DIY Electric Car Forums banner

TCCH Elcon charger troubleshooting and repair

120K views 399 replies 35 participants last post by  kennybobby 
#1 ·
Several of us dudes have been looking into some failed chargers to determine if repairs are possible. In the bad units we have seen that the input start relay and by-pass resistors have been overheated and failed. The main board has been traced and sketched up in another thread: http://www.diyelectriccar.com/forums/showthread.php/tcch-elcon-1-5kw-charger-schematics-89470.html

We have made some good progress on tracing up the control board, and in the process found a defective FET driver that was pulling down the 12vdc supply. This may be the mechanism for causing the relay and bypass resistors to get so hot.

The fet driver was drawing excess current from the 12vdc supply (the viper circuit) such that the viper went into re-start mode. But doing that would cause the start relay to lose hold-current and go open--now the full 120vac would be carried thru the 2 paralleled 150 ohm bypass resistors (~190 watts in 1 watt devices).

We hooked up an oscilloscope and applied power to the viper and found it was in an infinite start-up loop and never able to kick in to regulation mode due to the Fet driver load. When we isolated the driver out of the load path, the viper came up and went into normal regulation. It appears to be a good device but should be swapped out anyway since it may have been overstressed.

The control board has two separate 12vdc supplies generated by the viper--one for the analog section and one for the digital portion of the control board. The analog end shares power with the start relay and the digital end shares power with the output relay.

If anything should cause the analog-end 12vdc supply on the control board to get pulled down low, it will take out the start relay and cause catastrophic failure of the charger due to the by-pass resistor configuration.
 
See less See more
#2 · (Edited)
Found the culprit!

Got the control board traced and posted the schematic.

The 3846 SMPS chip has two outputs which control the two FET driver chips. The output stage of the B channel was internally shorted and causing the 12 volt supply to be pulled to ground. It caused the FET driver to turn on both the high and low side FETs of one totem pole (1/2 H-bridge) at the same time--that probably made some noise, it definitely blew some parts, but it should be possible to replace all the damaged components and repair this unit.

What would be the symptoms? Measured the resistance from +12 vdc supply to ground and it was only about 2 ohms. Removed the FET driver chip and the resistance went up to 90 ohms. Removed the 3846 SMPS chip and it went up to 1 Mohm.
 
#3 · (Edited)
Coulomb's index

Coulomb's index

I've decided to put an index here to what I consider to be the more useful posts in this thread. Yes, it's mainly for my own use, but others may find it useful as well. PM me if you find a post you think is worth adding to this index.

Re-assembly notes, including the heaksink clamp removing tool, and Reassembly tips.
Transformer transplant.
Discussion on the reliability of the bridge rectifier, and effect of disconnecting mains under load. More here.
Drill this through if replacing the rectifiers on the high voltage models.
8 kW model photos.
Using jumpers in troubleshooting; more troubleshooting tips.
Tests for while the MOSFETs are not yet replaced.
SOT-23 devices marked "A1t".
"Desaturation" protection schematic.
PCB with C38 (large capacitor near MOSFETS) removed. Also C2 and C46 small capacitors nearby.
Improving creepage distance between MOSFET leads.
Input relay photos.
The mystery of the faint crackling sound solved.
How NOT to replace the heatsink clamps.
Start of discussion on replacement pre-charge resistors (150Ω 2-3 W near input relay). Continues for several posts. Using 750 mA slow blow fuses instead of a thermal fuse.
Measurements possible without taking out the main PCB. Next post has possible repairs without taking out the main PCB.
Replacing the HF transformer.
Q7, Q8 PFC MOSFETs
Cleaned-up control board (daughter board) image (large)
A and B PWM outputs; the U12 latch.
Non-Viper power supply model


For convenience, some links to revised schematics (in another thread):

DC output section schematic.
Control (daughter) board schematic.
AC input schematic.

---------------------

WARNING.

The DC bus capacitors collapse to around 10 V soon after the input (pre-charge bypass) relay clicks off, but the capacitors in the output section retain a nasty, possibly lethal bite for roughly a minute after turning off the power.

While there is an output relay, its contacts are bypassed with a resistor, which would limit any shock from touching both outputs to a likely non-lethal belt, but it is still dangerous, and you should take precautions.


---------------------

As I told you both 150 ohm resistance of my 2.0kw charger were burned. This happened after the plug with the 7 pins (CAN-communication) has broken into two pieces (exactly there where the plug is sitting in the case/box of the charger)
So you're saying the plastic of the 7-pin round socket split in two? I've never seen that.

and I plugged them together but unfortunately in the wrong position then the resistance burned and I heard little noise some where. Could you help me?
I've already mentioned the bridge rectifier. Have you checked that for short circuits?

I like to know what could be happened? Which piece/s died and how could I localise it? As I told you I changed the resistance and they burned again. So there must be wrong something more?
Yes, a good chance is the bridge rectifier. Check also for flashover from one of the electrolytic capacitors to a small capacitor's lead. They sometimes use a larger-than-designed-for capacitor, and the clearance is sometimes too small. It could also be the main switching MOSFETS, or more likely the PFC MOSFET. You just have to trace the circuit through, using the schematic as a guide. No two repairs are the exact same.

Where is the difference between the 1.5kw charger schematics and the 2.0kw charger schematics?
I'm not aware of any 2 kW charger schematics. From what I have seen, there doesn't seem to be a whole lot of difference; they just seem to use larger inductors in the 2.0 kW design, and different limits in the firmware. [ Edit: they also use a different number and possibly size of various electrolytic capacitors, from memory. ] The MOSFETs may be different part numbers with higher specifications, but I suspect not. The layout is a little different to accommodate the larger inductors. We're lucky to have the 1.5 kW schematics traced out. If you end up tracing part of the 2 kW schematic, it would be great to see what differences or even what commonality you can confirm.
 
#4 ·
Thanks for your reply.
But I am still searching to find the problem.
I forgot to tell you that the 7 pin plug haven't been in the plug they were loose (see the attachment) , so without any protection to each other. I think that the minus pole/pin and the plus pole/pin were in contact, so a short circuit happened. If that happened, what influence could have had this fact?

Another thing I would like to know is: Were generate the thing the 12Volt?
Do you have an idea?

Thanks
Harald
 

Attachments

#6 · (Edited)
Another thing I would like to know is: Were generate the thing the 12Volt?
The 12 V is generated by a "Viper" chip. It's a tiny 8-pin DIP thing with a small copper heatsink sticking up.



You can tell this is a 1.5 kW charger; there are two large capacitors near the Viper. (2.5 kW chargers have three there.)

The schematic for this part of the circuit is in the schematics post; it's the one called low_voltage.JPG.

THe yellow thing with the "2" on it is the transformer that generates the two 12 V supplies. Note that the "isolated" output isn't all that isolated; there is some 3k of resistance from it to the negative end of the battery being charged.
 

Attachments

#5 · (Edited)
Oh sorry to hear--that is pitiful. The chassis connector has come loose and slid out and allowed the sockets to short circuit together.

The +12v and ground have a 2 Ohm resistor between the control board and the socket in the circular 7-pin, so those may have burned up.

i see in the second photo that the by-pass resistors near the input relay have started to scorch and burn the yellow staking compound--that is a classic symptom of the failure mode in which the relay gets turned off due to overheating of the low-voltage supply controller chip (the ViPer), and those resistors are left trying to carry all the current into the diode rectifier bridge mounted on the central rail of the heatsink.

i'll have to look at the schematic again to see what other damage from the circular 7-pin is likely...
 
#7 ·
Greetings - I have been following this forum for years and love the info. But I have a PFC-2500 that is now not charging and I have not found anything like this mentioned.

The charger does everything it should during power-on and in fact turns on the output relay and shows a solid red LED. But it never puts out any current and will sit there for hours before it does a time-out error. It has been working fine for 2 1/2 years charging everyday.

I have been digging into the circuit (thanks to all on this thread for the schematics) and have yet to find anything. It is almost like it thinks the cells are full as far as SOC but why? I have checked the feedback resistor dividers and they are fine, etc.

I am hoping someone has seen this or has some ideas and I would be grateful for any feedback. I am a power EE by trade and understand the circuit and the one thing I keep coming to is the program IC (which, without knowing the FW you are making a lot of assumptions). Thanks
 
#9 ·
I am hoping someone has seen this or has some ideas and I would be grateful for any feedback.
I have had the situation where I've set a voltage larger than the pack's voltage, it has a high current limit, and next to no current flows. I can't pin down exactly what's going on as yet, but it seems to want to see a moderate difference between the voltage setpoint and the pack voltage before meaningful current flows. But it sounds like in your situation, where it used to charge just fine, this doesn't apply.

It could possibly be the pack; if a cell is high resistance the voltage might rise with no effective current. Does the pack drive the vehicle? Do you see any current at all, or literally zero? If non-zero, is there any voltage rise on the pack? I think it can't be the output fuse, because then it would not see the battery voltage, and would bring up an error, and not close the output relay.

The next thing to do may be to set up a serial port connection to get information about what is happening. I don't have access right now to my schematics for the serial interface, and I can't remember if we published them. With a $10 circuit and a $20 serial to USB dongle, you could capture some 2400 bps data and post it here for us to analyze. That would tell us if it thinks it's too hot, or if it's trying to send current, and so on.
 
#10 ·
Thanks for the replies. I did check all of the fuses (including the SM ones on the CB) and all are good. And my battery pack has been running normal and exhibits good ESR drop under load, so I don't think that is any issue

A little more clarity now that I think back; I have noticed lately that it does not "start" charging right away and pulses current for awhile (like minutes). I never gave that more thought until your mentioning the voltage difference. And there is more; I have some mornings seen the pack not fully charged back but it was at 80-90%.

When I "faked" the charger into thinking a battery was connected, I placed 4V at the input to the CB (which equates to 3.2V per cell) and the charger was trying to ramp up current. But when I plugged it back into the pack, I measured the sensed voltage going to the CB for both the DCDC output and pack and they were only 10mV different. The charger then just sat there happy as a pig in mud doing nothing.

I have read a few online issues where people report not getting the full charge with this charger and even one where it only would charge if the pack was below 50%. So where is all of this going? Well....

My charger puts out near 400V since I have 114 cells. I also noticed the divider is made with 2-390K leaded, 0.25W metal film resistors. Being a Power EE I remembered something; voltage coefficient of resistance and LEV. Basically, these resistors have a MAX operating volatge of ~250V independent of the power dissipation. A long exposure to high voltage (~200V for each resistor) leads to a decrease in resistance (both due to applied voltage and aging for a continuous applied voltage). And guess what; the resistors on the battery side are ALWAYS exposed to high voltage!

If these resistors are loosing resistance over time because of this, a higher than expected voltage shows up at the CB and the charger thinks there is more SOC than real. It can even think it is full and then will not do anything (or it will pulse once and think it is almost full). This would explain what others are seeing and why mine is acting this way. I also notice it seems worse when cold, and these resistors are lower at cold temps.

I am going to carefully measure the effective resistance and then change it if it is off. If this does indeed fix the issue, I will replace these resistors with ones designed for high voltage applications (they make them).

Thoughts? A little off the wall theory but everything else seems fine.
 
#11 · (Edited)
Thoughts? A little off the wall theory but everything else seems fine.
Your theory sounds quite plausible.

It could also explain a few things. These chargers have EEPROM values that are supposed to compensate for the tolerance of the measurement resistors. When you have multiple resistors in a chain as there are here, the tolerance adds. So you'd expect the voltage measurement to be a bit better than it actually is. If these resistors are being affected by the effect you describe, that could explain the lack of accuracy. (Of course, poor calibration procedures or inaccurate instruments could also explain it).

So I'll be very interested to hear the results of your investigations.

The calibration issue also implies that you should ideally calibrate your new resistors (saving new values to the EEPROM), or alternatively you may need to trim your new resistors so that they are the same value as the original resistors were when calibrated. An easy way to do this would be to examine the serial data stream; the battery voltage is reported among many other values. So you can trim your new resistors until the value in the data stream agrees with the value measured by a trusted meter. Of course, you might also get away with replacing the resistors with their nominal values, and hope for the best.
 
#13 · (Edited)
As far as getting the right resistor value, I do have one saving grace and that is the 10 programs; they are for changing the number of cells. Right now it is set at the MAX so I can't play with setting it higher. But if I over compensate the resistor so I have to use a "lower cell count" which to the charger is a smaller voltage it expects to see, I can then go to a higher cell count setting to "fine tune" the output voltage.

I injected the 4V at pin 31, which is the battery pack feedback. As far as current ramping, the charger was actually slowly ramping its output voltage starting at the "perceived" pack voltage and ramping all the way 420V, then stopping and repeating. It would be ramping up the voltage (and it was slow like 5V per second) and watching the current. This is how it would work to setup the output current for the CC mode.

But when plugged into the car battery, it never ramped and sat with the output set to the battery voltage with 0 current (I verified it was switching and running with expected gate drive voltages for the H-bridge). It was doing everything it should for setting a 0A output which tells me it thinks the battery is full.

I will look into getting a serial device eventually if this ends up being a pain or does not fix the issue.

Thanks

P.S. The charger never gives any error code either plugged into the battery pack, which is another indicator it seems to think everything is A-OK
 
#16 ·
The 4V was referenced to the output ground (which is the local ground on the right side of the controller board). Yes, there are 3 "grounds" from what I can tell and they are not the same!

The output relay is switching but you have a very good point; it might not actually be closing the connection. I will check that and thanks.

P.S. It is down on my priority list since I have received my new Elcon charger so my PHEV is back up and running (but I want to repair the old one to have a backup).
 
#19 ·
That's a really disappointing part of the design; minor accidents like this should not cause things to blow up. They should provide some sort of current limiting to the power pin.

It sounds like this would be the Viper chip. It is certainly possible to buy this chip; I have a few myself, and they cost only a few dollars (excluding shipping). It's a bit of a pain getting the PCB out to get to this chip, and the black gunk they put on the PCB is irritating. But if you can overcome these two things, it is certainly repairable, assuming it's just that chip. This chip provides two sets of 12 V power, including power to the microcontroller via a small voltage regulator.

If you go poking around in there, be aware that the heat sink of the tiny Viper chip (8-pin DIP, the heat sink is a small postage stamp sized piece of copper soldered to four of the pins on one side) is at lethal potential. It's lethal when running, and probably for several minutes after the power is turned off. Caution!
 
#25 · (Edited)
No, I believe it's 2 ohms. Maybe that's to turn it into a fuse. [ Edit: I have suspicions that it's an actual SMD fuse. ] 2K would not fuse from 12 V (< 0.1 watts power dissipation).

If the micro is still working (e.g. the red/green LED still lights), then this might be all that's wrong.

Edit: also check R39, on pin 2, depending on how you shorted pin 3 to ground.
 
#28 ·
Oh. Well, the click means that the Viper is OK after all; it provides the power for everything except actual charging, including pulling in the power relay (the click a second or so after applying power). So it can only be something associated with getting power to the micro and the 3-pin connector. My feeling is that it's probably quite repairable, and I don't know why they'd say it's not. Perhaps they'd prefer to sell you a new one.
 
#33 ·
I got zero v at pins 22,23 and 1,5. Is microprocessor history?
Probably not. It can't be expected to work without power, and the fact that it has no power now suggests that it is probably safe from whatever caused the power to be lost. (Though it's still possible that there could have been some sort of surge that burned nearly everything.) In fact, if the processor is toast, it's not actually game over. It's a cheap part (a few dollars), and tricky but routine to replace it. We have images to reprogram it with, and even (in theory) the knowledge to replace the calibration values. We've only done about one of these processor replacements, and I can't remember where it's up to, but I think it was working well enough. But that's not game over.

What would be really tedious would be 75% of the chips being burned up by some "battery voltage across the low voltage power supply" scenario. We have part numbers for all of the chips in theory, and as long as none of the inductors are open circuit, all the burned chips and passives could be restored, but it might take so much labor that it would not be worth doing. I think this must be the scenario that Elcon are thinking of, but we disagree that it's necessarily at that point, at least with the clues we have so far.
 
#34 ·
Thanks for catching my error. Yes the isolated side runs on 3.3 volts not 5. You may have fried the voltage regulator but I still say it's that big cap on the back side since you heard a pop. You may even have fried the on by the viper chip and if that is so you may not have 12 volts to this part of the card. Caps usually short don't they? At any rate I think it's repairable and I would give it a look if you don't think you can manage.

Where are you cause I'm in Alabama? I think Adam shipped his to me for $12 or something.

The one we soldered a new processor on and reprogrammed works like new. We put a brand new blank chip on it and then programmed the EEPROM and the flash on it.
 
#36 ·
On one of the ViPer repairs we also had to replace the diode and cap on one of the 12V outputs.

The 3kW repair that pdove mentioned involved replacing and reprogramming the microcontrollers for both the master and slave units.

We have another repair in progress that involved the FETs of the switching boost section. This one is turning into the labor sink hole that Coulomb mentioned--replaced the fets, numerous passive devices, the fet driver chips, the SMPS chip, the ViPer chip, the input relay and bypass resistors. May need to replace the PFC chip, etc... The failure took out so much that it is a tedioous process to test, find and replace all the affected components.
 
#37 · (Edited)
I thought I posted this elsewhere, but I can't find it.

When re-assembling an Elcon/TC charger after having removed the main PCB from the heatsinks, it's important to use a suitable tool to prevent this sort of problem:



Note how the nearest MOSFET has been twisted, and is likely shorting its leads together.

The best solution seems to be to use a tool to prise the really strong heatsink clamps apart when positioning the clamps on the MOSFETS (also the diodes and bridge). Petrhaps a pair of spoons would do; I found a pair of screwdrivers was not terribly good. I found this pair of pieces of aluminum worked for me:



These are actually heat-sinks for diodes from an old welder, but obviously any scraps of metal about the right size would do. You can just squeeze the pieces of metal together to prise the clamps apart, and they are (just barely) not too big to get in the way of other parts of the charger. You may have to trim your scraps to size.

[ Edit:*the later models have slightly different clips, so 3 mm aluminium is too thick. So you'll need 2 mm, and aluminium may be too weak, so steel is preferred. ]
 

Attachments

#41 ·
I've got a working 156V (13S flooded cells - 180V 11A max) 2Kw unit that I swapped the transformer and inductor from a 2Kw 120V (160V Max) unit - but have run into troubles with trimming the output voltage. (The 160V unit was unrepairable)

Due to the divider - both transformers produce 180V output.

I've tried the method here - http://www.diyelectriccar.com/forums/showthread.php/elcon-charger-voltage-adjust-51100.html BUT no luck - it either runs at 180V or doesn't detect the battery (sigh).

Is it possible to reprogram a 155V 13A lithium curve/program into it? The lead one seems strange - it is doing an absorb or equalisation routine - voltage pulses at end, not what I need (and it doesn't have the BMS control contacts either).

It works fine on a 13 series lead pack, with either transformer. Sigh.

Was a PITA to dissasemble and reassemble these things.
 
#42 · (Edited)
I've got a working 156V (13S flooded cells - 180V 11A max) 2Kw unit that I swapped the transformer and inductor from a 2Kw 120V (160V Max) unit - but have run into troubles with trimming the output voltage.
Ooh - a transformer transplant! We've thought about that, but I don't think that's been done before.

(The 160V unit was unrepairable)
Heh. Paul, are you still in "I like the challenge" mode? :eek:

Due to the divider - both transformers produce 180V output.
I guess that's what we'd expect.

I've tried the method here - http://www.diyelectriccar.com/forums/showthread.php/elcon-charger-voltage-adjust-51100.html BUT no luck - it either runs at 180V or doesn't detect the battery (sigh).

Is it possible to reprogram a 155V 13A lithium curve/program into it? The lead one seems strange - it is doing an absorb or equalisation routine - voltage pulses at end, not what I need (and it doesn't have the BMS control contacts either).
The short answer is yes, we have the technology to to this sort of thing now.

Was a PITA to disassemble and reassemble these things.
Yes, they are a bit of a pain.

Unfortunately, reprogramming them is a different sort of pain. You will need to build two simple interface circuits, and you will need a serial USB dongle and an Arduino board of pretty much any sort. We can send the software you will need, but you'll need to have a bit of a software bent to do this. Having succeeded with the transformer transplant, you certainly have a hardware bent. Are you up for this additional challenge?

I was going to document what we know on the AEVA site once it's settled down (they're going through a painful database translation process at the moment). Maybe I should do it on DIY if you are ready.
 
#45 ·
I'm ok with software, used to do a heap of pic microcontroller programming.

The 160v unit I stripped down, kept the semiconductors and magnetics but got rid of the rest unfortunately.

I'd be quite interested in the process, would save quite a bit of money instead of another $500+ for an appropriate charger.
 
#46 · (Edited)
I'd be quite interested in the process, would save quite a bit of money instead of another $500+ for an appropriate charger.
The code is Intel assembly. First you have to build a circuit using an arduino.
Next you have to write over some of your code with a routine that dumps memory. Then we have to repair the section we overwrote. We may need to read the EEPROM which requires writing some more code in to dump it. Then we modify the parameters we want compile it and write it back into your unit.

This is not a trivial exercise nor for the faint of heart.

If this is a CAN enabled charger it is easier but I was under the impression it was not.
 
Top