Need help with this blown KT controller

kdog

10 kW
Joined
Feb 2, 2014
Messages
715
Location
hobart, tasmania
Hi all
A friend has asked for some help with his ebike (cause I put it together for him) it's a 500w DD hubby with a KT controller and LCD display. I built the battery for him which is a 13s6p.
All pretty standard, it's been going well for 3500 k's but he's had a series of failures in a week (which I'm suspicious are related).
I got it going for him and he took it out for a ride, started off ok, then he did some heavy regen braking down our hill. Then no power and cogging. Damn it!
Motor's fine no shorts.. But all phases cog when connected to controller
Taking the controller out it seems that all phases are shorted. I'm familiar ( but lack indepth knowledge )with testing fets.
Testing with my DMM I got this:
Battery + to all phases 0.00ohm
Battery - to all phases Aprox .45v drop
so am I correct in thinking...
That all fets are shorted but have intact body diodes, seems unusual to do ALL fets. The fets all look fine, there is no visual damage to the controller at all.
Is there some other process that could produce this result?
The controller still works in all other respects.
Cheers
 
When you say the
kdog said:
The controller still works in all other respects.
exactly what do you mean?

Do you mean it still drives the motor normally? If so, the controller is not damaged (at least, not FETs/gates/etc) but could be receiving a wrong signal from outside controls (such that when throttle is not applied, brake is enabled, like slip regen).

If it does not drive the motor anymore, then usually if it cogs when connected to controller, but not when it doesn't, then at least the FETs are blown in the controller, and possibly the gate drivers or other associated parts.


Otherwise, if it does not drive the motor anymore, but FETs/etc are not damaged, then an ebrake/regen signal is probably still being enabled on it, braking the motor.


I'm not sure I understand your getting a voltage reading during the FET testing. What setting were you using?

If volts, the controller must be powered on (and to get any voltage you must also engage the throttle and if it is not sensorless must also have correct hall signals coming into the controller from the motor).

If ohms, the controller must be powered off.
 
kdog said:
That all fets are shorted but have intact body diodes, seems unusual to do ALL fets. The fets all look fine, there is no visual damage to the controller at all.

The controller still works in all other respects.
I am being developing our OpenSource firmware for BMSBattery S / Kunteng controllers. With the development and test I did burn already some mosfets on S06S (6 mosfets controller) and S12S (12 mosfets controller) and did the replacement of mosfets.

If helps you, grab here the schematic of the controller: https://opensourceebikefirmware.bitbucket.io/Motor_controllers--BMSBattery_S_series--BMSBattery_S06S.html

Yes, I had some shorted fets and the controller would work like sending the information to the LCD but would fail as soon I increased throttle (down mosfets negative were shorted).

Sometimes more than 1 fet shorts, maybe you are in a situation were all fets from the 3 phases did short...
 
If the FET measures like a diode then it's not shorted. Your measurement of the high side looks like they're shorted.
 
I'll try to elaborate/ clarify ( but bear with me- it's not my strong point)
the controller seems to be shorted in all phases- there is cogging in all combos of phase wires. There is no ebrake signal on LCD ( and this does work when I engage brake) an the controller communicates with the LCD (responds to throttle/pas ebrakes on the screen but comes up with the error code for internal short)
There is no drive of the motor.
I think I did half the reading on diode and half on Ohms function cause I got distracted for a while between them. But redoing it in diode mode I get
0v btwn pos and all phases (shorted?-should be open?)
~.45v btwn neg and all phases.
I guess I'm looking for confirmation that all the fets are shorted. It seems unusual to have all fets blow simultaneously ( and that I'm not missing something)
Casainho- thanks for that info- It seems similar to your failure.
 
Zero ohm reading between a phase wire and battery wire indicates a shorted FET. When measuring, reverse the probes to measure both ways.
A good FET will read like a diode. Blown ones almost always short.
 
I've got 15 fets on order.. Should be here in a day or two( 3spare). Hopefully that solves it, but I'm not overly hopeful, there's always something more. :(
Is there a way to test the gate drivers? or do I just try the new fets and see.
 
kdog said:
I've got 15 fets on order.. Should be here in a day or two( 3spare). Hopefully that solves it, but I'm not overly hopeful, there's always something more. :(
Is there a way to test the gate drivers? or do I just try the new fets and see.

Once you remove the old FETs, you can try measuring resistance from battery neg. to each of the gate leg traces. None of them should look shorted to ground. Also measure the resistance of the gate resistors. Beyond that, it's hard to test unless you have an oscilloscope and a way to simulate the hall signals.
 
kdog said:
I've got 15 fets on order.. Should be here in a day or two( 3spare). Hopefully that solves it, but I'm not overly hopeful, there's always something more. :(
Is there a way to test the gate drivers? or do I just try the new fets and see.
Test by comparition. At least you can use the new one as comparition, on the signals. Look at the schematic and measure the signal output on microcontroller that goes to each driver. Than measure the driver output/gate of the power mosfet.
 
Ok my replacement mosfets arrived so I got started on the repair. I broke out the old ones and desoldered the legs one by one, then I drilled out the solder in the hole. Is there a better way to do this? ..It worked fine though.
Annoyingly just one mosfet in each phase was blown, but I couldn't know this until they were separated.
I'm suspecting a high voltage event during regen. We did just turn the regen level up to full just prior to them blowing, so I'm going adj that back down.
Gate resisters all measure 51ohm ( as their marking suggests they should)
Gate to neg readings are all 27Mohm
So hopefully this will work.
 
I use an industrial strength solder sucker, but for home use, doing the legs one at a time is a good method. I wouldn't drill out the holes, as they might be plated through. Solder wick or a wimpy solder sucker will clear the holes. Just make sure you get good solder flow on both sides of the board to reconnect things.

Even if only one FET blows, its good practice to replace all of them from a matched set.

Gate drivers sound hopeful.
 
As I said, just make sure the solder flows on both sides and it will be fine. Often times the through hole plating rips out when you remove the old FET anyway.
 
Yeha! She lives again... Hopefully for more than the first ride but so far so good.
Soldering all those little mosfet legs in is easier said than done... With all those smd parts right next door just waiting to float off or bridge to... Damn fiddly.
Thanks for the help everyone.
 
fechter said:
Solder wick or a wimpy solder sucker will clear the holes.

I just bought 3 solder suckers from Gearbest, after going through two rolls of braid.

Are the cheap china solder suckers reusable?
 
What do you mean by "reusable"?

My solder suckers just have a cap (white plastic, probably teflon, on the cheap one, and aluminum with teflon tip on the decades-old "better" one) that comes off (threaded on aluminum, friction-fit on teh plastic one). Once the tip/cap is off, just tap it on a surface to empty it of solder bits.
 
Time to revisit this thread :(
Same mate.... Blown another KT controller in much the same circumstance. This is the 3rd.
I'm hoping someone who knows these controllers well can give me some insight.
The details.
500w generic DD rear motor.
40amp KT controller with the lcd3 display
13s lion battery with a 60a bms.
Regen on and set to ~70% (to hopefully prevent big Regen spikes)with ebrake levers
What happened:
I had just replaced his bms cause the old one was shit, and did an off bike discharge test- all good bsttery side.
I gave the bike back to him, he charged it fully and left for work in the morning. He lives on top of a hill so he has to brake(regen kicks in) for about 700mts before he can use any charge. He made it about 300mts before it growled and stopped.
On inspection looks like he's blown a bunch of fets.
Same thing as last time.
Same distance down the road.
He's getting pissed at me for repeated failures.
All the other KT systems I've build don't use Regen and I've found them to be really reliable.
Is KT known for poor ability to regulate Regen current and voltages?
I'm thinking of ditching this system for something else but that's more $$ that I feel he shouldn't have to pay for.
Some days I just want to give up on this EV thing cause the continual failures do my head in (not just cheap Chinese stuff either)
Anyone got any ideas :?
 
kdog said:
Time to revisit this thread :(
Same mate.... Blown another KT controller in much the same circumstance. This is the 3rd.
I'm hoping someone who knows these controllers well can give me some insight.
The details.
500w generic DD rear motor.
40amp KT controller with the lcd3 display
13s lion battery with a 60a bms.
Regen on and set to ~70% (to hopefully prevent big Regen spikes)with ebrake levers
What happened:
I had just replaced his bms cause the old one was shit, and did an off bike discharge test- all good bsttery side.
I gave the bike back to him, he charged it fully and left for work in the morning. He lives on top of a hill so he has to brake(regen kicks in) for about 700mts before he can use any charge. He made it about 300mts before it growled and stopped.
On inspection looks like he's blown a bunch of fets.
Same thing as last time.
Same distance down the road.
He's getting pissed at me for repeated failures.
All the other KT systems I've build don't use Regen and I've found them to be really reliable.
Is KT known for poor ability to regulate Regen current and voltages?
I'm thinking of ditching this system for something else but that's more $$ that I feel he shouldn't have to pay for.
Some days I just want to give up on this EV thing cause the continual failures do my head in (not just cheap Chinese stuff either)
Anyone got any ideas :?
So that seems specific about regen. Try ask on the OpenSource firmware thread to KT, the developers there are very experienced and maybe they can give a clue about the regen. Maybe the OpenSource firmware is better on the regen??
 
Hi cassainho
I didn't realise kt people were involved in that thread- is that the one here on ES?
I found amberwolfs explanation for a potential method of destruction on another thread here. I have feeling some thing like that might be right (V spike from bms shutting off).
Re programming is beyond me at the moment so I'll have to look for some other solution.
Argghh! All this work....
 
kdog said:
I found amberwolfs explanation for a potential method of destruction on another thread here. I have feeling some thing like that might be right (V spike from bms shutting off).
I was about to suggest that here... it's the only thing I can think of that makes sense for the repeated problems.
 
It would be disappointing if this was the case. I would have hoped that the controller would have adequate protection... Other controllers I've had doing exactly the same thing have been fine. I might have to get a better controller. Regen is a good thing around here with the hilly are we live in.
 
kdog said:
I had just replaced his bms cause the old one was shit

This might be the problem. If the BMS cuts the connection to the controller due to overvoltage protection during regen, the voltage at the mosfets will increase dramatically.

I recomment to try the open source firmware, if you use sensored motors. Xnyle has added an overvoltage protection a few weeks ago.

regards
stancecoke
 
stancecoke said:
kdog said:
I had just replaced his bms cause the old one was shit
This might be the problem. If the BMS cuts the connection to the controller due to overvoltage protection during regen, the voltage at the mosfets will increase dramatically.
This makes sense to me!! I could not think of this possibility.

As the battery starts charged at the home of the user and then there is the hill, probably the BMS can cut due to the overvoltage and yes, there should happen the very high voltage at motors terminals when charging current is simple interrupted.
 
casainho said:
stancecoke said:
kdog said:
I had just replaced his bms cause the old one was shit
This might be the problem. If the BMS cuts the connection to the controller due to overvoltage protection during regen, the voltage at the mosfets will increase dramatically.
This makes sense to me!! I could not think of this possibility.

As the battery starts charged at the home of the user and then there is the hill, probably the BMS can cut due to the overvoltage and yes, there should happen the very high voltage at motors terminals when charging current is simple interrupted.
I think the best option is do as my Smart eletric car does, it disables regen when battery is charged. So, I think the firmware could have configurable battery overvoltage limit that is lower than the one on the BMS.
 
Back
Top