any of the others. |
A couple of the wire lugs are backed out for inspection too, and even under decent magnification there's still nothing notable or vaguely suspicious. |
So the plot thickens. This car has been on the road longer than a lot of second-gens under a reasonable variety of ambient conditions, including various roadtrips around the country to climates other than New England, although I can't say it sat soaking in high humidity or any other obvious persistent condition all the time. Electrolytic effects in different environments? You wouldn't think that appreciable current passes through any of these pins, as they're *voltage* monitors. And there shouldn't be any corrosive vapors drifting over from the battery itself, as the modules are sealed and the chemistry is fairly benign. For the moment, I'll just accept having dodged this particular bullet. |
A practical Prius history lesson
Back in 2010 I was hangin' with my fellow hybrid geeks at Hybridfest in
Wisconsin, where Dave Taylor had a booth for his startup battery
rebuild shop called ReInvolt.
He began by rebuilding failed first-generation Prius packs from the more
robust and newer second-generation modules, giving new life to the beloved
rides of some of the early adopters.
The business flourished and grew to cover many other types of hybrid
vehicles, and eventually became an arm of
Dorman Products,
and Dave's still involved in the field.
Anyway, we were chatting at the Fest and I mentioned the deposits I'd
found on my battery terminals, so he handed me a little bag of his
latest innovation -- new tin-plated copper straps, designed to reduce
dissimilar-metal issues at the interconnect points.
Given that no immediate action was needed on my pack, I squirreled them
away in some box back home and almost forgot about them, until now.
This is a chart mostly aimed at the construction industry, but is useful here too. Slap two different metals together in the presence of an uncontrolled atmosphere and moisture, and the degree of interaction can be roughly predicted by how far apart they are on the scale. Add electricity, and things can get a lot worse! What's most interesting about the white fuzz is that it's clearly heavier on the *negative* battery terminals, even though both studs and the copper are uniform. Given that both charging and discharging always involve a little loss as heat, one can assume that more energy flows electrically *into* the pack than comes back out, and thus electron current spends more time moving from the copper to the steel of the negative module studs than the positive ones. I have no idea what the reaction might be or what the white stuff is composed of -- it's sort of flakey and amorphous, not really sharply crystalline. The safest bet is that if metals are going to touch each other long-term, they should be as close as practically possible on the galvanic scale which the tin plating accomplishes. I don't know what process is used to plate the straps, but there are any number of simple-to-use plating solutions for the electronics industry. |
The safety plug has no signs of any issues, even though it undergoes the same heavy flows of current. They must have picked just the right plating for those bolts. |
Once things are together and functional again, it's time to go out for a little charge/discharge abuse-testing and then look for hot spots. If anything, the interconnect straps are the coldest part of everything. More warmth is seen down in the bottom of the cells, which may be slightly influenced by heat coming up through the floor of the car -- note the hint of a warm area toward the right where the exhaust pipe passes underneath, but if anything that's a minor factor. |
Finished but not forgotten
It was nice to have closure between those early discussions with Dave and some other battery-rebuild enthusiasts and finally getting this refurb done. The whole job was a little tedious but quite simple, and I was done with it well before I thought I would be and thus the car was *not* going to be down the entire weekend after all. Meanwhile, an acquaintance had been poking me to join him and some other hybrid folks at the big Nantasket car show, which was happening all of a day after I wrapped up the strap swap. So what the heck, I thought, why bother putting the car back together when this could be a great geeky display of things most people never see? After printing up up a couple of explanatory signs, I was good to go. |
That car show was a much bigger deal than I thought it would be -- must have been a half-mile of muscle cars all lined up along the beachfront road, and here I was right in the midst of it with my 12-year-old original battery exposed to the masses.  And people were interested. Not since the fuel spike of '08 dropped off did I have the opportunity to stand out in the hot sun all day handing out flyers and knowledge -- it was fun to be back in that event groove for a while, even if most folks go to shows like that to ogle all the "old iron" and street stompers. Like I reflected on in another article, I totally understand the draw of the classics and it's still fun to contrast all the different technologies and their evolution. |
What's especially interesting is that the lead to pin 22, from the negative
end of the pack, carries a little more current than the other taps,
and as I thought about this more one thing suddenly became obvious.
The ECU is a fancy voltmeter, and as such needs a ground reference.
This would logically come from the negative end of the pack, even though
the entire battery system is isolated from the *car* -- the point is that
as each tap is measured all the way up the pack, a small current must get
drawn between it and pin 22.
This would make more far more sense from a design and circuitry standpoint
than trying to only measure deltas between adjacent taps.
The slow pulsing nature of the current is also interesting, as it implies a periodic cycle of measurement with a set of samples taken every so often. With the car still powered up I waited until the current peaked and went back to 0 and then disconnected my meter, leaving the negative-point tap floating free. Right about the time I expected the next pulse, the car went "bip!" and lit the Red Triangle of Doom, and the battery *fan* came on at some intermediate level -- an informal confirmation that a break in a sense tap isn't detected instantly, but only during a measurement cycle. The error code on the scantool was P3030, "disconnection between battery and ECU", and a quick wander through Toyota's description made the likely function of pin 22 carrying "GBB0" even more obvious. I reflected the details on these findings back to the fellow who had originally given me the heads-up on the problem: ## So I've got my pack all strapped back up with happy tin-plated bars, and have been running around with the pack naked to the world in the back because it made a great demo at a car show last weekend. "Naked prius" revisited, in a way, which nicely bookends the whole decade... anyway, in the process of pulling together stuff for the inevitable webpage, I thought hey, why don't I go out and *measure* some of those sense currents and see if there's anything funny about pin 22? ... well, rather than messing with the orange connector it's easy enough to pull the tap lugs off the pack and just run 'em through a milliammeter on their way into the bundle. And I found something interesting. Sampling currents are on the order of microamps, but what happens at pin 22 is that it sees pulses of up to 25 uA about every 4 seconds. Inserting into the path of another tap farther "up" the pack [in terms of voltage] shows similar slow pulse behavior, but not as high a current -- maybe 4 or 5 uA, and bounces around a little more. So here's what I think might be happening. Pin 22 connects to GBB0, the negative end of the pack, and as the block-scanning happens, all the other tap point voltages are *referenced off it*. I.e. the tests are likely from a given tap to its effective ground, rather than taking each individual delta between adjacent ones. So pin 22 *does* get a tiny bit more abuse over the long haul, thus it's "special" in that way. It may be that pulling and reseating that connector a couple of times earlier in a given car's life might head off the whole thing, who knows. *Something* must have put stray resistance across that joint in some of the ECUs as the cars were built, or the choice of metals right there wasn't right, but that wouldn't explain how only some grow bad stuff. On the other hand, the "white fuzz" on the bus-bar nuts was quite a bit more than six years ago, so even though that's a much higher-current scenario maybe *any* potential across galvanically disparate enough metals is eventually going to have its effect. "Texashybrid" probably sees a lot more ambient humidity than I do in general. ## We know that to measure a system you must disturb it, however slightly, and a voltmeter has to draw some tiny amount of current through a completed circuit to move a meter or charge a sample point. Others observed that the series resistors going into the ECU's measurement chips are only 1K, so what if there's actually a very brief inrush of higher current as a given tap is switched in for sampling? That could lead to spikes of a hundred or more milliamps for a very short time, but it's still unclear if even that would be relevant. Both the pins and lugs appear to be gold-flashed at the contact surfaces, in theory lessening the dissimilar-metal issue. So while I think I'm on to the reason pin 22 specifically has been the "first to go" in most instances, the question remains, what's growing those crystals in the first place?? Maybe somebody should get in touch with those guys on the planet Krypton, since they're the recognized authorities on that sort of thing. |