The Troubleshooting and Repair for On-board Charger (OBC) Thread

[skylogger]

Hi KIEV / Coulomb and other distinguished quest:

It looks like I've been able to successfully repair this second OBC that I have been working on.
So the list of TRUE faulty parts were the 20 AMP fuse in the MCU, The two 1000pf snubber caps in the doghouse,
The Relay, and one of the 4.7R resistors in the doghouse.
I had to replace D301 Diode, but that was my fault for blowing it up during testing, applying a wrong polarity when testing the relay.

Since I had to look into things a bit deeper when fixing the D301 Diode, I have a suggestion for people that want to test the 4.7R resistors
and the relay in the dog house, WITHOUT having to remove the Main PCB. (you still have to lift the top controller PCB out of the way though.

1) You need to connect an ohm meter from the Neutral line going into the doghouse, and the other lead to the Neutral line coming out of the doghouse. There are a set of wires 1 black and 1 white on blade connectors that run from the top pcb down to the bottom PCB. If you look at the silkscreen, you will see one of these is Live and one is Neutral. This Neutral will be one of the ohm meter test points. The other test point is one of the pins along the white strip that attaches the main pcb to the waffle. You can see silk screen markings along the side of the strip, and you will see a "L" and a "N" along the upper left stip of pins. The "N" is the test point you want to connect the second lead of your ohm meter to. KIEV posted a picture of this test point circled in red earlier in this thread. You should see a value of just a bit over 9.4 ohms, because you are measuring across the two 4.7R resistors and a little bit of additional resistance from the inductor that is also in series. If you measure a higher value, this is an indication that the resistors are out of spec or have gone open circuit.
Leave this ohm meter in place connected to these two test points for the next test.

2) Now for the second test to check the relay, you will need a 5v power supply preferably with current limiting.
WARNING, This step requires making sure you apply correct polarity, and don't accidentally touch any other components with the probes.
if you have a power supply with banana lead connectors, you can attach some meter probes to the output of the power supply. This is because you want to apply 5v across the pads/end caps of a tiny 0805 size diode D301. BE VERY CAREFUL as you want to apply voltage REVESE BIAS to this diode, so that the + is connected to the Kathode (the bar on the silkscreen) and the - is connected to the anode.
If you get these swapped wrong, you will blow up the diode. The reason this is applied REVERSE BIAS, is that this is a protection diode, and you are not wanting the diode to conduct, you are just using the two ends of the diode as test points, because this is the only place to make
contact to the coil of the relay on the top side of the PCB with the main board still mounted in the box. (other than the flex strip connector which is too fine pitch to use as a test point).
So when you apply +5v across the diode, you will hear the relay click. Also, the ohm meter that you connected to the two Neutral test points earlier will change from around 9.4 ohms to ZERO Ohms when the relay closes. This shows that the contacts are correctly shorting out the 2x 4.7 ohm resistors. If you don't see the value go to zero, this means that the relay contacts are not closing properly.

It's a bit tricky applying the voltage across the little SMD DIODE, and it easy to make a mistake with wrong polarity, or accidentally bumping into another component with the probes which could cause other damage, but This test will allow testing the RELAY with out the hassle of disconnecting all the wires, and standoffs to get the main PCB removed from the box. If from these test, you find the relay or the resistors are faulty, then you will still need to not only remove the main PCB from the box, but you will also have to desolder all of the pins along the two white strips to be able to remove the waffle heat sink, so you can access the pins of the relay and resistor on the bottom of the PCB for replacement.

 

[bambou]

Hi all,

As you can see, I'm new here, and I'm very impressed by all your knowledge and capacity to explain difficult things.
So congrats and thanks to all of you

So my theory predicts two different outcomes, though one is more likely than the other.

You explanation made me wonder about something (though I may have misunderstood you, because my knowledge is very low about electricity and electronics).
Is it possible that stopping the charge without the car knowing it have this same effect as the one you describe (e.g. mains failure, programmable switch)?
I mean, when you press the button of the connector to unplug it (or when the car decides to stop the charge on its own), then the current decreases slowly.
But if the charge box is unplugged, then could the sudden variation imply the same destruction as the one you explain?

Thanks again to all of you.

 

[kiev]

@bambou, welcome to the forum, and i think yes to your question--if the cord were jerked out suddenly then the controller would not have time for a graceful shutdown sequence and it could cause the same sort of damage we are seeing.

@ skylogger, thanks for the detailed information on testing the AC relay and resistors, especially on how to do it without removing the bottom board.

@coulomb, thank you for the excellent relay theory discussion. i can see that if the contacts open or go frosty during charging then the snubber caps could be blown. Or it can prevent the onset of charging--the feedback voltage from the AC detection circuit will not meet the commence to jigglin' criteria. i still don't see how the MCU fuse gets taken out, but that may be related to something going on between contactors in the pack and the DC/DC converter.

My thoughts about a weak or old Aux battery were as a precursor for the AC relay going open. If the 5V relay coil power supply was disrupted by a flaky 12V supply, which itself is fed and controlled thru another set of those cheesy little relays in the fuse box up front, then this whole failure scenario begins.

i was taught that sliding electrical contacts are to be avoided, in some cases at all cost. Here we have the equivalent of a sliding contact in the points of the relay(s).

Some thoughts to consider:
Is it typical for relays to fail after 5 or 6 years?
Has the cycle life been exhausted?
Can we take preventative action by replacing the relays as a maintenance item?

Thank Goodness, knock wood, there doesn't appear to be damage to the FETs or other waffle plate components.

 

[coulomb]

It looks like I've been able to successfully repair this second OBC that I have been working on.

Wow! Two out of two, and not a semidonductor has died in either of them. Well done!

Kiev: I assume that when the doghouse capacitors blow, they are at least part of the time in a short circuit condition, putting two fuses with a handful of low-impedance components across the battery pack. The more-accessible MCU fuse is likely designed to blow first (a faster blow type), so that it blows before the fuse on the on-board charrger can blow. Perhaps the extreme heat of the plasma around the shorted capacitors eventually causes them to end up open circuit, using energy from the 200 μF capacitor at the output of the charrger. This current does not go through the on-board charrger fuse, since the MCU fuse in series with it is now open circuit.

Now that I type this, does the energy stored in the 220 μF capacitor seem to be enough to cause such extreme damage to the doghouse capacitors? I mean if you short it with a screwdriver, you'll wet your pants and maybe have a little chunk melted out of your screwdriver, but is that enough to cause the explosive damage we've seen to the doghouse capacitors? (Disclaimer: your wetness may vary )

 

[skylogger]

One really annoying thing that I have is the fact that so far, we can't get a OBC to actually run on the bench, without being re-fitted in the car. I've supplied the two 12v supplies to the top pcb, and 240vac to the input of the OBC, but the OBC sits there doing nothing. In particular, I see that the relay inside the doghouse does not even close when testing on the Bench. I added wires across D301 and ran them up above the top pcb to a LED and a 330R resistor
so I could have an indicator when voltage was applied to the relay. On the bench, I could see the Three big electrolytic caps charge to 347v but the relay never closed. This was not a problem at the time, because the IGBT section was also not running, so the only current that the 2x 4.7R resistors had to deal with was the electrolytics, and once they charged, there was not much current flowing through the resistors. Once I fitted the OBC back in the car, and the charge cable plugged in, you can hear the contactors click, the fan comes on, and within a second the LED that I added shows the doghouse relay closes. I guess this means that what drives this relay is not just a dumb timer circuit, but is reliant on the controller on the top pcb to be happy with everything before the relay gets activated. The Things that are currently not being connected to the OBC when its on the bench are the 360V Traction battery OUTPUT, 12v aux OUTPUT, and CANBUS lines. I would think that not having the 12v AUX OUTPUT connected would be a problem with the DC-DC Section running, but would not think this would keep the top OBC Section from running, as long as the 2x 12v supplies power the top pcb. I think it comes down to the OBC not being happy and running with no 360v traction battery load, since the output current and voltage sens values would all be wrong. The other problem would be not having proper CANBUS communications from the EV-ECU may prevent the charger from running.

Now that I have a running OBC in my test bed car, I will try purposely cutting one of the CANBUS lines to see if the OBC still starts charging. I think it will detect AC, the contactors will close, the fan will come on, the charging light on the dash will come on, but this will only last for a few seconds and then all will shut off due to a timeout error with no CANBUS Communications.

<EDIT>: I purposely cut one of the CANBUS Lines to see what the OBC would do. When I plug in the charge cable, I hear the fan start, showing AC was detected, but after a few seconds this stops and no charging even starts.
The LED that I added to show if the doghouse relay is activated, never lights up. This means that during the AC Detection phase, current is running through the 2x 4.7R resistors, but due to a CANBUS Connumications timeout, everything stops before the 2x 4.7R resistors even get bypassed, and the relay is never activated, and the IGBT's never run. So at the end of the day, we can't do full bench testing unless we can simulate the CANBUS Communications or simulate that the controller on the top PCB is happy with everything so it activates charrging.

 

[Gorfllub]

This is such great news!

A comment about pulling the AC out with no shutdown command..... I am certain there are people that do this every day.

I know people who run their chargers on a timer. Except the timer is on the house circuit and cuts the power to the evse. So these cars get this immediate loss of AC any time the system shuts down prior to the battery being full. This can be every time they charge depending on their schedules and needs.

Seems like these people should have a higher failure rate if that is indeed a cause.

Is that the case? Those of you with dead chargers..... Do you power your evse on a timer or separate switch?

 

[skylogger]

On the second OBC that I just finished working on, I've been told that the charging failed during a night that there was a storm and heavy rain. This could point towards a blackout, brownout, or power surge as the source of the falult.
The 12v AUX battery had been replaced 2 years ago by RAC. When I first had a look at the car, the 12v aux battery was sitting at 12.64v which sounded pretty good to me. I told the owner of the car to take it to one of the battery suppy places, and do a more complete test of the 12v battery under load. She came back saying they tested the battery and the results were "marginal". They also said that the battery was one of the smaller types, and a larger battery would be a better option. I've suggested to her that the next time she gets some spare cash and its convenient, it would be a beneficial to change out the battery, so as to possibly avoid obc failure in future. She has also put on order a Ampfibian Surge protector.

 

[footswitch]

Hi skylogger,
Thanks for your effort and congratulations on your success so far.
- When you cut one of the CANBUS lines, and when the AC kicks in, does the "charging indicator" in the dashboard flash on and off? Or is it steady?
- I don't believe any of the relays to be malfunctioning. Too little use... [EDIT] just now I read your previous post mentioning that the contacts wouldn't close to zero ohms, so sorry for my uninformed opinion :roll:

 

[bambou]

I know people who run their chargers on a timer. Except the timer is on the house circuit and cuts the power to the evse. So these cars get this immediate loss of AC any time the system shuts down prior to the battery being full. This can be every time they charge depending on their schedules and needs.

Seems like these people should have a higher failure rate if that is indeed a cause.

Is that the case? Those of you with dead chargers..... Do you power your evse on a timer or separate switch?

That's why I had initially asked the question about "hard" shutdown.
At least one of the French guys getting trouble with their OBC had been using a programmable switch.

 

[kiev]

i soldered some tiny wirewrap wires to the AC relay coil pins and measured 9.9R with relay off and 0.2R when closed. With 5VDC supply it was drawing 83mA. The spec is only .1R of contact resistance, but i didn't zero out the meter and know that my meter leads have .1 or .2 themselves--so it's nearly zero with this setup.

The relay would drop out when the voltage was lowered to ~0.7V, and would energize on the way back up at ~ 2.3 VDC. So it seems the relay in jay's failed box is ok. It doesn't make a very solid sounding clunk or snap, but i guess you can't judge a relay by the cover...

 

[coulomb]

i soldered some tiny wirewrap wires to the AC relay coil pins and measured 9.9R with relay off and 0.2R when closed.

I assume that's neutral in to neutral out.

So it seems the relay in jay's failed box is ok. It doesn't make a very solid sounding clunk or snap, but i guess you can't judge a relay by the cover...

I dunno. I think the lack of a satisfying clunk or snap is somewhat damning; I'd consider it somewhat suspect at this stage. It possibly takes less contact pressure to drive a multimeter to a zero ohms reading, but pushing up to 36 A peak into capacitors might require a little more force. There are mechanical things that can go wrong in a relay, which a multimeter test might not always find.

But I think your finding does weaken the "relay as prime suspect" theory.

 

[jray3]

Is that the case? Those of you with dead chargers..... Do you power your evse on a timer or separate switch?

Good theory, but I plead not guilty. Mr Bean’s charging sessions normally end automatically or via the proximity switch on the J1772 handle. I’ve unplugged 120V from the wall before disconnecting the handle a very few times, but not 240V. I never implemented the previously-discussed 240V contactor for scheduled charging, and EVen if, an input line contactor would only be suspect if used to interrupt charging. If used to start charging, it would work just like a normal EVSE startup.

 

[bambou]

and EVen if, an input line contactor would only be suspect if used to interrupt charging. If used to start charging, it would work just like a normal EVSE startup.

Yes, my worries concerned the ending of the charge, not the beginning.
When someone tries to optimize costs by charging only when the electricity rate is lower, then he may end the charge before the battery is full, and then enter into the destructive theory described above.
But I understand you have had trouble while not having unplugged the main socket, so ok.

 

[kiev]

... I’ve unplugged 120V from the wall before disconnecting the handle a very few times, but not 240V.

That's an interesting clue--It only takes once to cause damage.

i had assumed that nothing like that had occurred, but this raises a possible culprit for your box failure.

 

[skylogger]

Footswitch: When I cut one of the CANBUS lines, I was busy checking that the LED that I added to confirm that the doghouse relay was closing (which it did not do) so I was standing at the back of the car. I was able to hear the contactors and fan go on, so I knew that AC input was detected, but silly me, I failed to take note if the charge light on the dash was on or blinking.
Maybe if I get some time I will run the test again.

KIEV: I am thinking with the micro pitting situation on the relay contacts, when you run the relay test, you might have to open and close the relay 20+ times or more. You might find that out of the 20 relay closures, you measure 0.2 ohms the majority of times, but there might be 1 or 2 times when the relay closes that a high resistive value might come up if the contacts are intermittent or the micro pits line up.

 

[kiev]

Good idea, i'll have to run some cycles on it and see how it does.

Since the waffle plate is off there is no path to charge the caps, but i could rig something up to put ac thru the contacts. Too bad there is no way to inspect the contacts without destroying the relay. i was hoping this relay would show bad so we could have a culprit and fix it...

 

[Quixotix]

I apologize if this if this has been answered, but I"m struggling with the terminology in this thread -- vertical pcb (I don't see one), dog house???

Anyway, I'm attaching 2 pictures (I hope). I think that's a capacitor that's completely burned up. Am I right? Assuming it is, does someone know what size it is? Maybe it's the same size as the one next to it, but with the potting I can't read that one either.

About that potting, any hints as to how to replace the burnt part? I'm thinking I will: 1) carefully mill out a pocket where the burnt part is (milling is easy), 2) desolder and remove the heat sink assembly (what I pain) so I can get to the pins for the capacitor, 3) solder a new capacitor into the original position, 4) re-solder the heat sink assembly on.

Is their a better way to fix this?

Should I replace the other capacitor next to it (or any other components) while I have the heat sink assembly off?

Are there likely any small components under the potting that are within a mm or so of the burnt capacitor? I'd like to avoid cutting off any components with my mill other than the burnt one.

Should I re-pot the new component?

Thanks for your help.

 

[coulomb]

I"m struggling with the terminology in this thread -- vertical pcb (I don't see one),

Your charrger doesn't have one. Some (roughly half?) of the chargers have a vertical PCB with surface mount capcitors on it instead of the two through-hole capacitors.

dog house???

That's just the potted area walled off with plastic. Your burned capacitor is inside one, but there are others on other PCBs.

Anyway, I'm attaching 2 pictures (I hope).

I can only see the first one, and only on my phone. If it's hosted on something like Google Drive, you need to change the sharing settings so that everyone can see it. [ Edit: I can see them both now, thanks! ]

I think that's a capacitor that's completely burned up. Am I right?

Yes, that's badly burned (based on the first photo only).

Assuming it is, does someone know what size it is? Maybe it's the same size as the one next to it, but with the potting I can't read that one either.

The two are the same. They are 1 nF (1000 pF) 3150 V capacitors, as specified in this post: Murata DEHR33F102Kppp . [ Edit: the "ppp" might actually be three special characters that aren't rendering well, but in my PDF reader, these characters show up as empty boxes but match "ppp" in a search. ]

Whether they should be replaced with something "stronger" is debatable, but you are unlikely to find a higher DC voltage rating.

I'm thinking I will: 1) carefully mill out a pocket where the burnt part is (milling is easy), 2) desolder and remove the heat sink assembly (what I pain) so I can get to the pins for the capacitor, 3) solder a new capacitor into the original position, 4) re-solder the heat sink assembly on.

I would be careful with milling. You don't want to touch the PCB tracks. Others don't seem to have had much trouble removing the potting material with simple tools (perhaps a screwdriver or wooden chopstick).

If you're very lucky, you might be able to leave some pigtail exposed, and solder quickly to the pigtail, so you don't have to desolder the many connections to the Waffle Plate™. You really need a proper desoldering station to remove that plate. Be quick, so that you don't melt the solder under the board. However, it might also be possible to completely remove the capacitors from above, and re-solder them all from above as well, again avoiding removing the Waffle Plate.

Should I replace the other capacitor next to it?

I think that would be a good idea. It is in parallel with the other one, so it has seen exactly the same voltage spikes as the burned one.

(or any other components)

We still haven't found the Root Cause. So it's possible that eventually we'll know what else to replace. But unfortunately, we're not there yet. One thing we're suspicious of is a too-small or too-old auxiliary battery. There are also suspicions about the pre-charge relay. But the relay definitely can't be replaced without taking off the Waffle Plate.

Are there likely any small components under the potting that are within a mm or so of the burnt capacitor? I'd like to avoid cutting off any components with my mill other than the burnt one.

I don't believe that there are.

Should I re-pot the new component?

That would be a good idea. Cars are a high vibration environment, so the leaded capacitors could use some mechanical constraining. But it's also good to keep dust and grime away from the high voltage DC elements like those capacitors. It doesn't have to be the exact same material; I'd use neutral cure silicone. Neutral cure is definitely required, as acid-cure silicone will conduct electricity. Silicone has excellent insulating qualities. If you use some other material, make sure it is designed for electrical work (silicone generally is not), or is known to be a good insulator. I find it hard to source suitable potting material.

 

[Quixotix]

Thanks for the information Coulomb.

I would be careful with milling. You don't want to touch the PCB tracks. Others don't seem to have had much trouble removing the potting material with simple tools ...

If you're very lucky, you might be able to leave some pigtail exposed, and solder quickly to the pigtail, so you don't have to desolder the many connections to the Waffle Plate

The traces in that area are on the bottom side of the PCB, right? Anyway, I'll try the hand tool method. I had only LOOKED at the potting material and I assumed it was a hard epoxy. Now that know it's a rubbery material (silicone?), I can see that removing the potting with a hand tool would give me the chance of leaving pigtails from the old components.

You really need a proper desoldering station to remove that plate. Be quick, so that you don't melt the solder under the board. However, it might also be possible to completely remove the capacitors from above, and re-solder them all from above as well, again avoiding removing the Waffle Plate.

I noticed others with more skill than I were having trouble removing the waffle plate. I was thinking i might try to do the replacement all from the top side.

QUESTION: If I raise the new capacitor up so I can solder from the top side of the board, the longer leads won't be a problem from an electrical point of view (i.e., change the inductance, capacitance, or whatever)?

One thing we're suspicious of is a too-small or too-old auxiliary battery.

Regarding the auxiliary battery, here is what I know about mine (just in case it's a useful data point):

The failure in my onboard charger presumably occurred while the car was charging at home. After the failure, I drove the car -- with the 12V batter and the ! warning lights on -- for a total of maybe 45 minutes. I drove until the car shut off presumably due to low aux battery voltage. Half of the time I was driving I had the A/C and headlights on. So, the (original) auxiliary battery had a reasonable capacity prior to the charger failure.

I charged the auxiliary battery immediately after towing the car the last couple of miles home. I don't remember exactly, but when I checked the aux battery voltage it was in the mid 11 volt range before charging.

However, after the dealer had the car for about a month, they claimed the auxiliary battery was no good. The tech said it wouldn't last long enough for the car to finish its diagnostics (or something like that). They put in a new battery.

I have the old battery here. I just checked the voltage. It reads 12.7 volts after sitting (not connected to anything) for a week.

I don't have any real battery capacity test equipment. But, if it would be useful, I can hook an old headlight to it (along with an amp meter and volt meter) and I can give you an idea of its capacity.

 

[Quixotix]

They are 1 nF (1000 pF) 3150 V capacitors, as specified in this post: Murata DEHR33F102Kppp . [ Edit: the "ppp" might actually be three special characters that aren't rendering well, but in my PDF reader, these characters show up as empty boxes but match "ppp" in a search. ]

Did anybody notice what the Murata data sheet says about these DEH series capacitors:

See pdf page 60 (paper page 58) in this document: https://www.mouser.com/datasheet/2/281/c85e-522724.pdf
I added the bold.

Applications

Ideal for use on high-frequency pulse circuits such as
a horizontal resonance circuit for CTV and snubber
circuits for switching power supplies.

Do not use these products in any automotive
power train or safety equipment including battery
chargers for electric vehicles and plug-in hybrids.
Only Murata products clearly stipulated as
"for Automotive use" on its catalog can be used
for automobile applications such as power train and
safety equipment.

I guess we use them anyway because there isn't a better option? Do you think it was our MIEVs that caused this statement to be included?

Note: The 3 empty boxes (or ppp) at the end of the part number are where the code letters for the lead style (long, short, crimped, etc) and packaging type go. See pdf page 5 (paper page 3) for the code letters.