EBX board spontaneously resetting.

The last time I had a problem like this I was using an industrial PC enclosure in which the rear bracket was spaced too far behind the motherboard by about 1/8". When the PCI cards were pressed into place and their mounting brackets nudged slightly to get the screwhole aligned, it created a tension that tended to lift the card out the motherboard socket after time and vibrations.

In general, when the gold card edge fingers are fully inserted into the motherboard socket, the card bracket mounting slot must line up perfectly with the screw hole and the top of the mounting bracket must just touch the top of the rear enclosure bracket. If not, stresses will be introduced that can cause the card to eventually work its way out. Some enclosures are bad about this, and some are even adjustable so alignment can be corrected. In the situation I mentioned earlier, it was necessary to shim the rear bracket forward after drilling out the spot welds and replacing them with screws.

This is just one possible explanation that seems relevant since you reported the motherboard positioning was critical. Also ensure there are no vias or traces close to the motherboard mounting holes that could be getting grounded by the brass standoff, and use non-metallic washers on the screws.

TTaylor

Hi Tom,

Thanks for tak> The last time I had a problem like this I was using an industrial

That's a good suggestion, but in my case I'm not using any expansion cards. :(

It seems to me that something is happening to the board in transit or when it is being used in the client's environemnt, because before shipping it did not exhibit any pickyness about grounding or placement. Upon return, though, the symptoms arise.

Regards,

trena

Feels about right. Which would mean you need to check your customer just as thoroughly as you do your hardware, next time. This may mean you have to travel there and watch for yourself what they're doing to that poor board...

What you might try doing is to ship a board somewhere via the same carrier and just have it shipped back, then test it. A company I worked for had a similar problem although their product was much more complex. They would test it thoroughly, then ship it, and the customer would report it as non-working. When hand carried to the customer, the same type product would work. They redid the manufacturing process and fixed the problem.

It's almost like you have the need for some power isolation capacitors some where. It would be interesting to see if you could put your finger on the back of some chip leads through mounting holes and also have it work. Maybe you have noise that comes and goes with the topology of your PC board where slight pressure makes all the difference.

Try using insulated spacers on the motherboard. Also carefully inspect around the screws that fix the motherboard to the present spacers with a monocle, or some magnifying eyeglasses. Sometimes there may be a burr underneath the screw head which may cut through the masking layer of the pcb and create a short circuit to the chassis via the metallic spacers. Also check around the spacer locations underneath the pcb with a magnifier. Some problems may not manifest themselves until shipping to the client due to vibration when transporting the system.

Thanks, everyone, for the replies so far. I will have the EBX system back probably Thursday or Friday, and I do intend to let the group know what I find in the hopes it might help someone else in the future.

Regards,

trena

Tom - thank you again for your response (and others too!). Your suggestion to use insulating spacers between the screws and the board led me to try using nylon screws, and believe it or not, the resetting behavior has not repeated.

We got the system back on Wednesday, and it was immediately displaying the resetting behavior consistently. Using nylon screws to hold the motherboard appears to have done the trick. We have poked and prodded at the board while it's running, shaken the system up (NOT while it's running, :), run some CPU stress tests for hours to heat the board up and it appears to be stable again.

Tomorrow we'll be able to test the first EBX board to see if using nylon screws also removes the symptoms from that one.

Amazing how something so simple can make such a difference.

Regards,

trena

Why not while it's running? We used to test mainframes for loose connections and stray wire by raping on each card while the system was running. Of course you can use too much force but it took a fairly good hit to do it right.

Another win for "Percussive Maintenance!"

You're quite welcome, trena. Glad to be of help.

It appears that either 1) someone drilled out the original mounting holes to make them larger and the side of the hole has reached the inner power plane, 2) the PCB designer failed to allow an adequate clearance ring in the internal power plane around the hole, 3) the drilling machine or data table had an error that caused the hole centers to be out of tolerance, or 4) a hole was drilled in a board location that was never designed to have a hole.

In any case, your use of nylon screws should solve the problem.

Sincerely,

Tom Taylor

You really need to pin down the cause. The nylon screw trick is probably avoiding the problem for now, but without a pin something else may rear its ugly head later. Now you can tighten the trouble source by selective use of the nylon in your shop, and you will probably find that only one is needed. Then you can check the PCB for the above possibilities, or maybe something else.