Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Gen3 HPWC disassembly, with overheating issues explained!

TomServo

Active Member
Apr 10, 2014
2,601
1,922
Belleville IL
I don't recall seeing any reports of issues with them.
Thanks, and in addition I'm wiring a NEMA 14-50 appliance plug so my WC will only be pulling 40 amps. I was happy with that charge rate when I had my Clipper Creek, having sold that with our Volt I opted to splurge on the sexy WC, but still want a BIGGER safety margin and keeping the NEMA 14-50 gives me more options in the future. Again Thanks.
 

jebinc

Endlessly Vibrating MS PLAID
Supporting Member
Jun 19, 2019
10,302
15,677
Seattle area
Caution: Wall of text ahead!

So, I got a secondhand Gen3 HPWC (TPN:1457768-01-F) that had known issues at full power and of course ran it right up to 48 amps to see what she'd do(note it had updated firmware). For a little while it did 48 amps, but after a half hour in my 20f garage, it throttled back to 38-42 amps and stayed oscillating around there for a few hours until I turned it off when the outside of the case just above the identification plate reached 120f(still in my 20f garage!) Seeing a chance for a little electronic adventure, I opened it up to see what made it tick.

First, we have the familiar front. It is removed by removing two screws from the backside near the bottom, and then carefully sliding the glass downwards maybe a half inch.

Next picture is the backside, where the teeny screws(for the front) are seen at the bottom, the ground connector is the lower one, and L1/L2 are the upper ones. The circle near the middle is an optical temperature sensor, watching for the input lugs to overheat.

After the decorative glass is removed, we see eight screws holding a clear plastic plate(five were already removed in the picture!). It is gasketed and fairly weathertight, which is why there are eight screws. Remove them and the clear plastic comes off easily.

Now note the three connections at the bottom for the power and ground lines going out the cable to the handle. On mine, the screws for the power lines weren't as tight as I'd expect. Also note the small bundle of wires that clips into a black socket halfway up, they are signaling wires and a temperature sensor for the handle, and undoubtedly the button in the handle as well. Finally, note the wifi wire at the top. Be careful about pulling that off, in that it pops off perpendicular to the board, it doesn't slide off like a spade lug.

So here's where the overheating bits come into play... it gets ugly fast...
Those two big dark brown things are each two-pole relays, one relay for L2, and the other relay for L1. Tesla's hope seems to be that the two poles(called A and B in the pictures, in small blue font) of each of the two relays will take 24 of the 48 amps of that 120V leg. That hope is more of a dream, but I was surprised by how it happened. The incoming L2(I did all my analysis on L2 because it was closer to the edge of the board) spade is split into two smaller 'bars' that have a cross-sectional area between 9 and 10 AWG. I don't know exactly what the material is, but its nonferrous, I'm guessing tin coated copper or the like. I'm really surprised at this, since it'd be running 'hot'ish even with a perfect 24 amp distribution. Notably, the longer of the legs is around 4" long, the shorter is probably around 2.75" long. Right there, there's an imbalance that might throw things off, especially running the conductor at such a high current.


I removed the board from the case and was somewhat shocked(pun intended) that the PCB board was getting scorched on one of the two input poles on each of the relays. The immediate suspicion was that that pole was for some reason taking more than the desired 24 amps. The surprising bit is that for the left relay(L2), the pole connected to the LONGER input bar was the one that was discoloring, even though it should have higher resistance and therefore be carrying slightly less current.

Also note on the back of the board the optical temperature sensor, constantly looking at the input wires(from the house) to see if they are overheating due to being too small or not tightly attached. Luckily(or not), they also are checking the heat of the rest of the HPWC, even if only indirectly. This is a good thing, since it saves the entire HPWC from becoming a flaming decoration on your garage wall.

So, after some deliberation, I soldered five test leads to the underside of the board and reassembled it. The five leads were:

Arrival of L2 bus bar on the circuit board for pole A
Departure of L2 pole A relay connection(like a quarter inch of PCB board away!)
Arrival of L2 bus bar on the circuit board for pole B
Departure of L2 pole B relay connection(like 3/8 inch of PCB board away!)
Return of L2 pole A from relay.

With all these connections(and the metal bars that are relatively easily accessible from the front) and a fairly precise multimeter, I can see pretty clearly what the losses are along the way, although there's a decent chance my signals are a little polluted by the electric field of the relays. It turns out that the A pole of the L2 relay is taking around 44(!) amps of the 48 I'm charging with, and this is causing just stupid levels of heating in the box even though its not really all THAT much power being dissipated. I see around 0.21V AC is lost from the input metal bar to the output lug of the L2 side, which means that its dumping around 10 watts(0.21*48)(. Sadly, those watts have nowhere but heat to go, in a small brown package and the metal bus bar going into it.

I didn't have detailed measurements on the second(L1) relay, but saw it was dropping even MORE voltage, like 0.285V.

As an idea of temperatures in this region, the space between the two relays was reading at 250f. The relays themselves were 200f in spots, after an hour or so of running with all the covers OFF. On the bright side, it kept charging at 48 amps for around four hours, with no throttling at all.

Some might notice the rogoski coils on three of the four output lines from the relays. I assume that the software doesn't actively read them, or I'd hope something would actively shut down the system when it sees one pole of each relay is running 91% of the current. This HPWC shouldn't have been allowed out of the factory the way it is, IMHO. I also assume that the reason there are only three coils is that the HPWC could(if it cared to) see the current leaving on L1(cores 2 and 3) and assume its coming back in the same amount on L2(core not-there and 1), and thereby know the unknown not-there current individually.

So, the reason for the mismatch:

At such high current levels, even the negligible resistance of wires and relays becomes significant. Initially, I expected the longer bus bar(A) to be the issue, since longer means more resistance. Two identical resistances in parallel will cause half the current to flow each resistor, and the apparent resistance of the pair will be half each individual resistance. When one of the resistances is higher(like when a bus bar is longer) it should cause an increase in current to the lower resistance path, but clearly the longer bus bar is actually taking much MORE current than it should. This leads me to believe that either the other pole has excess resistance in the relay contacts themselves, or the very small bits of PCB board involved are just too resistant and inconsistent from one pole to the other. There may be something to this, since the PCB traces are TINY compared to AWG 9/10, and the PCB distance for the not-lots-of-current pole is somewhat longer, and the 'landing pad' for solder around the metal connection bars seems a bit small.. If I believe the 44/4 amp split of current, the itty-bitty chunk of PCB for the B side of L2 is showing about four times the resistance of the A side. It is unclear why they didn't link the A and B poles to each other at both the input and output sides for each relay, which would reduce the effect of the not-same-length feed bars. Maybe they were concerned about heating and cooling cycles cracking the PCB in between the poles.

IMHO, the design is flawed because it relies too heavily on balanced currents with nothing to really balance them, and it seems that even though there's enough electronics to monitor the imbalance, the logic doesn't seem to do much with that information. As a result, while its great to have redundancy, both paths through each relay should have been designed to be able to handle the full 48 amp load continuously without overheating.

I hope you have enjoyed this trip down Gen3 lane... and yes, it still works, but I won't be running it past 32 amps until I improve the B side connections.

View attachment 645417 View attachment 645422 View attachment 645598 View attachment 645423 View attachment 645428 View attachment 645430 View attachment 645432 View attachment 645433
Interesting writeup, thanks for sharing! Still trying to diagnose my gen 2 39A vs 48A issue. It’s looking more and more like a high resistance connection somewhere. More troubleshooting to do here!

@WilliamG
 

WilliamG

Hinge Fanatic
Apr 20, 2019
6,572
10,454
Seattle, WA
Interesting writeup, thanks for sharing! Still trying to diagnose my gen 2 39A vs 48A issue. It’s looking more and more like a high resistance connection somewhere. More troubleshooting to do here!

@WilliamG
Not having any such issue with my HPWC Gen 3. Runs for 5-6 hours at 48A at 70-80F ambient. The charge is reliably crapping out (stopping and starting over and over) in the 83-86% range due to the Plaid’s HV battery issue (which we know is the car issue given the HPWC Gen 2 had the same problem). This is also a very new build HPWC Gen 3 I have, so the OP’s issues are likely all addressed by now.

That 39A connection is an interesting one. Will be curious as to how it’s addressed, not that 39A is slow or anything!
 
  • Like
Reactions: Phlier
They apparently were able to see that it was overheating.

Owners with F-series wall connectors who charge at 32 amps or less may never know whether they have defective wall connectors. I charge at 40 amps because I built a new house before I owned a Tesla, and the wiring and breaker are insufficient for 48 amps. About the only thing about the house I didn't research myself and specify was the Tesla wiring. I told the electrician I wanted wiring to charge a Tesla at the maximum possible current. Big mistake! You have to research and specify everything to get it what you need. After I realized that the F-series wall connector was shutting down because of overhearing, I reduced the current to 32 amps and was able to charge without stoppage. With the G-series connector, I switched back to 40 amps.
 
  • Like
Reactions: kayak1 and TBrownTX
Has anyone had any overheating issues (Gen3 G-TPN suffix) with their wall connector? 1457768-01-G

I'm in Canada where it gets very cold and did not have any issues until 3 weeks ago. Constant 3 red blinking light, absolutely nothing is warm or hot from car all the way to electrical panel

Had electrician that installed come over, dismounted & measured everything and he confirmed that everything was kosher

Turn off the breaker and turning it back on sometimes work, most of the time doesnt....
 

Sophias_dad

Active Member
Supporting Member
Jul 29, 2018
2,351
2,738
Massachusetts
Has anyone had any overheating issues (Gen3 G-TPN suffix) with their wall connector? 1457768-01-G

I'm in Canada where it gets very cold and did not have any issues until 3 weeks ago. Constant 3 red blinking light, absolutely nothing is warm or hot from car all the way to electrical panel

Had electrician that installed come over, dismounted & measured everything and he confirmed that everything was kosher

Turn off the breaker and turning it back on sometimes work, most of the time doesnt....
Probably just a defective unit. At least here in the US, they are warrantied for FOUR years!
 
  • Like
Reactions: swaltner
Thanks for this information my unit - TPN:1457768-01-F - started having this issue after working fine for 3 months. I was able to call Tesla right now and they stated they will send a replacement in around 10 days. Assuming I will get the newer one with 24 ft cable also vs the 18 on this one.
 
Came here because my UMCv2 started overheating. It is summer and very hot in the garage (90F at it's coldest during the night), but it didn't really do this last summer; I'm throttling down to 16A every day now, even when I start the charge @ 4:00 AM.

A few questions:
1. We now have two Teslas again and I'm considering a WC. I'm generally against non-standard stuff (highly prefer 14-50) but with 2 cars and the overheating problems I'm having in summer with the UMCv2, I figured I'd bite the bullet and get a 48A option. Do not need the extra speed, but I can't do w/ 16A which is what the UMCv2 throttles down to once overheats (which comes in ~1 hour of charging @ 32A after resting for a while). 32A is fine w/ two Teslas for us, but not if I'm gonna spend ~$1k+ (WC+install, will need panel work), if I spend all that money I'd like the convenience of 50% faster 48A charging.

2. Sounds like the WCv3 is not really well engineered. I don't want to go thru the trouble (and expense) to have a subpar product and run into the same problems, and risks (read: electrical fire) after all said & done. Has the -F been solid?

3. Shall I be looking into non-Tesla options? Cost isn't a huge issue, I'd rather spend a bit more if it significantly reduces risk of electrical fire, as in, is there a really well engineered/over-engineered EVSE? While again I'm fine even w/ 32A, if I'm going thru all this trouble I would like to maximize the cars' on-board chargers @ 48A, so only really considering 48A or at the very least 40A options (have had my eyes on the Corded MC for 6 months, it's never come back in stock).

4. Who are you guys "calling" at Tesla regarding issues? I would at least like to try to talk to someone about the UMCv2, maybe they can replace it and I can try a new one see if it does better.
 
You should look at the wiring (if you can do so safely) where it’s connected to the house. It’s possibly a connection that is working loose due to the hot/cold cycles.

Does the plug feel hot at the car?
I've run thru all that. There is nothing wrong with the install, which was done by a Tesla-approved installer. I also don't do frequent 14-50 unplugs/replugs, I've probably only plugged/unplugged the current UMCv2 twice in 1 year. As a matter of fact, one of those unplugs was when I actually when thru and triple checked everything with the 14-50 install. The plug to the car doesn't get hot, just warm.

I've had two other Teslas in the past, both 2018s (a 3D LR and a P3D+) which we've since sold and now have the '21 Y7 and a '22 3D LR. The UMCv2 that came with the 2018 3D LR (the only one we used) never ever had any issues charging exclusively @ 32A/240V 24/7/365, often two cars/day. This is, of course, in the same location, on the same 14-50. I (unfortunately) included that UMCv2 with the sale of that car. We've only been using the UMCv2 that came w/ the '21 Y7 since last year (pretty much 1 year to the day). Last summer it threw 1 or 2 temp warnings, the entire summer. Once summer was over, did not have an issue until this June (2022). We're in Dallas where it's been 100F+ for the past 2+ weeks and now the warnings and throttling happens every day on every charge, even at night when temps are lower. Even tho it's hot hot hot, it wasn't much diff last year.

I've started a service ticket, will see if Tesla replaces it. They just pinged me that they need to "test the cable with the car."

Please, if y'all can address my other points, I'd appreciate it. The big secret why this is pushing me over the edge is that the UMCv2 barely reaches the other car. It reaches, w/o stretch, but I really need to pull the car in almost to the wall. The extra cordage from the WC would greatly simplify things (standards be damned 😟). But I'm down to go with a J1772 EVSE if there are choices substantially better engineered than the Tesla WC, sacrificing the push-button-to-open/unplug convenience of the proprietary Tesla system (which I currently enjoy w/ the UMCv2).
 
Last edited:

brkaus

Well-Known Member
Jul 8, 2014
8,660
7,285
Austin, TX
Im sorry, I misread that as the WCv2, not the UMCv2 since this was in a WC forum. My bad.

WCv3 is not as heavy duty as the WCv2, but the number of concerns here has gone down greatly with the more recent revisions. They also increased the cable length to match the v2 model, so that's a huge plus in my book.

The Tesla WC is the most cost effective 48a device out there from what I can tell. And it has more features that most with regard to power sharing. However it does not have scheduling but the car can do that.

I would submit a service request in the app for your car on your current UMC problem.
 
  • Like
Reactions: Dave EV

About Us

Formed in 2006, Tesla Motors Club (TMC) was the first independent online Tesla community. Today it remains the largest and most dynamic community of Tesla enthusiasts. Learn more.

Do you value your experience at TMC? Consider becoming a Supporting Member of Tesla Motors Club. As a thank you for your contribution, you'll get nearly no ads in the Community and Groups sections. Additional perks are available depending on the level of contribution. Please visit the Account Upgrades page for more details.


SUPPORT TMC
Top