LabPC analog input estrange behaviour

[Garcas]

Hi,

I have a LabPC board and I have found ?an estrange? behaviour when I connect an analog input signal to one channel (e.g.: to channel 0) and I leave the rest of channels open. I know this is not the way to do it, but I would like to know what?s happening !

HARDWARE CONFIGURATION: RSE Mode, Bipolar Input, Gain=1

PROBLEM DESCRIPTION
EXPERIMENT 1: I connect an analog input signal (3 Vpp 1 Hz sinusoidal) to channel 0 leaving channels 1 through 7 disconnected (open). I open MAX, CVI or LabVIEW and acquire the signal on channels 1 through 7 and I measure in these channels a signal, which is the signal in channel 0 modulated with a 60 Hz sinusoidal (line signal). Of course, if I ground the corresponding channel the measured s
ignal disappears.

EXPERIMENT 2: I connect an analog input signal (3 Vpp 1 Hz sinusoidal) to channel 0 and another signal (1 Vpp 4 Hz square wave) to channel 1, leaving channels 2 through 7 disconnected (open). I open MAX, CVI or LabVIEW and acquire the signal on channels 2 through 7 and I measure in these channels a signal, which is the signal in channel 0 modulated with the signal in channel 1 and a 60 Hz sinusoidal (line signal). Of course and once again, if I ground the corresponding channel the measured signal disappears.

QUESTION: I suppose this must be some kind of crosstalk, but with this frequencies ? I have studied the analog input stage of the LabPC but I haven?t found any explanation. So what?s happening ?

Thanks a lot

[Ben]

Cross talk yes.
Not in the wiring, but in the inpute detection circuits. If you had the schematics or block diagram of the board the answer would probably be obvious.

Let me do some speculating. An "open" input would appear as a high impedenece output. With sufficiently high samples rates, the capacitance of the input MUX, sample and hold circuitry, or A/D converters would serve as low pass filter. So waht is happening is the voltage read from the connected channels, does not have time to bleed off before the un-connected channels is sampled.

In general, I usualy do not sample channels that are not connected to any thing (:<)).

Ben

[Annette Perez]

Garcas,

This behavior is normal. If you want to measure 0V on all the other channels, you need to hook them all to ground by using a resistor.

What happens inside the board is that there is a capacitor in between the multiplexer and the amplifier. Everytime you read a voltage, this cap gets charged with it, but if the next channel doesn't have any signal connected to it, the cap will start leaking but this is way too slow compared to the speed at which we are scanning; this is why we see this behavior.

[Garcas]

Thanks for your answer, but I can not agree with it. I would if I would be scanning through the different channels, but this is not the case. May be the problem is that I didn?t express myself properly. I hope the attached figure clarifies the situation.

I have connected 2 signals in channels ACH0 and ACH 1, but I acquire only from channel ACH 2, which I left opened. In particularly I have acquired 100 points at 100 samples/s using LabWindows CVI?s DAQ_Op function from the Data Acquisition Library.

In this case, the only signal path I can think of is due to the multiplexer?s off channel impedance. If we assume that an unselected channel presents an impedance, in this case channels ACH0 and ACH1 are unselected, this impedance would generate a crossta
lk effect with adjacent channels due to their source impedance. What surprises me about this, is that we are talking about low frequency signals (1Hz and 3 Hz) and huge crosstalk levels (see measured signal in the attached figure).

Thanks again

[Ben]

If you kill the signalsto the other channels (0 & 1), the cross-talk goes away (I suspect).

If you put a 500 kohm resistor across ch 2 to gnd, the "cross talk" will be reduced (if not eliminated).

The cross talk is not due to EM wave propgation. I suspect the extremely high impedance of the DAQ channels is approaching that of an open circuit.

Does the magnitude of the cross-talk vary with humidity? I suspect it does.

Just trying to help,

Ben