KT controller defective, but why?

Hi guys, I believe that my KT controller gave up the ghost today, but I do not know why or what exactly happened. Maybe you can help.
First, the details: I have a 20" bike, running on 16s A123 (~50V), with a Heinzmann motor (rather large direct drive).
Gallery:
https://imgur.com/a/tTuFe
I have this controller:
http://www.groetech.de/index.php?main_page=product_info&products_id=849
With an LCD5 display. I modified the controller by installing IRFB4110 Fets and 100V low-ESR brand-name caps.
I programmed the maximum current to 5A. The system has worked perfectly since april, for about 1.900km. 2 or 3 weeks ago, I enabled regen braking at the highest setting. This also worked well until today.
At first, the speed displayed at the LCD froze at ~11kph, and the batter charge indicator graph at the top was blank. Everything else still worked. I could choose different power stages, and the total voltage was also displayed correctly at the bottom.
The motor worked, and the speed limit as well. Power cycling both the LCD and the controller had no offect. After power cycling, the speed display was stuck at 0kph. About 1km farther I tried another power cycle on the controller, but this time, the motor did not work any more. Another power cycle on the LCD, and the motor jerked once slightly, but still didn't work afterwards. I wanted to cycle home (thankfully not that far) but noticed that the motor was difficult to spin. I remembered that this happens when the phases are shorted. So I unplugged one of the three phase cable connectors, and now the motor spun freely.

I am totally at a loss why these problems happened. The phase cable thing at last points to shorted phases, this can only happen via the FETs. But this seems totally unrelated to the malfunctioning LCD. Can a general malfunction somehow destroy the FETs? Can the regen braking have something to do with it? Is this function known to fry controllers? I wonder why it worked flawless at daily use during all summer, and just 2-3 weeks after enabling regen things fail...

Thanks for your help in advance,
Julian

Sometimes these types of chains of failures, progressing from one to another, are caused by a wiring short inside the harness from motor to controller. Happens more often with wires that come out hte ends of hubmotor axles, because they get damaged from bending against the sharp edges of the axle, or pulled, and cut into the insulation, and can allow phases to short to halls.

So if that's what's going on here, what may have happened first is that the hall sginal (for the speedo) wire broke, or it's insulation failed enough to let it ground against the axle itself or one of the phase wires. That would kill the speedo detection, and potentially cause damage to the controller's MCU or other electronics, if the phase voltages shorted to it fed into those, and wasn't any protection against it (usually there is not).

Then the MCU might cause wrong signals to the FETs, and blow phases (or even just trigger the wrong ones, causing "locked up" wheel symptoms, and prevent the motor from turning..

The original cause coudl be unrelated to teh regen, or it could be that regen has rocked the axle back and forth enough to either wear thru the wrie insulation if it's tight against he axle end, or it could have acutally caused axle spinout, and the wires are now all twisted up.

Hi Amberwolf, thanks for your quick answer.
Thankfully, no wiring issues are possible the way my harness is designed. The motor has bearings with a very large inner race, so the aluminium spacer between the axle and the inner race has 4 separate holes for each of the individual phase wires and the 5-pole sensor cable.

The phase wires and the hall cable is then routed to the controller not only individually, but also individually shielded. So a cable-related malfunctin seemed unlikely.
Also, the connection of the axle and the fork is bullet-proof torque-wise.




But I then found a thread with a similar issue, and similar to this one you were also the first to reply (thanks for being so helpful generally ):
https://endless-sphere.com/forums/viewtopic.php?t=91386
This guy also blew his fets after switching regen braking on, after they worked well a long time without regen. So I measured all phases against the positive and negative connection of my controller, and the green phase gave different readings. I disassembled the controller and put in 2 new Fets to be safe, and lo and behold, it works flawless again! I also set regen to 0 again as a precaution. Now I can hopefully put a lot of flawless miles on my bike again.

Cheers,
Julian

Curious. Couldn't get much diagnostic/situational info from the poster in the other thread, so don't know what actually happened to cause the issue there.


Was your battery fully charged, or nearly so, when using regen at any time? And does it have a BMS with a single port that shuts off charge / discharge when full / empty, or upon overcurrent-during-charge conditions?

If so, there's a chance (though a small one) that the regen topped off the battery, then the BMS shut off it's input / output to protect it, which disconnected the controller from the battery. The controller would not shutdown in this case, because it's regenerating power, so it's own output voltage would keep it running until the regen ended.

But at the moment that the battery disconnects, the voltage inside the controller would spike, because most of the current now has nowhere to go. The spike could be high enoguh to damage electronics in the controller, like FETs, and cause wonky behavior permanently or temporarily in the low voltage stuff, depending on how well the LVPS filters stuff like that out.

If the controller has an HVC it'll also shutdown during that spike if it goes that high, but if not it'd just keep running till regen ends.

At that point, the BMS might turn back on, or it might take a moment afterward (or even a while, if it has to wait for a cell to drain down for balancing).


Anyway, the upshot of all that is that it's one possible mechanism for regen to damage a controller; and the only one I can think of for your situation.


THat is, unless the software in the controller is just poorly designed and alllowed some form of current path in the FETs that damaged something. (which might explain the other thread as well, same brand controller; if it had the same firmware revision and the same bug, if it exists). It's not a very likely problem.

Hi. At the time the controller failed completely, the battery was about half empty. I routinely use the brake the first time after charging around 150m from home when I cross an intersection. Then the battery would be almost full, but full in my case means that the A123 cells are only charged to around 3.48V/cell instead of 3.60V. I figured that this might add a bit of cycle life, and it costs about 0.5% capacity. My BMS would shut the pack off at around 3.65V/cell, so this means that at any point there should be ample of capacity left to take regen current.
To balance my pack, I crank the charge voltage up to the balancing treshold of 3.6V/cell of my BMS about once a month.
Also my route does not have any hills. My city is famous for rain and being flat. :wink:

Hm. Strange, all this. And then there's the issue that the replacement Fets look quite different from the one I soldered in originally:
https://imgur.com/a/Ygwip9Y
I'm not quite sure what to make of that yet...
Edit: Now I do...
https://endless-sphere.com/forums/viewtopic.php?f=2&t=10655
Edit again: I have now replaced all Fets with the genuine ones from the same vendor (Reichelt) I got my repair replacements from. I do not notice any differences whatsoever, but It feels better though. :wink: