Troubles with HSDIO NI 6556: Can't generate at high frequencies because of a huge rise time

[Mar1]

Hi, I have currently the HSDIO 6556 and I am encountring a probleme when generating high speed patterns. In fact, I am witnessing 30us rise time for any output.

In the first attachement, you can find an oscillo visualization of a 5Khz toggling  signal generated at DIO0. The second attachement represents a 50Khz generated signal...As you can see, there is no way to go further in frequencies since the single starts to loose its integrity.

I am using the HSDIO1.9 driver and have used MAX test pannel to generate this square signal.

Its very confusing. Is there a special configuration for HSDIO 6556??

Thanks for your help.

[Jesse_O]

Hello Mar1,

There is no special configuration required to generate data at the specified 200Mbps data rate.  The rise time should be much faster than the images you show. Because the PXIe-6556 is meant for high speed communication it is source series terminated with a 50 Ohm resistor.  You can find out more details on this inside the NI Digital Waveform Generator/Analyzer Help under the topic Source Impedance or Termination.

Can you provide more details on your setup?  If your cabling or load is overly capacitive this 50 Ohm resistance will create a low pass filter.  Additionally can you provide more details on your scope, and how you took the measurement?  What is the bandwidth of your scope/probes.  Were the probes set to 10x or 1x?

[Mar1]

Thanks Jesse for your quick reply.

I am using a high-end oscillo (Agilent  MSO9064A, 600MHz, 10Gsps) with an agilent probe (N2973A, 500Mhz, 10:1)

I'm using an NI flying VHDCI cable and making a measurment directly at its endpoints. No termination, no load... A very intuitive setup.

Do you think that termination will solve this problem?

Awaiting for your precious advices.

Thanks.

[Jesse_O]

A 600MHz scope sampling at 10Gsps should be able to show much faster edges that what you are seeing. 

This configuration should be capable of show much faster edges than what you are seeing.  Can you verify a few things? 

1)  Is your scope bandwidth settings set to maximum?  You might have a filter option enabled. 

2)  Can you try a different channel on the PXIe-6556 to see if you get the same filtering behavior?

3)  Can you measure the SMA Clock out signal?  You can export a sample clock to this pin, and it should have fast edge rates.  This would be another way to determine the source of the problem.

[Mar1]

Here is the Sampling clock (10KHz) exported at either the VHDCI connector (PF4/CLK OUT :yellow) and at the SMA Clock out (Green).

Its is the exact same signal exported twice at the sametime...

[AnalogKid2DigitalMan]

Try it without the probes and feed it directly into the scope via a BNC connector with the scope input set to 50 ohms.

-AK2DM

[Mar1]

hi AnalogKid2DigtalMan, the digital IOs are accessible through a 40pin connector. There is no way to connect them directly to the scope.

[AnalogKid2DigitalMan]

Then try putting a 50 ohm resistor in parallel with the scope probe connections. The 6556 requires 50 ohm termination in order to operate within specifications.

-AK2DM

[Jesse_O]

The PXIe-6556 does not require a 50 Ohm load impedance to operate at 200MHz.  The most common use case is to have a high impedance load, which is why the source is 50ohm series terminated.  The measurements you are seeing are not expected.

The next step is to determine if the problem is the probing setup, the cable, or the 6556.

I was unable to find the probe you referenced earlier, so I assume it operates just like a normal passive probe.  How are you connecting the probe GND to the I/O GND?  Do you have a VHDCI to VHDCI cable (i.e. 152870-01) and a 2163 or 2162 that will break out the signals?  Can you measure another I/O line?  Can you swap your cable and try the measurement again?  Has this ever worked?  In the plots above when you measured the SMA Clock out was that through a probe, or directly to the scope?  Can you use the same probe/channel to measure it?

[Mar1]

Hi Jesse, I am using both NI SHC68-H1X38 and NI SHC68-C68-D2 (+CB-2162) . Both of them gave the same result. I tested different IOs, I've tested different probes (even 3.5GHZ active probe)... The rise time is unbelivebly high... Its like there is a big capacitor coupled at the 40pin connector outputs.