22KW OBC FAILURE INSIGHT & POSSIBLE SOLUTION

[ct14garage]

Here’s a full breakdown of what’s going wrong and a possible fix and the DESIGN FLAW EXPOSED. Weak MOSFET chips. The Onboard Charger (OBC) has some complex circuits, including the Power Factor Correction (PFC) Circuit. This circuit relies on MOSFET transistors, and for the Taycan, it uses the STMicroelectronics 40N120AG – a 1200V, 32A rated MOSFET. Unfortunately, this component is the one that pops everytime (photos attached).

Blown MOSFETS example below:

To solve this, I’m suggesting using a different part when repairing it- the Nexperia NSF040120D7A0, a 1200V, 65A MOSFET that’s rated for 33A more than the original. It's a direct replacement.

We've already replaced the faulty MOSFETs in two cars with the upgraded NSF040120D7A0, and so far, the charging works as expected. Fingers crossed, this fix holds up over the coming months, unlike the original part that would have fried by then.

PS: Before anyone asks, the OBC is not meant to be disassembled, there are no screws to open it. You've gotta cut the back lid open with an angle grinder carefully to not damage anything inside and then once repaired you gotta weld the rear lid!

Sponsored

 

[daveo4EV]

I both love and hate the fact that all these failures in reliability inevitably come down to $0.05 parts…

 

[chun]

I both love and hate the fact that all these failures in reliability inevitably come down to $0.05 parts…

That's how porsche has the highest profits out of all car manufacturer's )

It's a pity they designed it to be "unrepairable" by having it in a complete sealed enclosure.

My dealer in switzerland confirmed that the part is discontinued for new cars installs, and they only keep stock for warranty and repair as mandated by law. He also said all broken ones aren't even shipped back to porsche (as they do with most other replaced parts), they are just scrapping them.

 

[ct14garage]

PS: The OBC is neither designed nor manufactured by Porsche.

It is designed and manufacture in Italy ?? of all places. By a company named MetaSystems

https://www.metasystem.it/en/

 

[ze_shark]

I both love and hate the fact that all these failures in reliability inevitably come down to $0.05 parts…

Not to nitpick, but a silicon carbide MOSFET would easily cost Metasystem 2-3$ a piece in high volume, and this is not a high volume application. It does not invalidate your point about component cost vs failure cost, but this is not a 3904 either ...

Another question is whether they fail because of an intrinsic defect, or a PCB layout reason (heatsink ground plane for instance).

 

[daveo4EV]

Not to nitpick, but a silicon carbide MOSFET would easily cost Metasystem 2-3$ a piece in high volume, and this is not a high volume application. It does not invalidate your point about component cost vs failure cost, but this is not a 3904 either ...

Another question is whether they fail because of an intrinsic defect, or a PCB layout reason (heatsink ground plane for instance).

your experience exceeds my computer science experience - so I sincerely bow to you - but yeah a $5 part vs. a $1680 "option price" part - which is $2400 as a service part - $5 component rendering a $2400 part "bad" and immobilizing a $175,000 vehicle…yeah they should invest is top quality parts for key components like this.

 

[ze_shark]

I agree with your point.
In foresight, the plan is always a 500 dppm failure rate, and STM is as reputable a source as Nexperia.
In hindsight, failure of a relatively cheap part, whether because of the part itself or improper design rules in its implementation, never makes economic sense.

The question, as for many other automotive components, is always how these deficiencies escaped Porsche's otherwise thorough validation protocols.

They are not disruption yahoos who think cars are software. They have a strong hardware culture. Yet they missed key hardware weaknesses in Taycan J1.1: OBC, HVAC, battery integration.

 

[ze_shark]

STM and Nexperia is "chinese stuff" ?
STM is no more or less chinese than Apple.

 

[chun]

do they fix it for who is out of warranty?

They replace the whole OBC, they do not cut it open to repair whatever got burnt.
The damaged OBC is scrapped.
What you get as replacement is identical to what you had before.
At least in Switzerland.

Worth mentioning that repairing the OBC by cutting it open is likely more expensive in hours paid in europe/usa than just replacing it whole.

I don't think Porsche themselves care to ask for an improvement of the design of the 22kw OBC anymore, as they don't offer it as a product anymore, just for warranty & repairs - and they likely have a deal with the supplier for that.

 

[chun]

Porsche will replace them until they are out of it!!

I may be remembering wrong, but in Europe it's 10 years, mandated by law, that they are obliged to have part for warranty and repair available.
So only way i see for them to improve it is if they bring back the option for a Taycan 2; or if MetaSystems goes bankrupt and they are forced to seek to supply a similar part for warranty/repairs.

 

[simcity]

I may be remembering wrong, but in Europe it's 10 years, mandated by law, that they are obliged to have part for warranty and repair available.
So only way i see for them to improve it is if they bring back the option for a Taycan 2; or if MetaSystems goes bankrupt and they are forced to seek to supply a similar part for warranty/repairs.

They may have the whole part. But it may be worth someone with car out of warranty getting an Indy EV component repair specialist onto the case.

Clearly @ct14garage does this for a crust, and has kindly shared his experience, but I'm aware of one or two others in Europe that do this type of repair work too.

 

[F1Ruaraidh]

Here’s a full breakdown of what’s going wrong and a possible fix and the DESIGN FLAW EXPOSED. Weak MOSFET chips. The Onboard Charger (OBC) has some complex circuits, including the Power Factor Correction (PFC) Circuit. This circuit relies on MOSFET transistors, and for the Taycan, it uses the STMicroelectronics 40N120AG – a 1200V, 32A rated MOSFET. Unfortunately, this component is the one that pops everytime (photos attached).

Blown MOSFETS example below:

To solve this, I’m suggesting using a different part when repairing it- the Nexperia NSF040120D7A0, a 1200V, 65A MOSFET that’s rated for 33A more than the original. It's a direct replacement.

We've already replaced the faulty MOSFETs in two cars with the upgraded NSF040120D7A0, and so far, the charging works as expected. Fingers crossed, this fix holds up over the coming months, unlike the original part that would have fried by then.

PS: Before anyone asks, the OBC is not meant to be disassembled, there are no screws to open it. You've gotta cut the back lid open with an angle grinder carefully to not damage anything inside and then once repaired you gotta weld the rear lid!

Typical component skimping leading to failure.

Sensible form fit function fix. I would use aluminium filled permabond to refix the lid rather than weld.

Thanks for posting.

Poor validation by Porsche.

 

[F1Ruaraidh]

PS: The OBC is neither designed nor manufactured by Porsche.

It is designed and manufacture in Italy ?? of all places. By a company named MetaSystems

https://www.metasystem.it/en/

Specified an Italian charger. Ohhh Porsche. (Shakes head)

 

[tycanmt]

Not to nitpick, but a silicon carbide MOSFET would easily cost Metasystem 2-3$ a piece in high volume, and this is not a high volume application. It does not invalidate your point about component cost vs failure cost, but this is not a 3904 either ...

Another question is whether they fail because of an intrinsic defect, or a PCB layout reason (heatsink ground plane for instance).

Interesting thread. I worked for ST a million years ago (not on power devices).
Some poking around on the Meta System and DEREN web sites suggests that this module's design was done in China, with manufacturing in Slovakia. I would assume they've long ago performed failure analysis and determined the design is marginal and discontinued its manufacture. If the failures had been due to component defects presumably they'd have continued manufacturing the module with defect free components and replaced the bad units in the field.

 

[ze_shark]

For context:
- the Nexperia reference which OP (@ct14garage) claims to be using as a replacement to the STM SCTH40N120 is an early 2024 part.
- the original STM SCTH40N120 is an April 2021 part.
SIC parts are fairly recent.

I wonder whether the RDSON spec (100m Ohms vs 40 typ 60 max for Nexperia) plays a larger role in this application than the current rating itself IF the Metasystem design is borderline on thermal dissipation.

Maybe this is an area where conventional design validation fall short. On such components, OEMs should peer through proof of thorough engineering verification and robust design margins. That a handful of validation samples cover N miles in harsh climate conditions does not tell you enough about design margin.

Sponsored