Hello Franc,

i saw you have PM disabled, so writing my question here:
i was repairing the controller of an e-scooter. One of the half-bridges were shorted, these were already replaced by me a few years ago. The damage occurred at a power-on event, not during heavy load. This is already strange but the more interesting problem came after i replaced the mosfets:
The original part is TK100E10NE, 100V/100A thing. The previous replacement part was IRFB4110, it served quite some years.
now i replaced the faulty ones to YJP70G10A. Tested the thing from a lab supply with 2A current limit. It was ok for accelerating the motor, current was about 0.3-0.5A
The problem came when i applied the electric brake. The bridge was either completely shorted or the top side FET was shorted, sometimes only its gate-source path was in short.
I did this several times, trying to figure out what the basic issue was. In some cases i had to replace the mosfet driver coz its high-side part went shorted.
High-side FET went shorted even if i just applied braking and turning the wheel by hand, ie very low currents flowing in reverse direction.
Then i got enough and tried a different FET: STP110N8F6, which is rated to 80V only but the scooter runs from a 10cell pack, ie 42V max. Even if electric braking generates higher pulses, this should be ok, surely for a test.
Guess what? These work fine. I was trying to spot what the problem with YJP70G10A could be, first i thought it is not designed to stand large amounts of reverse current (top side fet channel diode is used for charging the battery while braking). But the datasheet says 70A for this parameter, so i am pretty much out of ideas.
Do you have any idea what it could be, what i am overlooking?
thanks,

pepe

_________________
Adatmentés - Data recovery

I have no idea. You might like to ask here:

https://www.eevblog.com/forum/repair/
https://www.badcaps.net/homepage-community
https://www.elektroda.pl/rtvforum/forums.html

MOSFETs can be damaged by inductive kickback or improper drive, or shorted turns in inductive loads.

_________________
A backup a day keeps DR away.

Hello,

thanks, at least it is not something trivial i overlooked.
improper drive and shorted turns can be pretty much ruled out coz with the STP110N8F6 it seem to work fine.
I will try to find time to ask on those forums, thx

pepe

_________________
Adatmentés - Data recovery

Be aware that eBay and AliExpress are full of fakes. Try to buy from reputable sources like Digikey, Mouser, Farnell, RS Components, LCSC (China).

The problem you are experiencing is very difficult to solve. I see a lot of people struggling with MOSFET reliability issues when dealing with inductive loads (motors, transformers, solenoids). A momentary high voltage spike will instantly kill a MOSFET, but you won't see it. Check the snubber components, usually a reverse biased diode.

https://lcsc.com/product-detail/MOSFETs_TOSHIBA-TK100E10NE-S1Q-S_C396034.html?s_z=n_TK100E10NE
https://au.element14.com/Search?st=TK100E10NE (Farnell)

_________________
A backup a day keeps DR away.

This discussion might help:

https://electronics.stackexchange.com/questions/80137/how-diodes-protect-h-bridge-dc-motor-driver

_________________
A backup a day keeps DR away.

I got them from a local store, but admittedly one of the cheap ones. There is no snubber diode anywhere, the spikes are drained towards the battery thru the high side FET channel diode imo. Since i tested it from a lab supply, it did not limit the supply voltage but there are some caps (4x470uF) able to absorb them on board. The supply voltage slightly elevated when i turned the wheel by hand while braking (like 30->38V max) and soon the top side FET was in short. Also tried with battery connected, in that case both transistors died, probably high side first, then the low side due to large current.
It is still odd that other types of FETs work fine.
I know there are fakes on Ali, some months ago i bought some ML4824IS1 (no longer available at mouser, etc), then it caused me quite some headache when the lamp power supply worked for an hour or so, then the power stage blew up. Replaced it a few times until i found out that the controller is actually ML4824IS2 (marked as IS1). So it was driving the output stage at twice the freq it should have had... Thanks, Aliexpress

pepe

_________________
Adatmentés - Data recovery

for mosfets its better to use any test ckt and plot curves....if they match with datasheet then they are probably original and good.

Did you try hooking up a dso directly onto the line that gets shorted while moving things (break or handle as how you do)

Try also hooking it up without the fets and see if you are getting any impulses..

may be some other snubbers elsewhere are not working well and things are leaking back

_________________
THERE IS ALWAYS HOPE TILL YOU LOSE IT.

hello,

i am over this project with using different fet. it is very hard to measure such momentary glitches, not to mention how complicated it is to set up for such measurement... not worth the efforts. The control circuit is ok, the thing is working just fine with the STP fets, i'm not gonna invest more time into it. I was just curious if i was dumb or overlooking something trivial.

thanks,
pepe

_________________
Adatmentés - Data recovery

Most MOSFETs I see today are widely used on laptop motherboards where they are used to protect the CPU and GPU when a diode or cap fails.

8-pin MOSFET chips are inexpensive and better repair shops usually have lots on hand.

_________________
USB 3.5" enclosures are $19 USD postage paid from China, I own several

oh, man(?), why do you have to comment everything...

_________________
Adatmentés - Data recovery

engagement farming

Hello Franc,

a question again
This time it is related to ECL logic, which i am not quite familiar with. In fact, i have a circuit where the -5V supply line resistance to gnd is 27ohm. I identified that this rail feeds some ECL to TTL interfaces (MC10124 and 10125), but not sure if this is normal or not for such circuit. It is pretty hard to identify where else this rail is connected coz it beeps just as for GND connections, only distingishable by numerical reading of the multimeter.
I don't think this low res is normal, but can't say that for sure in this situation.
I appreciate any insight.

pepe

_________________
Adatmentés - Data recovery

I saw a small amount of ECL stuff 40 years ago. I don't think I ever encountered a faulty IC. ISTR that these were used in minicomputer graphics cards, or maybe a Control Data hard disk.

I would guess that your diagnosis is correct, but I have no experience to back me up. Sorry.

_________________
A backup a day keeps DR away.

ok, thanks. gonna have to dive a bit deeper then :s

_________________
Adatmentés - Data recovery

The datasheet specifies a maximum Negative Power Supply Drain Current of 72mA, and 25mA for the positive supply.

https://www.datasheetarchive.com/datasheet/MC10124/Motorola?term=MC10124&version=4

5 volts / 27 ohms = 185 mA

_________________
A backup a day keeps DR away.

yes, i was looking at that, but since not only a single IC is hanging on that rail, there could be a reason for such low res in theory.

_________________
Adatmentés - Data recovery