DAQ input impedance low

[K C]

Hello,

I use a PXI-6251 board to measure some signals during a test. Everything was OK until I tried to measure two signals at a high sample rate.

When I measure two signals in differential mode there is no problem when I set the sample rate at 35k or lower. Above this value the voltage at the first channel starts to drop. When I sample the two channels at 500k I found that the input impedance of the first channel is going down to a 140K !! 

I see it but I cannot believe it is true. I tried other channels and the result is the same.

Any ideas what can be wrong ?

[Jennifer O]

Hi KC,

I have tested an NI-6251 here and I definitely do not see that behavior.  If I create a differential AI task on channels 0 and 1 and set the sampling to 500k.  I measured the terminals of AI0 and AI1 and verified that the resistance remains high (exceeds the range of my DMM).

A few troubleshooting suggestions:

1)Verify the resistance on the channel with no input connected (anything connected would change the impedance measurement)
2)Check the resistance from either channel to ground, is it also very low?
3)Does the resistance drop if you scan just one channel, or only when multiple channels are acquired?
4)Have you always experienced this behavior or did this just start happening/is this a new board?
5)Can you read an analog output channel wired back to your analog input correctly.

It sounds like the analog input circuitry could be damaged, this is not expected behavior for the M-Series board.  I recommend visiting ni.com/support to create a service request so that you can discuss the repair process directly with an engineer.

Regards,

Jennifer O.
Applications Engineer
National Instruments

[K C]

Hi Jennifer,

Thanks for your reply.

It's already weekend for me, so most answers on your troubleshooting suggestions have to wait till monday.

But I can answer 3 and 4.

3) I see this only when I acquire 2 channels. With only 1 channel it is OK

4) I only have 1 board. I was building a test with mainly DC measurements. When I started to build a test when higher sample rates I discovered tthe failure.

The power-supply connected to the test-setup is maximal 30V. All voltages to the AI channels are lowered by prescalers. So I would supprise me if something is damaged.

Will get back to you next week with the other answers.

[K C]

Hi Jennifer,

Had a nice weekend ?

If I measure the inputs with a DMM the resistance is OK > 50M

I tried a few things. For this test the configuration is two inputs in differential mode. Both inputs have bias resistors of 1M.

When I connect only the first input I see that only the 1M resistors putting a acceptable load to my circuit.

When I connect also the second input (connected to a circuit with a 1k resistance) I see that the input resistance of the first input drops.  This depends on the sample rate.

Can you explain this behavior ?

[Hany]

KC,

Usually, if there is an impedance problem, it will result in a charge buildup causing a ghosting effect where signal 0 starts showing up on signal 1.

That it works on one channel only but not more than one, then that means the internal multiplexer in the card is the culprit. 

What are the voltage ranges of channel 0 and 1? Are you measuring a 1mV signal on one and 10V on the other?  You may not be allowing enough time for the PGIA (Programmable Gain Instrumentation Amplifier) on the board to settle.

Hany Ayesh

www.ayesh.net

[K C]

Hi Hany,

I know the things you are saying, but that is not my problem,  I think   

In the example I described I use two channels both in differential mode and at maximum scale.

With the first channel I read a 24V voltage through a resistor divider of 39k and 22k reading the voltage across the 22k. So I should read 8.66V. Because of the bias resistors this is 8.6V

When I connect a 1k resistor to the next channel and start to sample both channels at a rate higher then 35k the voltage at input 1 starts to drop during sampling.

At 100k the voltage is 8.38V, 8.08V at 250k and 7.63V at 500K !!!

I tested this with different channels and reversed order. All with the same result.

Message Edited by K C on 06-12-2007 09:39 AM

[ChrisR.]

I'm not surprised by these results. Your source impedance is 14k (39k || 22k), which is a bit high for fast scanning. I agree with Hany that it is a result of scanning. The output of one channel gets transferred to the other by the capacitance at the output of the input multiplexer. The simplest solution is to lower your divider impedance, if possible. And don't scan any faster than you need to. Also, if you're not going full speed, you could allot more time to the higher impedance channel by manually specifying the sample rate. Another idea is to precharge the input mux by sampling the expected voltage, for example by setting one of your DACs to 8.6V and inserting it into the scanlist.

Also you should try to minimize the capacitance on the input channel, as greater capacitance increases the recovery time from the scan-induced charge kick.

Sorry this isn't good news, but it's the nature of high-speed scanning.

Chris

[K C]

Hi Chris,

Thank you for the explanation and suggestions.