A primer on EV power contactors

[sukhoi_584th]

I learned today that a new co-worker was previously the reliability engineer for contactors on a well known EV, so I picked his brain. Most of what people have been discussing here is correct but there are definitely some misconceptions.

Background:

• Despite what suppliers say, contactors currently in use were not designed for this use case. They were designed for industrial electrical cabinets, motor controllers, etc. where they are under constant load. They are one of the last components to be properly redesigned for the transient loads of EV use.
• The basic design is a solenoid pushing a plate against the two pins to close the switch. The contactor is made up of the pins, plate, housing, potting materials to seal it, and it is filled with gaseous nitrogen to prevent arcing (this is key).
• They do not open/close under load, ever. The exception is e-stop type usage.
• This table is not applicable to EV usage. The red circle is where we are at for EVs. The table is intended for e-stop type usage on industrial equipment so you know how many times the e-stop can be done before expecting the contactor to fail.
  
  
• Solenoid current usually varies between high current to close the contactor quickly, and a maintain current to hold it closed. This is what the "economizer" does. For a vehicle with multiple contactors like the Mach-E usually an external optimizer as part of the Battery Management System is used to control all the contactors to save money over integrating economizers into each contactor.

Failure modes

Welding closed:

• By far the most common failure mode (vs failing open). Pitting leads to welding.
• There are a lot of different materials with different coefficients of thermal expansion in a contactor. These include the copper pins, potting material, and housing. Excessive CTE mismatch when the pins heat up causes cracking/leaks, the nitrogen vanishes, and pitting/welding quickly follows.
• DCFC is hard on contactors as it causes the copper pins to heat up the most and imparts the most thermal stress.
• If the economizer is not tuned properly the contactor can close too hard with the plate bouncing/vibrating off the pins, and cause arcing. Additionally the pins are usually a harder material than the plate and can cause physical deformation of the plate if it closes too hard.

Failing open:

• Rare and unlikely to be a hardware failure. It is probably software-commanded where the software thinks there is a dangerous high voltage fault and cuts off the battery immediately. Likely Ford was being very cautious about this, and the recall software update makes the software less likely to falsely identify a HV fault.
• The solenoid coils could be failing or there could be a problem with the electronics providing power to the coil.

Root causes of welding closed

• The contactors are probably leaking nitrogen then arcing. They are being used in a harsh use case with heavy transient loads very different than industrial motors. Industrial motor controllers are also not under pressure to reduce cost by pennies and more likely to be over-specced for the application vs our DCFC loads being close to the max allowed by the contactor.
• There may be minute product differences causing what worked fine during Ford's testing to now be failing. My co-worker had a supplier move production from one factory to another, and the contactors went from fine to failing after only a few thermal cycles. They eventually had to redesign the contactor to eliminate the failure mode as they couldn't figure out why the ones from the new factory kept failing.

My co-worker expects all the old HVBJBs will eventually be replaced assuming nitrogen leakage is the root cause. The new part number is probably Ford implementing a burn-in test or something like that on the contactors to be sure they aren't susceptible to the failure mode.

Sponsored

 

[Mach-Lee]

A couple things to add:

• Failing open can also be caused by the thermoplastic case melting/warping so the contact plunger can no longer move freely. This can cause the contactor to stick in either position.
• The heat dissipation of the design also has to be considered. A lot of heat is dissipated through the terminal connections on the bottom of the contactor. If the wires or bus plates are too small they won’t remove heat from the contacts fast enough which can cause contactor overheating. The design of the entire assembly must be considered as a system rather than individual components.
• Tolerance issues with the case or contact surfaces due to manufacturing variability can cause a snowball effect if the bad tolerance leads to a poor connection and then heating.

I don’t think the contactors are opening when they shouldn’t, rather they are overheating and/or arcing under load due to a poor thermal design and/or manufacturing flaws.

 

[Jimrpa]

Very cool info! Thanks!

 

[Blue highway]

Thanks for sharing.

One of the things that makes the fix slow ( in a calendar sense) to implement is building hours on the replacement parts to prove MTBF. If the first revision doesn't deliver, you change and start building hours and cycles with another generation.

One of the adventures of having Gen 1 of a new car.

 

[sukhoi_584th]

A couple things to add:

• Failing open can also be caused by the thermoplastic case melting/warping so the contact plunger can no longer move freely. This can cause the contactor to stick in either position.
• The heat dissipation of the design also has to be considered. A lot of heat is dissipated through the terminal connections on the bottom of the contactor. If the wires or bus plates are too small they won’t remove heat from the contacts fast enough which can cause contactor overheating. The design of the entire assembly must be considered as a system rather than individual components.
• Tolerance issues with the case or contact surfaces due to manufacturing variability can cause a snowball effect if the bad tolerance leads to a poor connection and then heating.

I don’t think the contactors are opening when they shouldn’t, rather they are overheating and/or arcing under load due to a poor thermal design and/or manufacturing flaws.

Yeah heat dissipation is certainly a consideration. Using the HV bus as a heatsink is about the only way to get heat out with the current design. Though in our case I personally doubt undersized heatsinking is the root cause. That would have shown itself quickly during reliability testing if it was such a fundamental design problem. But excess heating could definitely cause more deformation of the housing resulting in nitrogen leakage.

 

[PilotMark]

thanks

 

[dtbaker61]

A couple things to add:

• Failing open can also be caused by the thermoplastic case melting/warping so the contact plunger can no longer move freely. This can cause the contactor to stick in either position.
• The heat dissipation of the design also has to be considered. A lot of heat is dissipated through the terminal connections on the bottom of the contactor. If the wires or bus plates are too small they won’t remove heat from the contacts fast enough which can cause contactor overheating. The design of the entire assembly must be considered as a system rather than individual components.
• Tolerance issues with the case or contact surfaces due to manufacturing variability can cause a snowball effect if the bad tolerance leads to a poor connection and then heating.

I don’t think the contactors are opening when they shouldn’t, rather they are overheating and/or arcing under load due to a poor thermal design and/or manufacturing flaws.

good points....

I'd like to add that there ARE a whole differently designed class of contactors which are not dependant on a sealed case and inert gas... which is problematic with big fast temperature swings. Old School EVs commonly used open contactors for high current DC. These contactors rely on 'magnetic blowouts' designed to prevent arcing if the contact faces are required to snap open under load. When closing, there is typically a delay (precharge) before load is passed to prevent arcing and pitting.

considering that the HVBJB is a sealed area already, and does have pretty significant temp changes rapidly.... I am surprised the Design team went with sealed contactors.

I'm also surprised that the HVBJB space, fuses, and contactors cannot be reached, inspected, and replaced from the top down without dropping the battery tray. Having a removable lid for that section for access from the top down, perhaps ducted to/from the cabin even, would have enabled circulation of cabin temp air to help stabilize temps.

 

[sukhoi_584th]

The lack of maintenance access to critical components on EVs can be head scratching. From what I understand the original Model S had the contactors sealed in the battery like ours, and Tesla quickly realized that was stupid. They added a separate hatch so the HVBJB could be accessed without dropping the whole pack, and then on the newer cars you can get to it under the back seat. Yet both Ford and Rivian ignored that and have them sealed in the battery box. Are they sealed in the battery on the F-150?

 

[SuperRob]

Failure modes

Welding closed:

• By far the most common failure mode (vs failing open). Pitting leads to welding.
• There are a lot of different materials with different coefficients of thermal expansion in a contactor. These include the copper pins, potting material, and housing. Excessive CTE mismatch when the pins heat up causes cracking/leaks, the nitrogen vanishes, and pitting/welding quickly follows.
• DCFC is hard on contactors as it causes the copper pins to heat up the most and imparts the most thermal stress.
• If the economizer is not tuned properly the contactor can close too hard with the plate bouncing/vibrating off the pins, and cause arcing. Additionally the pins are usually a harder material than the plate and can cause physical deformation of the plate if it closes too hard.

Root causes of welding closed

• The contactors are probably leaking nitrogen then arcing. They are being used in a harsh use case with heavy transient loads very different than industrial motors. Industrial motor controllers are also not under pressure to reduce cost by pennies and more likely to be over-specced for the application vs our DCFC loads being close to the max allowed by the contactor.
• There may be minute product differences causing what worked fine during Ford's testing to now be failing. My co-worker had a supplier move production from one factory to another, and the contactors went from fine to failing after only a few thermal cycles. They eventually had to redesign the contactor to eliminate the failure mode as they couldn't figure out why the ones from the new factory kept failing.

Great post! I noticed you didn’t include WOT as a root cause. Given that Ford has included this as a contributing factor to these root causes, and it seems like a significant number of GTs have had this issue, do you think people should be concerned about trying to access the power in the car they paid for?

 

[sukhoi_584th]

Great post! I noticed you didn’t include WOT as a root cause. Given that Ford has included this as a contributing factor to these root causes, and it seems like a significant number of GTs have had this issue, do you think people should be concerned about trying to access the power in the car they paid for?

That's a good question. Per the math in post #11 the GT is 895 A at WOT. That's fine with the contactors for the amount of time one would do WOT per the spec sheet:

It looks to be some sort of exponential decay for current vs time allowed, so even 750 A is probably closer to 5 min allowed. But if the contactors are having CTE mismatch issues causing cracking/leaking then the rapid heating caused by WOT is more of a problem. Given that Ford specifically called out WOT in addition to DCFC for the recall (IIRC) it must be a factor.

Does anyone know what cable diameters are connected to the contactors? The time limits in the table above are only for 200 mcm cable, and if Ford used 400 mcm to allow enough heatsinking for continuous 500 A (necessary if they plan extended ~175 kW DCFC) then the time limits go up so much I can't imagine WOT matters.

To compound things, it really seems like we're dealing with out of spec parts vs Ford design team being incompetent. So that throws all these spec sheets out the window.

 

[Mach-Lee]

That's a good question. Per the math in post #11 the GT is 895 A at WOT. That's fine with the contactors for the amount of time one would do WOT per the spec sheet:

It looks to be some sort of exponential decay for current vs time allowed, so even 750 A is probably closer to 5 min allowed. But if the contactors are having CTE mismatch issues causing cracking/leaking then the rapid heating caused by WOT is more of a problem. Given that Ford specifically called out WOT in addition to DCFC for the recall (IIRC) it must be a factor.

Does anyone know what cable diameters are connected to the contactors? The time limits in the table above are only for 200 mcm cable, and if Ford used 400 mcm to allow enough heatsinking for continuous 500 A (necessary if they plan extended ~175 kW DCFC) then the time limits go up so much I can't imagine WOT matters.

To compound things, it really seems like we're dealing with out of spec parts vs Ford design team being incompetent. So that throws all these spec sheets out the window.

Those figures in the other post are off, because the nominal voltage under load is 348V not 400V. The GTPE draws 1050A under full load.

The contactors are bolted to metal plates inside the HVBJB, not wires. Size may be an issue there with the heat dissipation as I've said in #2. The short term rating is with size 4/0 (#0000) cables bolted to the contactor which are huge. Using smaller cables would reduce max amps before overheating.

 

[dtbaker61]

That's a good question. Per the math in post #11 the GT is 895 A at WOT. That's fine with the contactors for the amount of time one would do WOT per the spec sheet:

It looks to be some sort of exponential decay for current vs time allowed, so even 750 A is probably closer to 5 min allowed. But if the contactors are having CTE mismatch issues causing cracking/leaking then the rapid heating caused by WOT is more of a problem. Given that Ford specifically called out WOT in addition to DCFC for the recall (IIRC) it must be a factor.

Does anyone know what cable diameters are connected to the contactors? The time limits in the table above are only for 200 mcm cable, and if Ford used 400 mcm to allow enough heatsinking for continuous 500 A (necessary if they plan extended ~175 kW DCFC) then the time limits go up so much I can't imagine WOT matters.

To compound things, it really seems like we're dealing with out of spec parts vs Ford design team being incompetent. So that throws all these spec sheets out the window.

I suspect that the first pass current is not the main problem..... but heat building rapidly from 20C to 75-85C with nowhere to go in a sealed contactor, in a sealed compartment, with no decent heat sink BECOMES a problem.... cracks the contactor, lets out the inert gas and then at some point opens/closes under enough current to arc, starts pitting/welding... fails

so probably two issues.... undersized contactors, and inadequate cooling

if these issues can be fixed, then we can have both expected performance AND reliability

 

[ctenidae]

So, they crack open and quite literally let the smoke out.

Great post, makes a lot of sense. Seems to be the sort of thing they can sort out with enough hours of data. In the meantime, maybe I won't go WOT every chance I get.

 

[Eugene]

The lack of maintenance access to critical components on EVs can be head scratching. From what I understand the original Model S had the contactors sealed in the battery like ours, and Tesla quickly realized that was stupid. They added a separate hatch so the HVBJB could be accessed without dropping the whole pack, and then on the newer cars you can get to it under the back seat. Yet both Ford and Rivian ignored that and have them sealed in the battery box. Are they sealed in the battery on the F-150?

Sounds like Ford/Rivian were designing off old Tesla EV plans. lol

 

[Blue highway]

I suspect that the first pass current is not the main problem..... but heat building rapidly from 20C to 75-85C with nowhere to go in a sealed contactor, in a sealed compartment, with no decent heat sink BECOMES a problem.... cracks the contactor, lets out the inert gas and then at some point opens/closes under enough current to arc, starts pitting/welding... fails

so probably two issues.... undersized contactors, and inadequate cooling

if these issues can be fixed, then we can have both expected performance AND reliability

The parts capacity doesn't have much headroom.

500A for 7.5 minutes
1000A for 1 minute
MUCH less if its already hot.

It doesn't take a lot of imagination to see where this part would not have enough time to cool if you took the car to a track day. Foot on the floor is ~1000A, I bet it isn't hard to exceed an average of 500A over 7.5 minutes doing a few laps... or even just having some fun on the street.

It's easy to Monday morning quarterback, but I'm surprised a part this marginal was signed off for production.

Sponsored