Tie unused inputs to AIGND

[kenkar]

I am using BNC-2090 rack mount with my PCI-MIO-16E-1. I am having crosstalk problem. Now I need to tie all the unused inputs to AIGND. Would anyone suggest a way to do without soldering a wire (zero ohm resistor) on the circuit board of the BNC-2090. Thanks.

[John N]

Hello kenkar,

I think there is something you can try. The basic solution for this kind of situation is to try and slow down the channel rate (as opposed to the scan rate) to give your amplifiers more time to settle. You can set the channel rate with the AI Clock Config VI in LabVIEW.

The total input impedance of your channels is probably large and yielding the crosstalk symptom.

If this is the case, here are some documents about the subject:
Amplifier Settling Time
Discussion of different clock terminology

Hope that hel
ps. Regards,

John N
Applications Engineer
National Instruments

[kenkar]

Thanks, John. I have tried to increase the interchannel delay. The crosstalk did allievated to some extent. Please see http://exchange.ni.com/servlet/Redirect?id=5424182
for the background of my problem. The extract of it is as follows:

I was sampling at 5 kHz to acquire 500 samples with interchannel delay set to auto (-1). I found that one of the thermocouple channels had got an unusual step, which coincided with the switching of the slotted switch signal (from low to high and then high to low). I had tried to increase the interchannel delay, the step on the thermocouple signal did reduced in magnitude. The strange thing was that even I increased the interchannel delay to 50 microseconds (the maximum allowed without getting an error from LabVIEW), the
step still existed. The MIO board was supposed to settle to full-scale step within 3 microseconds with +/- 0.012% accuracy. How could my situation be explained?

Most of my inputs are from instrumentation amplifiers. They are supposed to have very low output impedences (fraction of a ohm). However, I don't know whether the output impedence of my slotted switch is large.

[John N]

Hi kenkar,
Looks like some good discussion already on that thread. I'm glad to see it.

I couldn't find in either post the actual order of your channels. Could you post that? If it was only a settling time issue, then I would bet your switch channel is immediately before your thermocouple channel that is off.

However, after looking at the data in your other post, I have a suspicion that there is a ground loop or some other similar problem affecting your thermocouple. None of the other channels seem affected. A good test might be to ground the channel immediately after the switch. Then scan including the offending thermocouple channel. If it's settling time, the problem should be gone. If it's not, then the ground loop or whatever it is will st
ill affect the thermocouple.

Does that thermocouple have a path to some reference to drain amplifier current?

Keep investigating. I bet we'll find the source.

Regards,

John N
Applications Engineer
National Instruments

[kenkar]

The physical connections to BNC-2090 were:
Ch0: TC2 Ch1: TC3 Ch2: TC5 Ch3: RJT Ch4: HW1 Ch5: HW2 & Ch7: SW

The channel order in LabVIEW, more specifically to AI Config.VI was:
SW, TC2, TC3, TC5, RJT, HW1 and HW2.

I have changed the physical connections to:
Ch0: HW1 Ch1: HW2 Ch2: RJT Ch3: TC2 Ch4: TC3 Ch5: TC5 & Ch7: SW

I was hoping by putting the SW channel (the one which was causing the problem) further away channel-wise from other channels and being separated by an unused channel,the problem would be solved. However it did not help. Then I kept using this physical connections but changed the channel order in LabVIEW to:
SW, HW1, HW2, RJT, TC2, TC3 and TC5

This was to put the strong signals (HW1, HW2 >1 V) in between SW and
TC2. Now I cannot detect any crosstalk. Would you explain to me why does the channel order defined in LabVIEW matter?

[PdB]

Sorry, but I would just ignore the contaminated readings during the switching (only 13 samples over a total of 500). I guess it is caused by some current flowing that lifts the TC-voltage a little bit (a TC signal is low-voltage that is amplified by say a factor of a 1000).
Replace these false readings of TC2 for example by the average of the last sample before and the first sample after the switching. It's not pretty but it works.

[John N]

Sounds good to me!

Kenkar, if you want to get rid of the bump in your data, I think you've got 3 or 4 directions to hit it from now. If you figure it out, or find something new I'd like to hear about it.

John N
Applications Engineer
National Instruments