PCI-6115 Digital Output crosstalk

[src42]

I am trying to create 8-bit output signals that are specified by various software-created waveforms that repeat indefinitely. A typical example of such waveform is a 280Hz sinusoid. The configuration is as follows: (1) used Ctr0InternalOutput as clock source; (2) clock rate is 50kHz; (3) connections made with SCB-68 wire-terminal connection box.

I use a Tek MSO 4104 1-GHz oscilloscope (4 analog input; 16 DIO inputs - 8 are used) to monitor the 6115 digital signals, either with the scope analog or digital inputs. I have been able to generate the signals, but the problem I see is systematic corruption of the individual bit levels by other bits. For example, the MSB (which changes less often) shows either HI-LO or LO-HI transients when using the scope DIO input. With analog scope inputs, this is evidenced by lower-amplitude noise spikes on given line. All of these transients are synchronous with the transition edges of other signals.

My guess is that this is some kind of termination or ground loop problem with the 6115 DIO connections. I first tried using a simple ribbon cable (8 lead plus ground). I got some improvement by replacing this with individual RG-174/U 50-ohm lines, connecting each coax ground to the nearest digital ground terminal. Then I terminated the output end of each coax line with 100kOhm and 100pF (parallel) to its ground shield. These latter measures improved the signals, but did not eliminate the problem. I am looking for tips on the best way to configure the DIO connections or condition the signals.

[enderWiggin10]

Hi src42,

When you mention you are outputting the 8-bit signal, this leads me to believe you are using digital output, but you also mentioned that it is a 280Hz sinusoid.  Are you using Analog or correlated Digital output from the 6115?  Are the problems occurent at frequencies other than 50kHz?  Can you post a screenshot of the results your are getting by any chance?

[src42]

Yes, I am using the 8 digital output terminals to convert my software-generated signals into 8 individual bit lines. I have used either low-frequency sinusoids or linear ramps to create the desired bit levels, which change every 20us.

I am using correlated Digital output, with the Ctr0InternalOutput as the DIO clock.

I am testing only at 50kHz, since that is the clock frequency I need for my experiment.

I have been able to reduce the crosstalk by using 8 individually-shielded coax lines, with 50-Ohm series loads at the SCB-68 terminals, and all digital ground terminals connected to cable shields. At the load end of each coax I use a 100kOhm || 470pF parallel filter to the ground braid. This reduces the initial pulse overshoot as monitored with DIO inputs to a Tek MSO 4104 scope. The remaining question is how to connect the terminating coax grounds. Grounding only P0 and P7 with the DIO connector grounds still leaves cross-talk, but adding the P6 ground removes it. What is the best practice here?

[enderWiggin10]

Hi src42,

What happens when you connect all of your coaxial shields to Digital Ground?  Is this a possible solution?  Perhpas I am missing something, but I am not sure why you are only grounding P0 and P7, and not all lines of the port.

Best,

[src42 · Accepted solution]

I wound up grounding P0, P6, and P7. This reduced the cross-talk to acceptable levels. I do not claim this is a unique solution, but grounding all coax shields did not produce as clean results. I found the on-line documentation from NI to be hard to find and inadequate in discussing such issues.

[enderWiggin10]

Hi src42,

Thank you for the follow up, and I will pass the information along concerning the lack of documentation.  We very much welcome and appreciate feedback to improve our products and documentation.  I apologize for the cross-talk and I am glad that the grounding has helped to reduce this issue.

Best,