On Wed, 19 Mar 2014 01:34:05 +0000 (UTC), wrote:

wrote:
One of my CNC mills has a problem. The machine, a Fadal 15XT, built in
1998, (I think), keeps showing an alarm for a malfunction the machine
can't have.

Does the problem board have any mechanical support other than the edge
connector? Is this support aligned correctly and connected? (Example:
the bracket on most PC expansion cards that screws down to the chassis.)

The board is properly supported.


You might also look for metal debris, oil, old grease, dried coolant,
etc, on both sides of the problem board and on the "motherboard" that it
plugs into. This may mean removing the motherboard from the cabinet,
which is generally a giant PITA. I know *you* would never run the mill
with the cabinet wide open, but you have no idea what the low-down
so-and-so who owned it before you did.

The board and the inside of the cabinet is clean. I have been running
the machine with the cabinet open though, just to get a couple jobs
done. I have made sure that nothing can get inside when open.

I like Cydrome Leader's idea of possible bad solder joints. It could
also be that the edge connector on the motherboard has one or more
contacts that have lost their springiness. This is partly determined by
eliminating other causes, but if there are other cards with the same
connector, you can get an idea by unplugging one of those cards from its
slot and reseating it. Compare the force it takes to do that with the
force it takes to do the same thing with the problem board. If the
problem board is way easier to remove and replace, you might suspect the
connector. If you conclude that the connector is bad, usually you can
replace just the connector, but this may involve unsoldering a bunch of
pins.

I'll try the card swap to see if I can detect any differences. If the
connectors are less springy if I can't bend them back where they
belong I will unsolder the connector and replace it. Unfortunately the
board it plugs into is buried behind everything else and will take a
lot of time to remove. And there 100 contacts in the edge connector.
I carefully inspected, with a 10x magnifier, the board for any bad
solder joints or cracked traces and found none.

If it's a bad connection, you might be able to induce the problem by
poking at the board(s) with a nice long dry plastic stick, as long as
you're reasonably sure that there isn't 120 V or 240 V running around
the board. Have a hand on the e-stop and/or circuit breaker when you do
this, just in case you manage to send a signal for "drive the table 20
feet that way and turn on the spindle to 100,000 rpm" by poking.

If I turn off the machine and let it cool for even 4 hours the board
will still throw alarms. However, if I remove the board and put it
back in the alarms stop.

If removing and reseating the board is an instant fix, I think this
points heavily in the direction of a bad contact somewhere. If you had
a failing component, usually you would have to wait for that component
to cool down enough to start working again.
Looking at the connector contacts on the board I can see that some
have more obvious marks than others on them from the contacts in the
female connector. Hmmm.

I like this idea less, but here it is: if the sensor is a relatively
high-resistance device, a couple of things could be happening. 1) Some
foreign glop on the board in exactly the right (wrong?) spot has
changing resistance with temperature, and eventually gets low or high
enough to trigger the alarm. This wouldn't tend to go away just by
unplugging/replugging, though. 2) The alarm input, itself, has a high
impedance, and eventually drifts to a high enough voltage to trigger
the alarm. This *might* fix itself by unplugging/replugging.

There is no foreign matter of any kind on the board and the machine
has no sensors. This is why the alarm is spurious. Could failing caps
be causing problems? Could unplugging the board allow some caps to
discharge which then makes the alarm go away?

Sometimes if there is a sensor that your model doesn't have, the input
for that sensor will have a jumper wire or resistor across it as a
dummy. Can you tell if anything like that used to be there and is now
missing? You might have to compare to photos of a similar machine -
this may or may not be in the manual.

I'll see what I can figure out about what may be different on this
board from one in a machine equipped with sensors.

I have cleaned the contacts with alcohol and "contact cleaner".

You might try taking the problem board out of the machine and sticking a
clean piece of stiff paper, like a folded index card, into the slot in
the edge connector that is still in the machine. Cut the paper to the
same length as the card edge connector - don't use a shorter piece of
paper and slide it back and forth, because you may catch one of the
contacts and bend it. It may also help to wrap the paper around a thin
piece of material (sheet metal?) - but you want the whole stack to be no
thicker than the circuit board that belongs there. Remove the paper and
inspect for signs of glop or corrosion in the edge connector.

I'll try your folded paper cleaning strategy and report back.

So after the long winded explanation above I am wondering if dirty
contacts could somehow make the board run warm and cause the alarms.

Dirty contacts will heat up if there is enough current going through
them. However, I would guess that a board that monitors pressure
sensors (and probably interlock switches and stuff like that too) is
running on relatively low voltage and current. If there was a board
driving the servo motors or coolant pump or something that takes
relatively more juice, then it would have a worse time with dirty
contacts.

If you can identify the low-voltage DC power supply (probably 5, 12,
or 24 VDC, but it could be a lot of things), it probably wouldn't hurt
to measure its output voltages with a multimeter, both when the mill
is working OK and when it is throwing alarms. This probably isn't the
problem, but low (or high) power supplies can cause many weird effects.

Matt Roberds
Thanks for the suggestions Matt.
Eric

---
This email is free from viruses and malware because avast! Antivirus protection is active.
http://www.avast.com