Cross talk with simultaneous sampling (not multiplex)

[hs713]

Does anyone have any experience resolving cross talk issues when using a DAQ with simultaneous sampling? From what I have read, I can see why cross talk arises when using a multiplexer, i.e. a device that has to scan across channels (not simultaneous) to get all the data. I can see how having an input source with too high of impedance could be a problem for the accuracy of that channel (settling time is too long), and if the channels had to be scanned through then there is some residual voltage left, causing the next channel reading to be inaccurate. However, I'm not quite sure how this applies when the sampling is simultaneous, since it's not switching between channels.

In particular, I'm seeing significant cross talk when trying to acquire a square wave signal, presumably because the square wave is composed of a large spectrum of frequencies -- it seems that this causes "blips" to appear on my other channels that correspond with the uptick and downtick of the square wave. I think it does happen with simple sine waves and other types of signals as well, but it seems to be the most obvious/significant at least when using a square wave input.

For reference I'm using PXIe-6368 cards in a PXIe-1078 chassis.

Curious if anyone has any experience resolving issues related to this. Thank you!

[crossrulz]

The crosstalk is likely happening  in your wiring.  AC signals can induce currents on other wires.  Use shielded cables to prevent this.  Having a low source impedence will also help.

[hs713]

Thanks for your message. The BNC inputs we are using are shielded up until the point that they go into a CB-68LPR block. Those blocks are unshielded. The BNCs go into a panel which does have insulating rings, and we have two short wires going to the screw terminals for each channel. I also took the block out of its housing box and attached the BNCs to their screw terminals without going through the panel, and still had the noise issue. 

Do you happen to know of a simple way to add shielding to these blocks/wires to check if that would help?

Also I noticed that the problem extends across cards of the NI chassis, too (i.e. if I put the square wave signal into a channel on a completely different card, I still get the noise on all the other channels on other cards).

I'm not completely sure if lowering the input impedance of the source by using a unity gain buffer or not would help. Since this is a simultaneous DAQ (it's not switching between channels), I don't really understand how ghosting due to high impedance of one source on another channel could possibly occur. The suggestion of adding a unity gain buffer seems to be for mutliplexing devices only. Do you have any explanations on that?