Need help troubleshooting M7264

[shirsch]

My Heath H-11 is fitted with a quad-height M7264 LSI-11 CPU board holding 4k words of memory. It's operational and passes all the xxdp diagnostics. While troubleshooting issues that appear with Q-Bone in the system I discovered a very noisy BIRQ7 signal on the backplane (see scope photo below). While this causes no problems with the CPU (it's simply dead-end terminated on the board), the bigger question is what's causing it. There were no other boards in the system at the time that trace was grabbed. The +5 supply rail is quite clean (see scope photo) and ground is equally so. Since there is nothing driving the BIRQ7 line where in blazes are these spikes coming from? I have verified that both termination resistors are showing correct values. Anyone have thoughts?

 

[AK6DN]

How are you probing? Is there a long ground lead on your scope probe? That can cause a false reading from induced noise in the ground loop.

 

[shirsch]

How are you probing? Is there a long ground lead on your scope probe? That can cause a false reading from induced noise in the ground loop.

Good point. The ground lead on my x10 probe is about 6" long and I had another foot of clip-lead to extend it to the power supply ground point. After reading your note I wrapped the ground clip through a ferrite loop about 5 turns to add some filtering. Made no visible difference. Do you have a suggestion for how I can make this measurement with more assurance? Keep in mind that I grabbed the ground and +5 traces using the same setup. Also, none of the other dead-end terminated IRQ lines show any sign of this spiky noise.

 

[AK6DN]

Try and use as short a ground lead to a ground point near where you are probing (ie, an adjacent backplane ground pin). If if it a high impedance probe (ie, 1M to 10M) input it will be very sensitive to induced noise. Another possibility is to use a low impedance probe setup using just a length of 50ohm coax attached to the probe point and a close ground. Then set the input to your scope on 50ohm instead of high impedance.

 

[shirsch]

Try and use as short a ground lead to a ground point near where you are probing (ie, an adjacent backplane ground pin). If if it a high impedance probe (ie, 1M to 10M) input it will be very sensitive to induced noise. Another possibility is to use a low impedance probe setup using just a length of 50ohm coax attached to the probe point and a close ground. Then set the input to your scope on 50ohm instead of high impedance.

I forgot to mention what started this entire easter-egg hunt: I built a Q-Probe (http://retrocmp.com/tools/qprobe) to help in system diagnostics and noticed the IRQ7 LED was on almost all the time. It has a pulse-stretching circuit and this is certainly being triggered by the BIRQ7 spikes getting down to a logic 1. So, no external wiring at all in that case. All connectors have been cleaned off and treated with De-Oxit, power supply plug ditto. As I showed above, +5 and ground are clean a whistle. Due to the on-board termination in the CPU, the impedance of that line is quite low and directly induced noise seems unlikely. To be clear, the Q-Probe is not causing the problem as it was not in the system at the time I took that scope picture. Given that the entire circuit consists of a two-resistor voltage divider between +5 and ground, I cannot fathom where these spikes originate if ground and +5 are clean.

UPDATE: Grrrrrr.... The blasted extender board was picking up noise from the CPU. Never even thought about it since it's a dumb piece of PCB with connectors. If I move it to the furthest slot things get much quieter. Thanks very much for getting me thinking in the right direction! Since I often need to use the extender in a crowded chassis it may be worth fabricating a shield for it.

 

[AK6DN]

Does that extender board have an embedded ground plane? Or is it just traces between the connector and fingers?
If the latter, it should be possible to add an external shield using some one sided PCB material cut to the right shape and grounded in multiple places (otherwise it becomes a big antenna too).

 

[shirsch]

Does that extender board have an embedded ground plane? Or is it just traces between the connector and fingers?
If the latter, it should be possible to add an external shield using some one sided PCB material cut to the right shape and grounded in multiple places (otherwise it becomes a big antenna too).

Just a double-sided board with traces on back and front - no ground plane. I don't have a suitable piece of PCB material here, but I probably can sandwich a piece of aluminum foil between two sheets of thin plastic and jerry-rig a workable RFI shield.