Re: G201 death/repair/questions/problems (long post)(long answer)

> Questions/thoughts:
>
> I've got the motors set to 3amps via the current set resistor.
>
> I don't have any fuses between the G201s and the power supply. I

was

> under the impression that this wasn't necessary but will install

them if

> it will help. Any thoughts?

In this circumstance the fuses might have prevented the destruction
of the FETS. FETS are pretty rugged devices and will take abuse. Two
things that will kill them are voltage spikes that exceed the ratings
across drain and source (probably not the case here). I have seen
spikes of just a nanoseconds that greatly exceed the ratings case a
junction breakdown. The only way to see spikes of that duration is
with a high speed triggered storage scope. Most failures are
evidenced in a shorted FET. One FET shorting could cause the cross
bridge FET to drive into zero resistance. I would have to go with
the earlier suggestion you are getting a short between the case and
one of the devices. The other goes out in sympathy.

>
> I only have 10,000uF of capacitor on the power supply. When I

built the

> supply a year ago, I remember doing the math to make sure it was

within

> the specs in the Gecko literature. Will a high ripple voltage cause
> damage to the Geckos?

Ripple would only be a problem as the total load increased and then
probabaly not an issue with the FETS. You can actually use a FET as
a low loss rectifier and two of them back to back will make a pretty
good AC switch anthough driving them at high speed is a challenge.

>
> I do not have the disable pin connected on the G201s. Is this

something

> that should be pulled to ground until everything is powered up and

I'm

> ready to move motors? I was planning to do this but haven't yet.
>

I don't think that is an issue. I have a 201 running with no
diasable signal and never a problem.

> Voltage spikes? I've done tests with my power supply that suggest

there

> are no spikes when it is switched on. Any thoughts?

I doubt that a turn on of the power supply into a heavy inductive
load would cause a spike bad enough to short a power FET.

>
> Is there a correct power up sequence? I've been leaving my parallel
> port connected and the software running while only turning off the

power

> supply/Geckos/motors.

My DC to my motors is on a contactor that is controlled by several
limit and e-stop switches. I often leave my computer running and
turn off the DC to the motors. That is the safest way to make sure
nothing goes anywhere when it should not.

>
> I'm hoping someone has seen this or a similar issue and can help me

find

> the problem prior to me making more smoke.

One of the things you probably need to do is get the Gecko's onto a
bench with a powersupply and some non-inductive loads. You will need
a source for pulses and a load across the output so you can get a
scope on the situation. You should be able to see if they are
switching correctly. If something is messed up in the drive
circuitry then things will conduct at times they should not. I do
not know the details of the Gecko bridge but I assume it is of the
charge pump type to generate the voltage for the upper FETS. Here
are some issues. If for any reason the wave form on the gates of
either lower FET is not crisp with rapid turn on or off you can
generate lots of switching losses. The positive charge turns the FET
on and since the gate has a large capacitance it "stores" the on
signal. A simple resistor to ground will drain that off but most
designers use an active driver scheme (totem pole) that stuffs
voltage into the gate and rapidly pulls it out to turn the FET off.
Flakey drive will cause FETS to be on at times they shouldn't.
Conduction across the bridge (upper to upper OR lower to lower) in
anyway will have a bad outcome.

>

>
> Last week I posted about letting the magic smoke out of 2 of my

G201s.

> I did some surgery on one of them this weekend and got it back to
> partial health.

There is no such thing as "partial health" with semiconductors. It
like saying an airplane with one wing is partially flying! If its
not working 100% correctly turn it off and determine why.

One of the phases is still acting odd. I only had

> IRF530N power mosfets on hand and only enough to replace the parts

that

> tested bad. I've ordered more IRF530's and some resistors. Some
> questions here:
>
> I've been led to believe IRF530N and IRF530 fets are

interchangeable.

> Any wisdom here?

In looking at the specs of both, the N version is an improved
version. The on resistance is slightly lower (.11 VS .16); no big
deal. Input gate capacitance is slightly higher; no big deal and the
switching characteristics are similar. I don't see why the
substituion would be a problem. I would replace all of the bridge
with the same type: either 530 or 530N just to eliminate the
possibility of mismatches

>
> Is there an easy way to test the mosfets in circuit? I've got a

nice

> Fluke meter that does diode testing. (If not, I'll replace them

all)

Most FETS fail shorted between the Drain and Source (Zero OHMS)
pretty easy to see on a meter. There is an intrinsic Drain/Source
diode that helps with commutation that on a good FET you will see as
lower resistance in one direction than the other. The gate is hard
to test since the drive circuitry can effect the readings. Best
thing is to record the readings of a known good set (if you still
have one left) and use that is a criteria.

>
> When running the partially repaired G201, the red LED flashes and

the

> motor is jumpy as if one phase won't swap polarity. Has anyone seen
> this behavior before?

Something is shorting out the power to the circuitry. I think the
LED on a 201 is just a power-on indicator. If the regulator on the
board that supplies the regulated voltage to the logic chips is
flakey it could cause real problems.

>
> Should the reddish-brown capacitor be replaced when the .5ohm 3watt
> resistors go?
>
> My meter reads .2 ohms on the resistors so I'll replace them.

I doubt your meter is accurate enough at less than one ohm to detect
the difference between .2 and .5 ohms. Power resistors are basicly
either coils of resistance wire or chunks of carbon. In the first
case the failure is almost always obvious (burned) and the unit will
be open (no resistance). Seldon do the turns short together and form
a resistor of a lower value. Carbon can change value if it is
overloaded but will normally display burn marks as well. Measure the
other resistors in the other units. See if they read the same. Make
sure you have removed any shorted FETS prior to the test.

>
> Are there any tests I can do on the motors that will tell me if they
> have problems?

Check for shorts between the two sets of windings and between each
winding and the case with an ohmeter.

>
> When things were running, the system was working well. The axes

were

> smooth, turbocnc's pulse train was smooth, calibration was going

well,

> and I was happy and so close to making my first cut. It will happen
> soon.

You will discover the error of your ways (;>) and get past this
issue. The probability that there would be multiple Geckos that fail
in the same way without there being an external case are so low as to
put that with winning the lottery. Your root cause is external for
the initial failures but first you must get the units repaired and
working on the bench 100% Don't put them on highly inductive loads
(motors) until you have let them run a resistive load under power.

>
> Thanks for reading my long winded post and for any wisdom you may

have

> that will help me get things fixed/solved/running.

The hardest thing to fix is a issue where there is more than one
problem at the same time. We will help as best we can.

>
> Chris