Yes, I know, yet another ICCU/12V thread. But I'm an electrical engineer who does failure analysis and troubleshooting for a living, so I can't stop wondering about this.
Edit: All of this is murky enough that I put together a graphic to explain it.
Most people know an EV still needs a low-voltage battery to run vehicle systems and survive long periods asleep. On E-GMP cars, the ICCU is what charges that 12V battery. It connects straight to the HV battery, the 12V battery, the rear-seat AC outlet, and the AC pins of the charge port. The DC pins skip the ICCU entirely; they go through the rear inverter first and then reach the battery through relays.
Inside the ICCU are two main circuits. One is a bidirectional AC-DC converter, used for AC charging the HV battery and for V2L. The other is a DC-DC converter (the LDC) that steps the HV battery voltage down to 12-14.5V for the auxiliary battery. That converter is indifferent to where its input comes from, since it just hangs off the DC side of the AC-DC converter, so it can be fed by the HV battery or by the AC-DC converter depending on the situation.
Both circuits tie to the HV battery through a fuse, and that's the one that blows in an ICCU failure. That explains why AC charging and 12V charging both stop when it fails, although in theory DC charging should still be electrically possible.
The NHTSA recall campaign says warranty parts that were inspected had failed FETs in the LDC circuit. If that's accurate, then how we AC charge or use V2L shouldn't matter, because that circuit isn't what's dying.
What, then, is the failure mode? HMG looked into over-current conditions as a possible cause of LDC failure, but the result seems to have been inconclusive. My own guess is a built-in weakness in the buck converter design that produces transients, over-voltage or over-current, which destroy the FET. Once the FET fails it becomes a short to a lower voltage or to ground, and that blows the fuse between the battery and the ICCU. A 170A fuse sits between the LDC output and most circuits; exceptions are the interior 12V sockets (each with its own 20A fuse), the front and rear electric oil pumps, and the tailgate motor. I've not heard of any fuse other than the main ICCU input fuse needing replacement, so the short to ground is probably inside the LDC circuit itself and not out on the 12V rail, given that the main fuse for this circuit has to be larger than the rail fuses.
What could trigger a failure? Battery rail voltage fluctuating, the LDC's output current or load changing, shifts in the LDC's target output voltage, and the temperature of its components. That's four constantly varying inputs affecting the buck converter, and it only takes one edge case pushing voltage, temperature, or current too high to blow a FET. I'm assuming a short to ground, because a short to the 12V rail would likely do catastrophic damage everywhere, though strong over-voltage protection on the output might also draw enough current to blow the fuse. Either way, it's hard to pin down exact triggering conditions, but there are a few things owners can actually do or keep in mind.
First, LDC temperature is directly linked to the AC-DC/DC-AC converter's temperature. If there's a lot of heat there and the LDC suddenly has to work hard, trouble could result. It's a paranoid step, but cutting 12V loads right after long V2L sessions or AC charging may reduce that risk.
Second, the condition of your 12V battery. My 12V battery died fairly quickly not long ago, so my ICCU was working hard to keep it topped up. After I switched to an ohmmu LFP battery, which sits at a notably higher voltage, the 12V charging pattern changed from "slow but frequent" to "infrequent but fast." I'm not sure why, but however the car calibrated, it now alternates between pushing about 1kW through the LDC and putting out exactly zero, leaving the battery to carry all 12V loads for stretches. That's strange, and I never saw it with two vehicles that had their original lead-acid batteries. My gut says this on/off behavior isn't good, so if someone must replace their 12V, I'd lean toward recommending AGM (assuming the behavior comes from ohmmu/li-ion batteries with their built-in BMS).
Third, sport mode plays into this somehow. The 12V charging logic in sport mode differs from eco or normal. For some reason it pins the LDC output at maximum and charges the 12V as aggressively as it can. Nothing I know of links this to any ICCU failures, but it's worth knowing.
Fourth, charging on a 400V architecture heats the OBC/LDC. I'm not totally sure why. Even in winter temperatures, the LDC quickly climbed above 50°C during a 30 minute Supercharger session. The circuitry doesn't explain it, because the boost conversion occurs in the rear motor inverter. Best I can tell, the ICCU shares a coolant loop with the rear motor inverter, so their temperatures track each other. Long 400V charging sessions or doing it in hot weather could overheat components.
My main conclusion: there's very little reason for anyone to alter how they drive or charge. The one real recommendation is to keep the 12V battery healthy so the LDC doesn't need to work as hard, as often, or under odd, unpredictable conditions. People should just use their cars as they need to. And if anyone has OBD sensor logs from a car that suffered an ICCU failure, I'd really like to see them.
On the subject of a healthy 12V, I suspect those of us (me included) who have had to jump our cars are victims of the car's algorithm. One user said the algorithm ignores 12V drain while the vehicle sleeps and only wakes the car to charge when a timer, set by when it expects the 12V to be low, runs out. Apply that to a failing lead-acid battery and the car assumes the battery will hold charge for a set time, but it drains far faster than expected and is too low by the time the car next wakes. That value comes from some calibration, so disconnecting and reconnecting the 12V leads might recalibrate it for your weak battery and buy a bit more time. More generally, I think a good fix is a high-quality battery plus forcing a calibration routine now and then as it ages, perhaps once or twice a year.
TL;DR
If the ICCU failure reports are right, the faulty circuit is the LDC DC-DC converter, which has a single job: when the car is on, convert high voltage to 12V for the aux battery. The best ways to reduce risk are probably keeping a healthy 12V (AGM looks like the best choice), limiting 400V charging, and keeping 12V loads low right afterward (within about 10 minutes) after 400V charging or long AC charging or discharging.
For repeat offenders that have killed several ICCUs, it's difficult to say. Loose wiring in the power relay assembly, at the LDC output or ground, or some other fault on the vehicle's 12V system could be responsible.
