Infineon Controller 35A Dead!

[ic3wall]

Hi everyone,

I bought a kit from ebikes.ca last Christmas, used it maybe 10 times and now controller seems dead!?

Here's what happened, at the end of a ride 2-3 weeks ago, my battery was quite low 40V (out of 48V full charge) and was pulling around 3-4A max at full throttle. I arrived home and decided to just keep the throtle on until the battery dies, suddenly I heard a small ding! Motor stopped turning, so I thought that's it the battery is dead. I put it on charge and then plugged back everything, cycle analyst comes on, but when I turn the throttle, I hear a click but there is no motion at all and cycle analyst says 0A. I noticed as well important rolling resistance. When I unplug the controller, resistance disappear.

I wrote to ebikes.ca to have some advice and they suggested the following tests, which I did:

-veryify all connectors (done, everything's fine)
-measure voltage of hall sensors (It varies between 0-5V while turning the wheel, which is normal)
-Finally, measure continuity between each phase and ground, and each phase and +. I found a short between + and phase A !

Apparently, that means a blown mosfet. I found that blowing mosfets uccurs when controllers are used over nominal voltage/current, and this is not my case. (I have a 40A fuse on the battery)

I'm waiting for an answer from ebikes.ca about their warranty, but meanwhile, what can I do ?

How do you find the blown mosfet, I didn't see any smoke/fire ??? Everything in the controller seems perfect.

[cor]

You already have located the blown MOSFETs: between + and A-phase!
I have a small (6-FET) controller on a standard Chinese E-bike.
The first time I ran it, within 100 meters it stopped and the battery self-protected (switched off).
I checked and it had a similar problem - one shorted FET between the battery DC power and one phase.
So, I looked at the number on the FET and ordered new ones (on Ebay).
When they arrived, I un-soldered the bad (shorted) FET and put a good FET in its place,
then re-assembled the controller while making sure that transistor heatsinking was as good as possible
(I added a very thin thermal pad to guarantee good contact with the heatsink).
Bike is running fine now.
I am guessing that the reason that my controller blew was that one phase wire did not make good contact.
When it loosened, it created a spark and blew the FET.
I was lucky that the battery has an over-current protection, so it did not burn up the controller.
Cor.

[ic3wall]

The connections of the phase wires seem good, but I could indeed order new mosfets, which would be a low cost solution. But I'm afraid I blow them again ..

I started looking for a new controller, would the ones sold by method are plug and play on a 9c motor (w/ hall sensors), twist throttle and Cycle Analyst ?

This looks like a good deal if it fits my motor: http://www.methtek.com/2011/11/12/72v-sensorless-30/

Thanks for your help!

[dnmun]

did you open up your controller yet? if you can read the label on the mosfets, then someone may have a replacement for you.

maybe the brainiacs will comment, but i would think you should replace all the high side mosfets on that phase that shorted, even if only one is shorted. jmho

post up pictures!

[ic3wall]

did you open up your controller yet? if you can read the label on the mosfets, then someone may have a replacement for you.

maybe the brainiacs will comment, but i would think you should replace all the high side mosfets on that phase that shorted, even if only one is shorted. jmho

post up pictures!

Yes I did open the controller, and everything looks fine.

According to ebikes.ca, mofsets are 12x AOT460.

[dnmun]

here is the spec sheet: http://www.aosmd.com/res/data_sheets/AOT460.pdf

did not find them at mouser but digi-key has them: http://parts.digikey.com/1/parts/150243 ... ot460.html

ps: if you used them by holding the throttle on and restraining the bike from moving, that is not considered normal use. could not do anything more damaging. imo

[Jeremy Harris]

Worth noting that if you run the battery down to a very low level then you can cause the controller to fail, it's something I've accidentally done on one of my low voltage controllers.

The failure mechanism is that the FET drive voltage drops below the minimum level acceptable, because the voltage regulators inside the controller drop out of regulation because of the low battery voltage. This is one good reason for always having the LVC set correctly for the battery that you're using.

An 8 V voltage drop on a 48V nominal battery pack is way, way too much - at that point the battery is well and truly discharged (and maybe even damaged).

This looks to be a failure caused by running the battery down too low, so after having changed the FET for a new one I'd suggest making sure you have the controller LVC set correctly, ebikes.ca will be able to advise on this.

[ic3wall]

ps: if you used them by holding the throttle on and restraining the bike from moving, that is not considered normal use. could not do anything more damaging. imo

I didn't "restrain" the bike from moving ... it didn't move because power was insufficient due to the battery being ver low. I know the result is the same.

There is a low cut off voltage and a low cut off current on the cycle analyst, I thought that these limits existed to prevent this kind of situation ....

[dnmun]

i was thinking that holding the phases locked in one position, that the charge stored in the hi side pump up capacitor drained down and then the mosfet had a reduced gate voltage (from the cap draining the charge off) while still having to shunt a lot of current through that hi side mosfet to the phase wire. i thought that the low gate voltage would cause big problems with the mosfet heating up.

[ic3wall]

Worth noting that if you run the battery down to a very low level then you can cause the controller to fail, it's something I've accidentally done on one of my low voltage controllers.

The failure mechanism is that the FET drive voltage drops below the minimum level acceptable, because the voltage regulators inside the controller drop out of regulation because of the low battery voltage. This is one good reason for always having the LVC set correctly for the battery that you're using.

An 8 V voltage drop on a 48V nominal battery pack is way, way too much - at that point the battery is well and truly discharged (and maybe even damaged).

This looks to be a failure caused by running the battery down too low, so after having changed the FET for a new one I'd suggest making sure you have the controller LVC set correctly, ebikes.ca will be able to advise on this.

If 8V is too much on a 48V pack then you have to explain that to me. I designed the battery pack myself, and at 40V, each cell was approximately at 3.3V. The minimum nominal voltage of the LiMn cells I'm using is 2.7V. I understand that going under 2.7V could damage the cells ... but at 3.3V ???

By the way, the low current limit was set to 36V, which corresponds to 3.0 for each cell inside the pack.

[Jeremy Harris]

There is a low cut off voltage and a low cut off current on the cycle analyst, I thought that these limits existed to prevent this kind of situation ....

Well, all I can say is that the limits were set way off. Allowing the battery voltage to drop by 20% is a pretty much sure fire way to cause the controller regulator to drop out, causing a loss of FET drive and subsequent failure.

[Jeremy Harris]

If 8V is too much on a 48V pack then you have to explain that to me. I designed the battery pack myself, and at 40V, each cell was approximately at 3.3V. The minimum nominal voltage of the LiMn cells I'm using is 2.7V. I understand that going under 2.7V could damage the cells ... but at 3.3V ???

By the way, the low current limit was set to 36V, which corresponds to 3.0 for each cell inside the pack.

The problem is the controller voltage range. If you have a 48 V nominal controller, then it will have an internal voltage regulator that is set to work at 48 V, maybe a few volts less. If you then let the supply voltage drop too low the FET gate drive voltage drops, causing the FETs to run hot and ultimately causing failure.

LiMn cells are 3.7 V nominal, but usually sit at around 3.8 V or so. At 3.3 V the cell is completely flat, and letting it get to 2.7 V is asking for cell failure; 3 V is the lowest safe voltage I'd go for. Either way, the controller should be set to cut off at around 3.3 to 3.5 V per cell in my view. I never let my packs get below about 3.6 V per cell.

Jeremy

[ic3wall]

If 8V is too much on a 48V pack then you have to explain that to me. I designed the battery pack myself, and at 40V, each cell was approximately at 3.3V. The minimum nominal voltage of the LiMn cells I'm using is 2.7V. I understand that going under 2.7V could damage the cells ... but at 3.3V ???

By the way, the low current limit was set to 36V, which corresponds to 3.0 for each cell inside the pack.

The problem is the controller voltage range. If you have a 48 V nominal controller, then it will have an internal voltage regulator that is set to work at 48 V, maybe a few volts less. If you then let the supply voltage drop too low the FET gate drive voltage drops, causing the FETs to run hot and ultimately causing failure.

LiMn cells are 3.7 V nominal, but usually sit at around 3.8 V or so. At 3.3 V the cell is completely flat, and letting it get to 2.7 V is asking for cell failure; 3 V is the lowest safe voltage I'd go for. Either way, the controller should be set to cut off at around 3.3 to 3.5 V per cell in my view. I never let my packs get below about 3.6 V per cell.

Jeremy

Thank you, I just learned something. I will for sure increase the low voltage and current limits on my CC.

Now, if I wanna save my controller, changing the mosfets should do the job ?

[Jeremy Harris]

There's a very good chance that changing the FETs will fix it. The standard FETs aren't great, and tend to run warm, as they have a high on resistance, so changing them for some better ones would be a good idea. My favourite FETs for a controller in this voltage range are IRFB3077, they have a nice low on resistance and will tolerate the current being upped a fair bit for better torque. You'll find the controller will run much cooler with these FETs than the AOT460 FETs the controller has as standard.

[ic3wall]

There's a very good chance that changing the FETs will fix it. The standard FETs aren't great, and tend to run warm, as they have a high on resistance, so changing them for some better ones would be a good idea. My favourite FETs for a controller in this voltage range are IRFB3077, they have a nice low on resistance and will tolerate the current being upped a fair bit for better torque. You'll find the controller will run much cooler with these FETs than the AOT460 FETs the controller has as standard.

Assuming changing the mosfets solves the problem, it would costs 12*5.24$ + 20$ shipping = 82$ for IRFB3077. The controller is worth maybe 100$. I'm thinking of maybe just upgrading for a better controller like this one:

http://www.methtek.com/2011/11/12/36-72v-25a-30/

Or I could change only the 3 mosfets on phase A with original aot460 for 15$...

What do you think ?

[dnmun]

there is another guy with blown mosfets on that controller with a thread active now, jansevr. buy enuff of them for both of you and split it so the shipping costs are lower. i mail them in a regular first class envelope, taped onto a piece of cardboard, and buy extras for both of you up to the deal point so you get a price break and then maybe if you have enuff sell them here to others later.

better yet buy enuff for the good price break, then you can sell the extras to someone else who appears with the blown AOT460 problem, keeping the overall costs low for everyone. that's what i do. that might keep the cost for the mosfets down to about $2 each with shipping.

[stuartjbecker]

I could definitely use some of those 3077's if you buy extra.