Hooked controller reversed - blown

[kZs0lt]

Had a melted battery connector that I was fixing.
However, after the fix, I've hooked up my 206 Infineon controller to reversed polarity.
I saw a big spark in my switch, and a short current burst heated up my connector.
Now controller won't work and CA DP displays 56A no load.

What could have been blown? How should I fix this? I have spare FETs.
Thanks for your advice...

 

[Ricky_nz]

What could have been blown? How should I fix this? I have spare FETs.
Thanks for your advice...

With the power supply reversed the FETS internal diodes provide a short across the incomming supply causing very large currents to flow. These currents could have burnt/destoryed PCB tracks or the current shunt so check for this sort of damage. You could have damaged the voltage regulator circuit and possibly even the microprocessor.
The input capacitors could be damaged but this is less likley due to the FETS shorting to protect them.

Start simple look for burnt tracks / open circuit and check the current shunt is intact.
Failed FETS should measure as a short circuit between drain and source.

 

[kZs0lt]

Thanks for your tips.
I do feel resistance when motor is rotated, just like regen, so some FETs are blown. I will try to disconnect phase by phase, so I will see on which phase are blown. I don't know what's the difference between high side and low side fets, and which are which, can they both blow at the same time?

Failed FETS should measure as a short circuit between drain and source.

I'm not sure which is drain and which is source, but I think I can figure it out.

I hope I didn't damage the voltage regulator circuit or the microprocessor.
How should I check the voltage regulator? I should see some 12v+ on the lm 317 and 5v+ on the 7805 right?

 

[Ricky_nz]

Thanks for your tips.
I do feel resistance when motor is rotated, just like regen, so some FETs are blown. I will try to disconnect phase by phase, so I will see on which phase are blown. I don't know what's the difference between high side and low side fets, and which are which, can they both blow at the same time?

Yes it does sound like at least some FETS are blown.
You could blow both.

An easy way is to Disconnect the motor and then meter from each phase wire back to battery - and battery + terminals assuming that yoy have confirmed no broken/burnt tracks.
Shorts between the phase wire and + being high side devices and to - being low side devices.
Remember in a working FET there is a diode between source and drain. (its Anode is on source and cathode on drain) whick may confuse your readings (use diode check on your meter if you have it). A good FET will read say 0.6V in one direction and open circuit in the other. if it reads 0V across the fet its dead.

The drain of the top side fets is connected to + battery and the source of the bottom side fets to -ve battery through the shunt. The source of the top side fet is connected to the drain of the bottom side fet and to the respective phase wire.


Looking at the front of the FETS labeled side with tab to top and pins to the bottom
the pins left to right are
Gate drain source

I hope I didn't damage the voltage regulator circuit or the microprocessor.
How should I check the voltage regulator? I should see some 12v+ on the lm 317 and 5v+ on the 7805 right?

I'm not sure of the voltage regulator circuit of that controller, perhaps someone else may know but if it has a 7805 then yes it should put 5V out +/- 2%. any more than 7V and you can probably consider the micro dead. The LM317 is a programmable device setup up with a pair of resistors to determine its output voltage.

 

[kZs0lt]

Thanks a lot Ricky for your detailed explanations.

Sigh, the shunt literally evaporated and all 3 low side FETs died. Could not see any trace problem(had beefed up traces). Measured resistance between source-drain and indicated 0 Ohms for the 3 low side FETs.
Replaced the 3 FETs, inserted new shunt and tested again.
Little spark, no motion. Then next to the switch I put a thin wire in case of a fault.
This time the small wire evaporated and after reinspection, I found the drag is back, and I have blown 2 low side FETs and 1 high side (one on every phase) this time. This time the polarity was OK, I have double checked. Shunt is Ok.

Now what to do next?
I still have 5 FETs, but the next mistake can let me without controller...
By the way I have a spare board, but it has only the surface mount parts, and the microcontroller, no caps, no shunt, no regulators, nothing. It would be a pain to replace all the wiring too.

Is it possible that the gate drivers or the microprocessor or the voltage regulators are damaged and that caused the second FET fault?
How could I troubleshoot the problem in a way I won't kill any more FETs?

 

[dnmun]

the input capacitors would have been reversed too, so they are shorted now, the voltage regulators may have some input protection, how black is it and where did it burn? but if it is just the FETs which may yet be ok, it could be just a matter of replacing the caps, maybe the regulators, but if the regulators won't put out 5V, test the processor with another 5V source before hassling too much with the repair. if it will function with an external source then replace the regulator and caps before attacking the mosfets. jmho

 

[amberwolf]

The 12FET currently on CrazyBike2 was reversed by it's previous owner, and it was lucky that only the input caps were blown, along with the shunt evaporated.

The most likely things to die, AFAICT, are the FETs, the caps, the shunt, and the low-voltage regulator(s). Sometimes FETs die in a way that doesn't seem apparent in testing, until a voltage is applied across the D-S junction (with or without a gate signal, depending on the damage. The caps may not appear to be bad visibly, but might still be shorted internally. Run them long enough (a few seconds to a minute) under normal voltages and they will heat up from that, potentially exploding from electrolyte boiling if they don't vent properly. They also may not fail shorted during a capacitance test, but fail under higher voltages where current isn't limited.

It's also possible but not that likely that the gate driver chips or transistors are toasted, causing shoot-thru by forcing both upper and lower FETs in a phase to be on at the same time.

Do you have an adjustable-current-limited adjustable-voltage PSU you can test with? If so, try testing at 15V with current limited to 500mA; then if nothing smokes use a voltmeter to test the voltages across the low-voltage regulator, and at each of the phase FET gate inputs.

There should be no FETs turned on because 15V ought to be well below any LVC, so any that are on are probably due to failed gate drivers, assuming that the low-voltage section is otherwise operating correctly.

If it all seems normal at that point, you could raise the voltage up a few volts at a time and retesting voltages with the meter, until either you are back at normal pack voltages or there is a problem you can find (and then fix).

 

[dnmun]

before you apply voltage again you need to figure what has shorted. if the caps are shorted, remove them and replace them with caps that will work, and have the right polarity and voltage range, they don't have to be full sized, just something so you can power it up, if the voltage regulator survived then the stuff like the gate driver on the 12V rail will be ok i expect. and the 5V rail should be ok too, along with the processor, but the caps will be shorted so do that before anything. jmho.

 

[hydro-one]

good luck its a tough call to throw fets etc at a suspect board i was lucky on the road , i used stripped battery wire-- a few strands-- to test for correct polarity. big spark bad . small spark that diminishes the more you play with it good. then i wired it up . i felt i would blow the controller if i got it wrong.,........

 

[fechter]

You might get some hints here on page 4:
http://endless-sphere.com/forums/viewtopic.php?f=16&t=8969&start=45

The power stage on the Crystalyte is very similar to most other brushless controllers. The gate driver setup might be different, but the same troubleshooting procedures would apply.

 

[kZs0lt]

Thank you guys for you responses, I didn't have time to get back to this, but will troubleshoot it again ASAP.

First I will check and replace the input caps. I know there is a big one, but there might be a little one too? Need to check.

Next I will try out if reprogramming with my USB harness works. If it works can I suppose that the microprocessor is fine?

Then I want to check the voltage regulation. Can I test it without the FET rail connected, just the ignition wire? I don't want to blow any more FETs.

I don't understand why did I blow additional FETs, after replacing the bad ones and the shunt, and connecting with correct polarity. Could it be because of bad input caps, that weren't checked and replaced?

 

[amberwolf]

Without enough capacitance across the FETs, the inductive spikes coming back from the motor each time the FETs are switched on and off can be very high--often higher than the FETs rated voltages (much higher than the battery voltage). That's one of the main things the caps are there for--to smooth out those spikes.

There are other things that they help with, too, by smoothing out input supply ripples (as large currents are drawn momentarily by the motor thru the FETs during each of the starts of a PWM pulse, for instance).

 

[fechter]

Yes, if the main input cap is blown, the FETs won't live long under load, but they shouldn't blow just from powering up.
Yes, you can test the voltage regulator and microprocessor with the FET rails disconnected.

If the gate drivers are shorted, the FETs might be on when they are supposed to be off, causing shoot through and destruction.
If you measure resistance from each FET gate to ground and to the 12v supply line, you should not see anything that looks like a dead short.

 

[kZs0lt]

I've took a look on the controller today.
It has the main cap(100v470uF) slightly puffed and the legs melted, one of them was totally disconnected.
But there is another smaller (100v100uF) after the shunt that seems OK.(reverse polarity blew the shunt).

I have separated the ignition wire and powered up only that part without the FET rails, and the voltage regulation seems fine.
Battery was at 61.xv, and the power resistors dropped only 9.3v (480 Ohms). This equates to 19mA a little low, I know that when everything is connected, motor included it should be somewhere around 60mA. But there was nothing connected and the FET rail was disabled.
I'm not sure what is the LM317T topology on this 6FET controller, it does not see the full 61v-9.3= 52v, was only about ~40v and 1.2v across the leads, but the voltage reading on 7805 was 12.18v and 5.07v respectively, so I think the voltage regulation stage should be fine.

I have tried to reprogram the microprocessor and it worked, so it should be fine.

At this point the main cap needs to be replaced.

But still don't get what happened second time. Replaced the shunt and the blown FETs.
If I remember correctly I have powered up the controller with everything hooked up (and probably blown main cap, could not see that because of lot of wires there) and throttling it did not move the motor. Probably tried this twice. If I remember correctly the wheel was still rotating without resistance by hand.
Than I introduced a thin wire at the battery hookup in case something goes wrong and next time I powered it up, blowed this little wire in a big spark, and I found the wheel has resistance again and after measuring them I have 2 low side FETs and 1 high side (one on every phase) shorted this time. The polarity was OK, I have double checked. Shunt is Ok.

I still have 5FETs, so I can replace these ones, but if I kill again more than 2, I am left without controller.
I was thinking to get some cheap FETs locally, just for testing and troubleshooting this. But could not find anything above 60v (18sLiFePo4) with reasonable current rating. What FETs would you suggest? What 75nxx FETs are the chinese controllers using?
I also have a spare 6FET board, maybe the best idea would be to just migrate everything on that...Trying to troubleshoot this could just make more damage.

If the gate drivers are shorted, the FETs might be on when they are supposed to be off, causing shoot through and destruction.
If you measure resistance from each FET gate to ground and to the 12v supply line, you should not see anything that looks like a dead short.

I will have to check this, if I don't see any short here can I assume everything is fine? Still not sure the blown cap is the only problem...

 

[fechter]

My thought was that if a driver was stuck on, it might make both high side and low side FETs on at the same time, shorting the battery and blowing the FETs. A current limited power supply might be a good idea, or a hefty light bulb in series with the battery as a current limiter.

A new part on the market is FDP036N10A. A little better than IRFB4110.

 

[dnmun]

you should be able to measure the gate voltages while the mosfets are not connected through the source drain, which is where that last blown cap goes, across the S/D busses. unless it is one of the small 16V caps that carry the charge up to the high side gate drivers. pictures would help.

so replace all the caps, check the gate voltages while rotating the wheel to see if it will switch the gate voltage on the mosfets, but i am just guessing.

 

[fechter]

If you measure across the gate to source on the FETs (two outer pins), it will be near zero when it's off and near 12v when on. The high side gate drivers usually have a bootstrap circuit, so you won't see them turn on unless the motor is moving. Spinning the motor by hand with the halls connected will make this happen. What you're looking for is one that's stuck on all the time. Normally they should turn on and off as the motor is turned. Any that stay on would indicate a blown driver.

 

[kZs0lt]

Ok, thanks for your responses, I have learned some from this mistake.
I have installed every component on the new pcb, so I am using it for now. I will come back and troubleshoot the old one, if I get some new FETs and voltage regulators.
New pcb cuts out from a stand on high current(set to 40A battery, 100A phase, and yes, this is a 6FET). The old one did not have this problem. Maybe it is because the shunt is higher, aprox 5.95mOhm, and the old was more like 4.1mOhm