2nd PCS2 failure

[mongo]

Also the PCS is apparently design to handle both DC and AC high power - which you would normally avoid as they complexity of AC or DC presents their own unique design challenges at PCS power levels.

It's a bidirectional DC-AC converter, no way to avoid it.

AC has an additional challenge in that the current flows in both directions. This can cause tremendous stress on components unless the circuit is design to protect against that.

The reported failures seem to be on the DC side MOSFETs, not the AC back to back pairs. The converter switching frequency is way higher than 60Hz, but likely uses ZVS/ ZCS internally.

Sponsored

 

[HaulingAss]

The widespread PCS failures in 2024 and 2025 Cybertrucks are attributed to thermal fatigue in the MOSFETs — and this was not a PCS-isolated problem. The drive inverter recall (NHTSA 24V-832) used similar faulty MOSFET components from the same production era.  So the root issue is that the MOSFET spec chosen for the original PCS wasn’t robust enough for the sustained thermal cycling demands of the CT’s 48V architecture under real-world charging loads.

AI gets a lot wrong. We don't know whether the spec chosen for the MOSFETs was wrong, or if the MOSFETs don't fully meet the spec. It's also possible there was a software issue, or other questionable components, that worked the MOSFETs harder than intended.

MOSFET manufacturing is interesting in that the manufacturer tries to make them all the same, but due to various factors, including impurity distributions in the silicon, no two are alike. Each one will have unique efficiency and the more efficient ones will run cooler. The manufacturer tests them and sorts them into batches before selling them. The ones that generate less heat are priced higher because they are more efficient. Because each batch will contain a range of specs, even those in the same batch will have different failure rates depending upon where they fall in the range of that batch.

It could be that the spec chosen for the MOSFETs was borderline, or that the MOSFET manufacturer was pushing the limit of how they graded them. In any case, due to these natural variations, some will fail and some will last the life of the vehicle.

 

[Flynn]

I don’t disagree, but it often seems like people think others don’t already know this. (To be fair many don’t). I should have qualified my post that my ai had been drinking all day

AI gets a lot right too, and is at least a place to start. It sounds like mosfets manufacturing/grading are a lot like cpu/gpus.

I wonder if munroe or anyone else has or will get their hands on some of these for teardown/analysis. Does it really come down to component temp control? Are failures less common in cold climates?

 

[pae1andonly]

Should we do a group order?

Sponsored