Re: [CAD_CAM_EDM_DRO] Gecko failure/repair

resosys wrote:

>It seems as though one phase output was blown. There are no burn
>marks or blown parts on the top of the board that I can see. Since
>these are the open case selloff units with no warranty, I removed the
>heatsink/mounting plate and examined the drive FETS. One of them next
>to the mounting screws shows a slight burn mark on the leads.
>

>Are there any specific tests to perform?
>
>

Well, if you pull the fets out, and hold them with the plastic side
toward you,
and the leads down, then the left lead is gate, the middle and case is drain
and the right is source, at least on most power fets. Check from gate
to source
and you should read infinite resistance. If not, you can chuck it right
there,
no further tests needed. Then, check from drain to source. You might put a
clip lead from source to gate to make sure the gate is discharged. You
should
read like a diode one way, and infinite resistance the other way.
(Different
meters read diodes differently, some show the voltage of a forward-biased
diode, about .62 V, others don't use that high a voltage in the Ohms
mode, and
you get inifinte. Those meters usually have a diode test function.)

If the FET reads some modest resistance both ways from source to drain, and
the gate is discharged, then it is bad. (Leaving the gate lead open can
fool you,
the gate charge on a good FET can last for hours.)

You should check the current sense resistor, too, to make sure it is good.
It will be a very low resistance 0.1 Ohm or less. Just make sure it reads
lower than your meter goes, and it is probably good.

Checking the circuits that drive the FET's gate is harder, and will be
specific
to the design there. I seem to vaguely remember Mariss used discrete
bipolar
transistors there. The problem is that when a FET blows up, it can cause
DC bus voltages to come through the gate lead and fry the driver components.

Jon