Trouble shooting a broken bike

[Jeremy]

I think you're right, going slowly up a hill was almost certainly the cause. The slow rpm, part-throttle, relatively high torque load condition is the one where the phase current (and hence the FET current) will be at least two or three times higher than the battery current, may be more. As the vast majority of cheap controllers use rather poor FETs, with a high on resistance (and very often they are cheap counterfeit copies anyway) then this is the very condition that might well cause FET failure. I've never seen a FET failure from a loose connection, especially with the controller you have, which has pretty good current spike protection (as long as you don't mess with the shunt to modify the current limit).

If you buy a Xiechang (i.e. Lyen, e-crazyman, cell_man, etc) controller then go for the 500W one. If you want a very cool-running controller then go for one of those that cell_man or Ed Lyen sell with genuine IR low on resistance FETs.

Even a tiny 6 FET Xiechang controller, fitted with genuine IRFB3077 FETs, will happily run at 30A and 48V and only get very slightly warm after a long uphill climb at full throttle (30A at 48V is 1440 W). The snag is that these genuinely good FETs are expensive, so this pushes the price up (I recently bought some 3077s in bulk, and even then they cost me around £2 each).

 

[NRG]

This is exactly the same as what happened to my original controller on my Alien GSII, climbing a long hill on part throttle places the FETs under greater stress and one popped. I replaced the controller with one from keywin as its programable and I also upgraded the FETs and caps. An expensive thing to do but the controller has been rock solid since...used these, had two controllers to do:

 

[Jeremy]

Those IRFB3077s are probably the best FET going for a sub-60V controller, IMHO. I have three controllers with those FETs are they all run cool, even the little 6 FET that I've had running at 28A max for the past couple of years.

newguy, if you want I could replace the FETs in your existing controller with 3077s. Mind you, the FETs (even at the price I bought them for) would cost you about £20 (they retail for well over £3 each plus VAT, I think), and then there would be postage on top.

 

[NRG]

Blimey! £3.53 now from Farnell.

 

[Jeremy]

Blimey! £3.53 now from Farnell.

And a good clue as to why the Chinese make so many counterfeit FETs! I've got samples of three different IRFB4110 counterfeits here, ranging from the obvious fake (the writing rubs off) to ones that look pretty close to genuine IR ones, except for a dodgy date code (and performance that's nowhere near as good).

 

[Deleted member 4366]

Back to the reason for the blown FET. I just can't see that going slowly up a hill at part throttle would cause that, otherwise we'd all be blowing FETs on a daily basis. If it were a very steep hill at full throttle with the motor nearly stalled, there could be a small chance. I still think that the cause lies elsewhere. These motors and controllers are very robust. I took a KU93 up to 30amps by soldering the shunt and gave it loads of abuse, and often went slowly up some very steep hills. Never a blown FET. Also these controllers have temperature protection via an internal NTC.

 

[NRG]

Could be the FET was substandard (OEM controllers use very cheap parts from dubious sources) or perhaps it had some sort of internal damage that was exasperated by the thermal stress of running 2.5x battery current in PWM mode. The heatsink interface could have been poor allowing the internal junction temperature to rise and trigger the failure...an intermittent connection is an unlikely cause of failure but you never know...

Edit. The NTC has no control over FET internal junction temperature. That can only be controlled by passive or active heatsinking. By the time the NTC is triggered the FET junction temperature will be a whole magnitude greater than the trigger point. If the part is substandard or at the lower end of manufacturing tolerance it could fail under stress or fail if the heatsinking was inadequate.

 

[Jeremy]

Back to the reason for the blown FET. I just can't see that going slowly up a hill at part throttle would cause that, otherwise we'd all be blowing FETs on a daily basis. If it were a very steep hill at full throttle with the motor nearly stalled, there could be a small chance. I still think that the cause lies elsewhere. These motors and controllers are very robust. I took a KU93 up to 30amps by soldering the shunt and gave it loads of abuse, and often went slowly up some very steep hills. Never a blown FET. Also these controllers have temperature protection via an internal NTC.

OK, first an NTC thermistor is a very, very slow acting protection system, it only works once the heat has got from the FET die, through the FET case and heated up the spreader bar (assuming that's where it's located). That takes minutes.

The primary protection from over-current events is the battery current limit, but this does not (and cannot) limit the FET current. The reason is down to the current multiplication that occurs when the motor back EMF (and hence motor voltage from the controller as demanded by the throttle) is low and the current demand (from the torque loading) is high. Under these conditions you can easily get two to three times the battery current flowing in the FETs, as short duration pulses at around 15 to 16kHz.

The short duration of these high current pulses causes the FET die and bond wires to locally overheat, particularly if the FET has a poor internal thermal conduction path to the case tab, as many cheaper FETs do. No amount of case cooling can prevent the FET getting too hot internally under some conditions.

If the controller has sufficient margins designed in to work with a particular motor/controller/battery pack/load combination, then the FETs will take this in their stride.

Problems arise when a controller is built right down to a price and then run under conditions where it's close to failure. The 9 FET controllers are slightly worse in this regard than the 12 FET ones, as they only use a single FET on one side FET and two FETs on the other side. This arrangement is economical to manufacture (it saves three FETs per controller) and works OK, but it does mean that the single high side commutation FET has to work hard and carries all the phase current. Once the high side FET fails it invariable blows the two low side FETs before the controller current spike detector kicks in to prevent further damage.

 

[Deleted member 4366]

There's a finite probability of anything happening. Some people win the lottery. Others get struck by lightning. After spending most of my career finding and fixing problems in manufacturing, I found that causes were nearly always simple things. People often used to use these complex technical theories to mask the obvious.

The reason I added the bit about the NTCs. Iran two KU63s so hot that the NTCs cut the power, but no FETs blew.

OK, first an NTC thermistor is a very, very slow acting protection system, it only works once the heat has got from the FET die, through the FET case and heated up the spreader bar (assuming that's where it's located). That takes minutes.

The primary protection from over-current events is the battery current limit, but this does not (and cannot) limit the FET current. The reason is down to the current multiplication that occurs when the motor back EMF (and hence motor voltage from the controller as demanded by the throttle) is low and the current demand (from the torque loading) is high. Under these conditions you can easily get two to three times the battery current flowing in the FETs, as short duration pulses at around 15 to 16kHz.

The short duration of these high current pulses causes the FET die and bond wires to locally overheat, particularly if the FET has a poor internal thermal conduction path to the case tab, as many cheaper FETs do. No amount of case cooling can prevent the FET getting too hot internally under some conditions.

If the controller has sufficient margins designed in to work with a particular motor/controller/battery pack/load combination, then the FETs will take this in their stride.

Problems arise when a controller is built right down to a price and then run under conditions where it's close to failure. The 9 FET controllers are slightly worse in this regard than the 12 FET ones, as they only use a single FET on one side FET and two FETs on the other side. This arrangement is economical to manufacture (it saves three FETs per controller) and works OK, but it does mean that the single high side commutation FET has to work hard and carries all the phase current. Once the high side FET fails it invariable blows the two low side FETs before the controller current spike detector kicks in to prevent further damage.

 

[Jeremy]

There's no attempt by me to use "complex technical theories" to mask anything at all here. You're the engineer, not me, and so should be very familiar with FMECA and similar methods.

All I've passed on here is hard won information from having blown stuff up, spent a lot time finding out exactly why it failed and then learning from it.................

 

[NRG]

Its also not the first time a post has been made on the forum and the fault traced to a FET failure, mine, this one and cwah (Lyen controller) off the top of my head, there may be many more unknown about where the controller has simply been replaced with a new one...who know!?

 

[Jeremy]

Off the top of my head I think I've probably repaired around a dozen controllers with blown FETs over the past couple of years. All were on "ordinary", rather than tweaked ebikes (none belonged to me!).

Each time I tried my best to find the exact cause of the failure, really for my own peace of mind before handing the bike back to it's owner after a repair. Only one FET failure was caused by an external problem, and that was a motor that had spun and torn the wires off, shorting the phase wires together. All the others were just failures that happened whilst the bike was being ridden, and pretty much all the failures occurred at low speed and high load (pulling off from a standstill more often than not).

 

[newguy]

Have a new controller on the way from Spencer at electric-hybrid-bikes.com, new controller for £24 including delivery. His ebay reviews are great and at that price not sure I've much to lose. We can settle the debate once it arrives thanks for the help

 

[Jeremy]

Sounds good to me, and far easier than getting a new one from China. The price is pretty good too, certainly nice to see a UK supplier that's selling parts at reasonable prices like this.

 

[newguy]

Completely agree great to see a uk supplier, if i knew about the site a couple of months back i would have bought one of his bikes, great prices

 

[Gdread]

Completely agree great to see a uk supplier, if i knew about the site a couple of months back i would have bought one of his bikes, great prices

That's exactly how I felt, I bought a kit from Greenbikekit didn't understand about the little red wire issue and wasn't sure that the controller with the kit was ok so got one from Spencer along with some other bits and it was all sorted.

 

[newguy]

To wrap this up....

I installed the new controller and it worked first go, and as a precaution I rewired most of the bike just encase it was caused by a bad connection somewhere. Now I just need to wait for this snow to melt

 

[jackhandy]

It's a shame this thread's gone - The title's made me smile oftentimes....

Is he asking for advice on putting a troublesome bike out of its misery?..

Or bewailing the fact that he Just Can't Do It?

 

[Old_Dave]

I was going to recommend a shotgun... That should do it

Pretty hard to miss


Sent from my iPhone using Tapatalk

 

[Gdread]

To wrap this up....

I installed the new controller and it worked first go, and as a precaution I rewired most of the bike just encase it was caused by a bad connection somewhere. Now I just need to wait for this snow to melt

Nice one, don't forget to bridge the three speed connection if need be - orange and black - if you aren't using a switch on the connector. It was a tip posted dv8eh, wheel spins faster

And snow adds to the fun! 2wd with a front hub and pedalled rear drive