Adding 6 IRFB4110 to a 72 Volt 20 Amp crystalyte controller

[Almasi]

Can I just add 6 Mosfet in parrallel to the one there, or do I need to add so resistor to drive the gate ? (I have never built anything with mosfet before)

I will for sure change the shunt to enable it to go to 40 Amp instead of 20 Amp right now.

I plan on doing this...cause 2 Mosfet have blown Yesterday (on power-up for no reason)

Robin

 

[smithinparis]

The drive resistor that you find there is matched to the capacitance of the MOSFET gate, so if you add another MOSFET you do need to add a new resistor in line with the new gate. Otherwise the RC time constant will be off. Also, be sure to add 4110 only if you have 4110's in there already - the parallel MOSFETs should be matched.
Before you go adding a new MOSFET though, you should get an understanding as to why the other ones blew. Those are pretty pricey and it would be a shame to have the new ones blow too. Which 2 FETs blew? Are they 2 that are side-by-side or part of separate phase drivers?

 

[fechter]

I'd recommend completely removing the original FETs and replace them with the 4110's.
You don't want to parallel different types.

There's a good chance the gate resistors also blew along with the FETs. Check the resistance across them. There is also a chance the gate driver blew. This is a bit harder to check. Measure resistance from the gate to both Bat+ and Bat- lines and look for shorts. Compare to an unblown FET.

 

[Almasi]

The OEM 72 Volt 20 Amp allready have 6 IRFB4110 (this is why I bought it)

I have the 2 first Mosfet are shorted, it happen on power up, but I did over 1000 miles with it before, it dicided to died!

I tested the resistor and diode (that are in parrallel to drive the mosfet gate)...and found out that they are perfect, so I tested the IR2101 Output 5 and 7 are giving 0 volt and 0.030 volt with power ON just like the other phase 2 and 3. (so my guess it's that it's not burn)



Robin

 

[Almasi]

BTW there is a PCB layout error on the 20Amp 72 Volt version I have.... the 100 Ohm resistor in parrallel to drive the Mosfet gate is not attached to anything on one side for the first Mosfet, only the first phase, motor phase 2 and 3 are OK...like somebody forgot a line

 

[Almasi]

This is the error they made on it (BTW it's resistor R54)

I guess that this might have lead to my blown Mosfet that where on Phase 1 ouput (exactly where this error is)

Robin

 

[smithinparis]

If, when you say it's the first two MOSFETs that blew, they are immediately beside eachother, they are likely the high and low side of the same phase. If the Source (S) of one MOSFET and Drain (D) of the other MOSFET are connected to the same output line going to the motor, then they are part of the same phase.
If this is the case, then what probably happened was that you had a short (with moisture maybe ... it happened to me...) that caused the high and low side to turn on simultaneously causing a direct short for your battery - the current didn't flow into the high side, to the motor, and then down the low side of a different phase - rather it went into the high side and low side of the same phase, and never saw any real load (the motor).
That said, adding greater current handling by adding a parallel MOSFET won't solve your problem. The same short would have just as easily blown 4 MOSFETs as it would have blown 2.
Was the controller wet at all?

 

[Almasi]

If, when you say it's the first two MOSFETs that blew, they are immediately beside eachother, they are likely the high and low side of the same phase. If the Source (S) of one MOSFET and Drain (D) of the other MOSFET are connected to the same output line going to the motor, then they are part of the same phase.
If this is the case, then what probably happened was that you had a short (with moisture maybe ... it happened to me...) that caused the high and low side to turn on simultaneously causing a direct short for your battery - the current didn't flow into the high side, to the motor, and then down the low side of a different phase - rather it went into the high side and low side of the same phase, and never saw any real load (the motor).
That said, adding greater current handling by adding a parallel MOSFET won't solve your problem. The same short would have just as easily blown 4 MOSFETs as it would have blown 2.
Was the controller wet at all?

The bike was inside at room temperature, next to a fire place, so there is no moisture issue (the controller had opening...so moisture would have came out of it after a week)

They were the high and low side drive of the same phase...that same phase that has the error on PCB that make the low side driver only driven trought a diode and NOT the diode in parrallel to a resistor!!!

I know that having more mosfet would have solve this issue...but I might as well upgrade for more Amps

Robin

 

[fechter]

That missing trace could certainly be why it blew up in the first place. The FET could possibly switch on half way and over dissipate (fry). It's kinda hard to believe they actually did that. I wonder how many there are like that? Seems like it wouldn't be firing on all phases either.

If you want to drive pairs of FETs in parallel, connect the source and drain leads, but use a separate resistor and diode for each gate lead that goes back to the driver chip.

For single 4110's, I think you could go up to at least 30 to 35 amps without blowing them.
Check your main capacitor(s) too. That is another common failure point that can blow the FETs.

 

[smithinparis]

That missing trace could certainly be why it blew up in the first place. The FET could possibly switch on half way and over dissipate (fry). It's kinda hard to believe they actually did that. I wonder how many there are like that? Seems like it wouldn't be firing on all phases either.

So basically to turn the FETs on, the diode would actually enter the breakdown region. Not really good practice ...

 

[Almasi]

That missing trace could certainly be why it blew up in the first place. The FET could possibly switch on half way and over dissipate (fry). It's kinda hard to believe they actually did that. I wonder how many there are like that? Seems like it wouldn't be firing on all phases either.

So basically to turn the FETs on, the diode would actually enter the breakdown region. Not really good practice ...

So the efficiency of that phase was affected right ?

 

[Almasi]

That missing trace could certainly be why it blew up in the first place. The FET could possibly switch on half way and over dissipate (fry). It's kinda hard to believe they actually did that. I wonder how many there are like that? Seems like it wouldn't be firing on all phases either.

I checked 2 different 72 Volt 20 Amp (odered at the same time in April 2008 from Ebike.ca) the first one was mine that blew 2 Mosfet (low side and high side from phase 1) that same phase that has the error, and the second controller (my friend) that blew up on the second power-up at 33 Volt A123 setup???? (it blew up at start up just like mine....but he didn't have any fuse...so it burnt almsot every component inside including the 6 IRFB4110) (Ebike.ca sent another one on warranty very quickly!!! which didn't blew) They both don't have one side of R54 connected to anything, the diode that it's supposed to be parrallel to is on the other side of the PCB and it's only one missing 'VIA' to make the connection or a trace that would be 4 mm long.....weird that they forget a small trace like that???

Robin Almasi

 

[Almasi]

Here is the PCB from the controller that was completly dead, I took out many damaged component...there is still more to remove

The missing trace is show here (where the blue line is)

 

[Almasi]

Here the other side of the PCB that show all the channel coming out of IR2101 going to the gate through a diode attach in parrallel to a 100 Ohm resistor , except for the first mosfet (low side driver of phase 1) that one as a resistor one the top side, and the diode on the other side where all the other 5 Mosfet have their resistor and diode.

Robin Almasi

 

[fechter]

Now that all the parts are removed, try measuring resistance across where the trace is supposed to be with a meter. It's possible there's a hidden between layer?

 

[Almasi]

Now that all the parts are removed, try measuring resistance across where the trace is supposed to be with a meter. It's possible there's a hidden between layer?

Don't worry, I tried that many time (I'm an electronic technician)

There is no 'via' either that goes to that pad.

Robin

 

[fechter]

OK, just checking.

That's really retarded. I wonder how many they made like that? Easy enough to fix...

 

[billvon]

That's really retarded.

Yep, but works well enough to pass a screening test, so out the door it goes.

 

[Almasi]

That's really retarded.

Yep, but works well enough to pass a screening test, so out the door it goes.

I did about 1600 KM with that controller with the missing trace...before 2 mosfet died (probably caused slowly by this) so I guess it will feel better or at least get more efficient once it's fixed

Robin

 

[dnmun]

can you tell if the pcb is missing the trace because of poor patterning or resist liftoff during silkscreening and developer? maybe their mask was damaged at a certain point and all the boards after that have the trace gone.

 

[Almasi]

can you tell if the pcb is missing the trace because of poor patterning or resist liftoff during silkscreening and developer? maybe their mask was damaged at a certain point and all the boards after that have the trace gone.

I did not find any sign of 'half trace' it's seem like it's just not there at all??? (even looking trough the PCB with strong light, I don't see anything that goes there)

Robin

 

[Almasi]

Good news for other Crystalyte 72 Volt 20 Amp Pedal First controller, newer one from around June/July 2008 and up have this issue fix.

Here is a picture, they put more solder to make a bridge from R54 to Mosfet Low Side phase 1 (yellow)

Good news from Ebike.ca ,I Emailed Justin about it, he will send me free replacement Mosfet for it Very good support/warranty from them!!!

It would probably be a good idea to make this little PCB error a 'sticky' post or added to the other Crystalyte fact ....no?

Robin

 

[BocaBen]

I checked my 72V 20A pedal first controller after the warning on this thread and I do have the soldering blob around R54.
Thanks for doing the research and providing the info.

 

[OzCad]

The missing resistor would allow Cdg capacitive coupling between drain and source to generate transient voltages on the gate which could exceed the gate to source breakdown voltage.
In this case, when the upper FET turns off, motor inductance drives the lower FET drain abruptly negative, from positive near Vbat to 1 diode drop below ground. The now-blocking series gate diode does nothing to prevent the gate from driving negative by almost the full drain transition swing, blowing the lower FET gate, possibly shorting it. The next time the upper FET turns on, you have a shorted totem pole.
In higher power circuits a (typically) 15 volt zener or transzorb is commonly connected between gate and source to limit such capacitively coupled over/under voltage gate transients.

 

[Cobber]

The Power MOSFET IRFB4110 is rated for maximum VDS of 100 volts. In a 72 volt controller the main power rail can be as high as 84 volts when the battery is fully charged. As the MOSFET turns off the motor will generate an EMF from the magnetic field collapsing. This can be as high as 30% of the voltage applied. This on top of the 84 volt rail will exceed the maximum VDS of the MOSFET and cause instant failure. To build a 72 volt controller you need MOSFETs that are rated at a higher voltage. Some manufacturers try to solve this problem by suppressing the spikes with a snubber circuit.