dermot
100 W
- Joined
- May 15, 2007
- Messages
- 172
My Crystalyte 36/72V 35A controller (married to a front wheel X5) shuddered to a halt a couple of weeks back.
I have a Lyen controller as a spare, but it didn't drop in at all well (more on that later*).
Yesterday, I had a go at fixing it - when I took the board out of the housing, I found that two of the three large capacitors had dropped off and the third one had split with goo leaking out of the opposite end to the wires. (So, if you go offroad, it might not be a bad idea to apply some hot glue to fix the caps securely - there was just a small blob of glue on mine)
I'm assuming that with only one cap in circuit, the higher ripple current caused it to fail.
There were signs of arcing adjacent to where one of the caps used to be - looks like it caused a short when it fell off. This is a 12FET controller, 4 FETS per phase, 2 high and two low. Some of the legs on two were blown off (and the FETS measured dead short) the other pair were still in circuit but dead short.
Fortunately I had a box of dead controllers donated by Jozzer some time back (Thanks mate!!!) so checked that I had 4 spare FETs with no signs of short circuit and soldered them in and replaced the caps likewise.
Without refitting the heatsink hardware, I suspended the motor above the ground and fired it up, fed with 60V from a bench PSU set to 1A current limit. For a second or so, it spun up with no signs of distress, but then ground to a halt saturating the PSU and pulling the volts way down. One parallel pair of FETS were short again.
Replaced them again, checked more carefully, but got the same result with the same two FETS dead short.
Then looked at resistance checks across all 12 FETs - each pair of FETs had 20 Ohms between the gates - traced that to a pair of 10ohm tiny surface mound resistors, the common point of which goes to a driver transistor. The resistors were perhaps 100mW and both open circuit on the faulty FET locations.
Looks like the initial flashover failure popped the 10Ohm gate resistors, and if you fire up a controller with open-circuit gate connections, it looks as if leakage blows up the FETs - even current limited to only 1A. So if you are faced with swapping FETS, it is a good plan to check the gates resistors are still OK. I took a pair out of a dead controller (Thanks again Jozzer) - a pair of tweezers and doing this *before* having a beer at lunch helps quite a bit, as a steady hand is required to swap resistors small enough to go through the eye of a large sewing needle!
And got a result - controller ran fine for 5 mins with no significant temperature rise on the FETS and around 0.8A drawn from the PSU. Put it all back together and rode for a few miles at full throttle pulling 35A for quite a bit of the time.
All of this is probably old news to many people, but thought it worth posting my experience.
Dermot
* And the Lyen controller? I'd initially ran through all the colour combinations and found a combination where it run forward smoothly and pulled no more than 1A on no load. On the road, however, there was very little torque and it pulled 25A with almost no torque. I'd assumed that there was only one combination that gave smooth low current operation - either this is not the case or my controller came with a fault - anyone have experience of this?
I have a Lyen controller as a spare, but it didn't drop in at all well (more on that later*).
Yesterday, I had a go at fixing it - when I took the board out of the housing, I found that two of the three large capacitors had dropped off and the third one had split with goo leaking out of the opposite end to the wires. (So, if you go offroad, it might not be a bad idea to apply some hot glue to fix the caps securely - there was just a small blob of glue on mine)
I'm assuming that with only one cap in circuit, the higher ripple current caused it to fail.
There were signs of arcing adjacent to where one of the caps used to be - looks like it caused a short when it fell off. This is a 12FET controller, 4 FETS per phase, 2 high and two low. Some of the legs on two were blown off (and the FETS measured dead short) the other pair were still in circuit but dead short.
Fortunately I had a box of dead controllers donated by Jozzer some time back (Thanks mate!!!) so checked that I had 4 spare FETs with no signs of short circuit and soldered them in and replaced the caps likewise.
Without refitting the heatsink hardware, I suspended the motor above the ground and fired it up, fed with 60V from a bench PSU set to 1A current limit. For a second or so, it spun up with no signs of distress, but then ground to a halt saturating the PSU and pulling the volts way down. One parallel pair of FETS were short again.
Replaced them again, checked more carefully, but got the same result with the same two FETS dead short.
Then looked at resistance checks across all 12 FETs - each pair of FETs had 20 Ohms between the gates - traced that to a pair of 10ohm tiny surface mound resistors, the common point of which goes to a driver transistor. The resistors were perhaps 100mW and both open circuit on the faulty FET locations.
Looks like the initial flashover failure popped the 10Ohm gate resistors, and if you fire up a controller with open-circuit gate connections, it looks as if leakage blows up the FETs - even current limited to only 1A. So if you are faced with swapping FETS, it is a good plan to check the gates resistors are still OK. I took a pair out of a dead controller (Thanks again Jozzer) - a pair of tweezers and doing this *before* having a beer at lunch helps quite a bit, as a steady hand is required to swap resistors small enough to go through the eye of a large sewing needle!
And got a result - controller ran fine for 5 mins with no significant temperature rise on the FETS and around 0.8A drawn from the PSU. Put it all back together and rode for a few miles at full throttle pulling 35A for quite a bit of the time.
All of this is probably old news to many people, but thought it worth posting my experience.
Dermot
* And the Lyen controller? I'd initially ran through all the colour combinations and found a combination where it run forward smoothly and pulled no more than 1A on no load. On the road, however, there was very little torque and it pulled 25A with almost no torque. I'd assumed that there was only one combination that gave smooth low current operation - either this is not the case or my controller came with a fault - anyone have experience of this?
I have repaired many! I have spare surface mount gate resistors and fet drivers chips and fets for that matter, yes i would routinely change the gate driver resistors they always go, I used to desolder the old resistors however most folks simply piggy back them over the dead ones which saves a bit of time, I used a very thin piece of blu tack rolled up in to a point to hold it in place whilst soldering. Yes the caps always come loose in those controllers, hot glue them back in place they will be fine, another thing to watch on them is the cheap pass through plastic rivets they use on the fets, I used to have lots of problems with them failing, I replaced all mine for quality ones from farnell, there are loads of posts if you dig right back in the archives on what we did, I am glad I dont have to do too much these days, they are pretty reliable those controllers, I have 2 brand new ones here still, I also have one that I modified for 100V which is handy, replaced all the fets in that one.
