Solectria (Brusa) AMC230 Repair

On April 17th 2013, A Solectria AMC230 motor controller S/N 514 came in for repairs. It was no longer working.
This is one of two AC induction motor controllers in a 1993-94 Solectria E10.

April 19th 2013, it made it to the work bench.


Hooking up high voltage, resulted in a large BANG during the pre-charge, So I quickly removed power.

Taking all the covers off the controller to see if there is any sign of arcing.


One thing that stands out is the bloated caps, resulting in the shrink wrap exposing the aluminum can.

Current sense board removed, to get a better look at the caps.


At the time I thought that one of the capacitors was arcing over at around 170V, (it later turned out to be a diode failing). The spark happens at 1:41 
Despite being warned ahead of time, it gave my wife, who was recording the video quite the scare.

April 23th 2013, Since it looks like the spark came from inside the gate drivers. I took the fan assembly and heat sink off.

With the gate drivers removed, you can see where the arcing was happening. Both screws go to one of the bulk caps + and - terminals.


The gate driver took some collateral damage.


I placed some 220 ohm resistors across the gate and source pins on the MOSFETs (this keeps them from turning on, since I pulled the gate driver boards).


I then hooked up the battery pack thru a pre-charge resistor, and it still sparks. This points to one of one of the half bridges is failing (punch thru) at around 170 volts. Standoff rating for the MOSFETs should be at least 300 volts (it could also be one of the diodes that is failing).

April 27th and 28th 2013, Next I removed all the bulk caps, and broke down the power stage into individual half bridges.


Then I diode tested across each individual half bridge, the one on the left is reading more than a volt higher than the others.


Diode testing the lower legs, they all look fine.


Diode testing the upper legs, the one on the left is still reading higher then the rest, but technically it is still working.


But since I know that something fails at a higher voltage, I built up a test rig using two 100W lights in series, connected to the 216V battery pack.
I also added some wires to the half bridge to prevent damage to the circuit board from arcing.


Connecting the lights and battery thru the half bridge causes the bad diode to break down and conduct.

Using a current meter to check which side of the half bridge was conducting. 


I then marked that side bad and checked the other half bridges, they all tested good.


I noticed that when the bad half bridge was conducting, it was getting warm, but with the heat spreader connected it was not possible to determine which part was bad, so I took it off.


Connected back up to the test load.

This diode got so hot you could not touch it.

After removing the bad diode, the half bridge no longer conducts.

On closer inspection, the diode has a hole blown in it.


Testing with a 9V battery across the 220 ohm gate-source resistor, showed that the MOSFETs are still working.


One thing I noticed, is that when the half bridge was built they did not get enough heat on the center pin to properly flow the solder. First pic is before, second pic is after I re-flowed it.


This might have been the cause of the failure, since the bad joint could have started to arc resulting in it overheating the diode, which caused it to fail internally.

Next I checked all the caps, leakage current, bulk voltage, and capacity.


All the caps where close to each other except the one that is slightly bloated, although it does hold voltage, it does have higher leakage current, and less capacity then the rest. 

This was the cap that was located closest to the bad diode, so it probably overheated too, resulting in vented electrolyte which reduced its capacity. Since the bulk caps are 20 years old (This motor controller was made in 1993), I decided to replace all the caps with new ones.

So, after only finding one half of a half bridge that was bad... there has to be a path for the current to flow... And that path was thru the gate driver.
The last one is shorted.


Each part I removed tested shorted...


Since it looks like the 15 volt supply on one half of the gate driver was exposed to 216V, everything is shorted, or blown open. So I decided to remove all the parts:


May 2nd 2013, new parts came in.

May 4th and 5th 2013, replacing the bad diode with a new one (closest equivalent diode I could find, since the others are obsolete), using the heat spreader to space it properly.


Securing the heat spreader, using the heat sink to space it properly.


Cleaning off excess silicone from the inductors.


Reassembly of all the half bridges onto the power bus bar.


New caps, 50% more voltage in 50% less package, what a difference 20 years makes.


New caps installed in the power stage.


Checking for leakage current/voltage... unfortunately, the section I just replaced still had a bad diode in it. I watched as the voltage slowly climbed up to 42 volts before I disconnected the battery pack.


With the heat spreader removed, pack voltage connected, and using a hot air pencil to heat each diode the bad one was located.


Checking for leakage current/voltage again... this time it looks like all the bad diodes are removed.


Given the fact that two diodes had gone bad, and current sharing between the old diodes and the new diodes might not be ideal. I decided to replace the other 6 diodes.


New diodes soldered, except the middle pins, using the heat spreader to line them up, and the heat sink to line up the heat spreader.


Heat spreader removed, to solder the middle pins.


Securing the heat spreader, using the heat sink to space it properly.


Finished half bridge power section.


Half bridge installed back onto the power bus bar, with a dab of thread locker.


Caps installed, also with a bit of thread locker.


Final check for leakage current/voltage... The new diodes have around 10 times less leakage... but that is what you would expect from new modern diodes.


May 11th and 12th 2013, new parts to rebuild the gate driver.

Populating the circuit board, using one of the other gate drivers as a part location map.


I could not find one of the transistors in a TO-92 package, so I had to use a SOT-23... I had to make an adapter.


More parts soldered into the gate driver board.


Finished gate driver board.


All the gate drivers, control electronics, and current sense board installed.


Ready for bench testing... And after all that, it still does not work. Something is shorting out the main supply.

Isolating the power supply board from everything else and hooking up the battery pack, the standby supply is still working.
But as soon as the main supply is commanded to come on, it drops to 0.23 Volts (should be around 18 Volts).


Since there are lots of parts on the 18V main power supply, I used an LED string to current limit a 12V battery to 600mA.


These guys get really hot... and yup, both MOSFETs are dead shorted.


Tracing out the circuit to and from the bad MOSFETs, and it turns out that they supply power to all the gate drivers... 
So it looks like the following conduction path happened: 
Battery pack positive --> shorted diode in top phase --> gate driver board --> main power supply --> Battery pack negative.

Power supply board back in, same test as before... commanding the main supply to come on results in 17.88V.


A quick scope check shows that transistors that drove the bad MOSFETs survived intact.


New MOSFETs installed.


Ready for bench testing again... And this time it works.

Video of the repaired AMC230 spinning a motor.



Old heat sink compound removed, surface cleaned.


New heat sink compound applied. The excess squeezed out and cleaned up, after torquing all the heat sink bolts.


Inductors and capacitors bonded with silicone.


Control electronics and current sense board installed.


May 13th 2013, removal of old silicone gasket material from housing.


New silicone gasket material for housing.


Motor cables and power cables hooked back up.


Final cover installed, with the empty parts box on top.


Final bench test.

May 15th 2013, packed and shipped.


Total repair time was about a month, with actual time spent on the weekends.

More EV Projects