Replacing MOSFETs in a Grinfineon controller.

[ions82]

I recently delivered a handcycle to a customer after adding a Bafang/Grinfineon e-drive system. It went together well and worked BEAUTIFULLY when I test-rode it. Same for when the customer took it for a brief spin. Unfortunately, some poor manufacturing/design of the handcycle itself led to a chain jam at the cassette. It stopped the wheel from turning, and the rider attempted to use the throttle to get the bike rolling again. This mishap resulted in a dead controller. Apparently, one (or more) of the MOSFETs gave up the ghost. Needless to say, the maiden test-ride was unsuccessful.

So, I now need to replace the six MOSFETs in this controller. The originals are AOT460. Would it be a good idea to replace them with IRFB3077? I am an untrained hack when it comes to electronics and repair, but I am confident I can stumble my way through this repair. None of the MOSFETs looked/smelled burned. No evidence of anything getting hot. Would different MOSFETs be any less probe to such failures? Or, is this a result of the basic controller design? If the user were to go full-throttle from a dead stop, might this controller die again? The user is paraplegic, so reliability is paramount. If this thing fails while he's out for a ride on a summer day, being stranded could result in a dangerous situation. He is in his 70s and really wants to enjoy every bit of life he has left. I'm hoping this drive system will help him be able to do that.

 

[amberwolf]

So, I now need to replace the six MOSFETs in this controller. The originals are AOT460. Would it be a good idea to replace them with IRFB3077?

are the ones you want to use the same capabilities as the originals? same d-s voltage, d-s current, rdson, gate charge (qg?), etc? if they're not, then are they higher capability, or lower? if lower, is that lesser capability detrimental to the usage scenario?

None of the MOSFETs looked/smelled burned. No evidence of anything getting hot. Would different MOSFETs be any less probe to such failures?

depends on why they failed.

if they failed because the controller asked more of them than they can handle, then to make it less prone to such failures you would need to use parts that can handle more than the controller will ever ask of them. but you first need to know what the controller will actually be able to ask of them, so you'd need to contact grin tech on that regard.

Or, is this a result of the basic controller design? If the user were to go full-throttle from a dead stop, might this controller die again?

depends on the design itself. you'd have to ask grin tech what that specific controller model is designed to do if the throttle demand is high but the motor is stalled.

The user is paraplegic, so reliability is paramount. If this thing fails while he's out for a ride on a summer day, being stranded could result in a dangerous situation. He is in his 70s and really wants to enjoy every bit of life he has left. I'm hoping this drive system will help him be able to do that.

if you really need it to be dead reliable, you would need a controller that detects a locked-rotor stall condition and ceases to try to power the motor until the condition is removed.

otherwise, even if it does not damage the controller, it could (probably will) damage the motor itself, or the phase wires between them, by overheating, since a locked-rotor stall condition means the motor is drawing maximum current continously, rather than just momentarily before it starts rotating faster and faster, which decreases the current it draws.

if you have a cycle analyst v3 on the system, you might be able to use some of it's limiting features to prevent or at least minimize this type of condition, but i am not sure which ones you'd need to use to do it. but grin tech could almost certainly tell you easily enough.

the below might not sound helpful...but it is intended to be, intended to help you see things from a different perspective, so you may make better things.

but whatever you do, then before you hand it off to the customer, if it must be totally reliable, you will have to test it as hard as possible under worse conditions than he will use it under. thus, if it does fail, you can fix it now, and determine how to prevent that failure. then retest under the worse conditions, and fix each failure that occurs, modifying the system to prevent each one in turn.

eventually, when you have bulletproofed it, it'll be totally reliable. but in my experience, it takes quite a lot of work to do that.

you must waterproof everything. you must snagproof everything. you must heatproof / coldproof everything (depends on which extreme(s) it will be used under). and so on....

but:

most importantly you must *user*proof everything. the user must be able to do literally anything "normal" (for them, *not you*!!!) and have nothing go wrong. this is very difficult, unless you work with the user, back and forth, to see exactly what they do, how they do it, etc., and then come up with ways to mitigate problems with doing stuff that way. (telling them not to do it won't work...you have to make the system unable to fail when they *do* do it).

as an example, i can cause almost any system to fail, fairly easily, because i use things in ways the system/object/etc designers could never have imagined.

(especially since most of them appear to have very limited imaginations, and expect all users to be as rigid as they are in their thinking, and to always do things the same way, just exactly as the designers intended...which of course, the users cant' do, because they ahve no idea what the designers intended, as the designers didn't tell the users, they just assumed the users would do it the same way the designers did...and when the designers find out the users didn't do that, they get all upset and blame the user for the problem instead of the design, rather than the designers' failure of imagination.... )

 

[bww129]

Apparently, one (or more) of the MOSFETs gave up the ghost...
None of the MOSFETs looked/smelled burned. No evidence of anything getting hot.

First things first - MOSFETs usually fail short circuit (all pins end up having a low resistance between each other) and often fail catastrophically when being forced to deliver high current and dissipate excess power. It would be easy to see and smell if they failed catastrophically, but not so easy if they failed without the telltale signs. Are you sure the MOSFETs have actually failed? Check them with a multimeter. Once you diagnose that then you can think about replacing them.

 

[AHicks]

When you figure out what happened and get it going again, with the history of one chain jamb disaster already under your belt, wondering if a circuit breaker in the circuit might/should be part of the recovery plan?

 

[fechter]

The originals are AOT460. Would it be a good idea to replace them with IRFB3077?

I would. The 3077 has about 1/2 the Rds and a higher voltage rating. There are even better ones out there if you don't mind spending more money.

 

[john61ct]

And assuming a capable user, a bit of "how to refrain from breaking your bike" training should go a fair way?

 

[ions82]

And assuming a capable user, a bit of "how to refrain from breaking your bike" training should go a fair way?

Indeed. I don't know how he ended up stopped while in the highest gear. I think going for the throttle was more out of frustration than anything else. The build for this bike has been frustrating for all involved. Handcycles aren't very well designed, so there are always at least TWO cans of worms that must be opened when trying to customize them.