NI USB-6259 BNC DAQ: analog input signal cross over issue

[veritas]

Hello,

       I'm using the NI USB-6259 BNC DAQ to aquire an analog signal from four channels, and i'm having an issue with one signal affecting the others. The circuit I'm running is the following:

I have a wire connected to a battery(two AA at ~1.5V), which then feeds to the signal wire of a BNC cable, that feeds to an analog  BNC channel of the DAQ. The ground of the BNC feeds to another wire, that is attached to a conductive plate.  The idea is this: when I touch the wire connected to the battery to the metal plate, I complete the circuit, and thus I get a binary step from nothing to around 3V.  When I tested this with an osciloscope using two channels (each ground connecting to the metal plate) i get indepedent steps whenever i touch either of the wires from the respective batteries to the metal plate (i.e. it works as expected).  However, when I use it with the DAQ, whenever i touch one wire, I get a response in all the other channels (the other 3), even though their respective wire is not touching the metal plate.  

Does anyone know why this is happening, and how I might be able to stop this "cross-talk"?

Thank you,

veritas

[NIquist]

This "ghosting" is actually normal behavior with the multiplexed DAQ cards.  It won't really hurt your measurements when you have active signals attached to the other channels.  For best results though you can put an extra channel between your actual signals and ground the inputs.  It also helps to avoid putting large signal inputs close to low-level signals in your scan order.  If you do you'll need to slow down the time between channels to allow the sample and hold circuit to settle.

There's some good documentation on this somewhere but I foget where right now.  I'll post it if I find it.

[NIquist]

http://forums.ni.com/ni/board/message?board.id=170&message.id=159690&query.id=90519#M159690

http://zone.ni.com/devzone/cda/tut/p/id/2825

Also see the M-Series DAQ user's manual. 

[veritas]

Thank you for the quick reply.

Well, the "signal" I'm looking for actually is the occurrence of a completed circuit.  That is, I want to detect when the wires touch the metal plate, and I need to have each channel independent. So, when I touch down one wire, and I get a step to 3V in all the channels, this definitely is confounding my desired signal.  When I run two channels through the oscilloscope, it works fine (i.e. each channel with step to 3v completely independent of each other), so "theoretically" it should work.

i'm a little confused about this statement:  "put an extra channel between your actual signals and ground the inputs".  By "between" do you mean don't use neighboring BNC channels on the DAQ?

thanks.

veritas

[NIquist · Accepted solution]

I see and you're right.  That application will have trouble with the crosstalk.  Fortunately the grounded channel trick should help you. 

Configure your DAQ to collect twice as many channels as you need.  Connect your wires to the odd channels and short out (ground) the inputs of the even ones.

Now when you scan the channels there will still technically be crosstalk but it will come from a grounded channel so there won't be anything to interfere with your measurements.

At least that's the theory.

[veritas]

Well, this time, theory has transfered nicely into practice!   Your solution seems to work very well. I get minor bleed over between the signals, but definitely will not confound any thresholding that will greatly clean up the signal.

awesome! thanks!

veritas