Signal noise cDAQ-9172

[rmichels]

Hello,

I have a cDAQ-9172 receiving analog voltage signals through an NI 9205.  My signals are 4-20mA signals so I'm using 100 ohm resistors to convert them to voltages.  There are 3 signals from a gas analyzer (O2, CO and CO2) and 4 from pressure transduers measuring air velocity.  If you look at the attached screenshot you can see that the gas signals start with no problems and then after a minute or so I start getting massive spikes sometimes as well as constant periodic changes in the signals.  A multimeter at the input into the 9172 shows very consistent signals.  Also, the velocity measurements as you can see are very noisy (the top wind speed chart), which we try to smooth out using the average and then a time average (middle chart with red and white lines).  Shortly after this screenshot was taken, the signals from the pressure transducers just drop to zero.

This whole problem is repeatable.  If I turn off labview and turn it back on, I'll get a normal signal for about a minute, then the up and down signals from the gas analyzer, then velocities drop to zero.  I've tried it on 2 XP-based computers with different cables.

Thanks in advance for any help.  Much appreciated.

Regards,

Richard Michels

[rmichels]

That last figure might be a bit confusing.  Here is a figure cropped to show specifically what I'm looking at.

[Margaret F.]

Hello,

What is your sampling rate and are you using a laptop or a desktop?

Regards,

Message Edited by Margaret F. on 01-06-2009 09:52 AM

[rmichels]

Sampling rate is 1/second and normally I'm using a desktop but I tried with my laptop and it was the same problem.

Thanks

Richard

[Margaret F.]

Hello,

From the behavior you are seeing, I am suspicious that there may be a grouding issue involved here. What kind of configuration do you have (differential, reference single ended, etc)? Also, could you provide a link to the specifications page for the sensors that you are using as well as can you please describe the behavior that you would be expecting? Thanks!

[rmichels]

Hi,

I just wrote out this whole message, clicked send and lost it.  So I hope this one is as useful.

We have differential inputs going into the 9205 unit using 4-20mA signals converted to voltages with 100ohm resistors.  We are mainly using this unit instead of the 9203 because we don't like the common end in the 9203.  We think it causes problems.

Specs:

ABB 3020 Gas analyzer: http://www.abb.com/product/seitp330/4ffe9d9bb68e3defc12570750047cf0b.aspx

Precision Digital PD 684 Propane flow meter: http://www.predig.com/PD684-9/index.php

Omega PX-274 Pressure transducers (4): http://www.omegaeng.nl/ppt/ppt_nl.asp?ref=PX274

We are taking these measurements in a fire research lab to calculate heat release rates.  All we need is the output from the sensors to match what labview shows us, which it does not.  Spikes and weird oscillations and drifts happen seemingly randomly.  We are desperate to get this fixed.  I have to start running experiments to graduate and I've wasted weeks on this.

Any more info you need, just ask. 

Thanks,

Richard

[Margaret F.]

Hi Richard,

This still seems like a grounding problem to me. When configuring a differential measurement for a floating source, you need to include bias resistors. This may be the source of your problem and is in line with the behavior that you are seeing. Please see Table 1 in this link to view a schematic of the correct way to configure differential with a floating source.

You may also try reading the signal in a test panel in Measurement and Automation Explorer (Start>>Programs>>National Instruments>>Measurement and Automation Explorer.) Expand "My System" and "Devices and Interfaces." You will see your devices there. You can right click on a device and select "Test Panels." Use the test panels to confirm what you are reading.

[rmichels]

Hi Margaret,

Sorry I'm so late in responding to your advice.  Thank you for the help.

It says in the link that you sent that bias resistors of 10,000 - 100,000 ohms work well.    Our concern is with our pressure transducers, gas analyzer and flow meter, which output 4-20mA signals.  Table 2 indicates that pressure transducers are considered high impedence.  So we should use resistors anywhere in this range from both positive and negative signal wires to the ground?

Also, we recently found that instead of just taking 1 measurement/second, it eliminates a lot of the noise if we sample at a high rate and take an average.  Just arbitrarily playing with frequencies and numbers of samples, I found that 400 samples at a rate of 2000Hz works fairly well.  Our main problem now is the strange drift that we have.  When the system starts up, we get about 1-2 minutes of data that looks good and then the numbers just drop to strange values. The attached screenshot shows the problem well.  Could this also be due to not having bias resistors?

Thanks,

Richard

[Margaret F.]

Hello,

How are you exciting your sensors? When you begin to see the readings drift, could you connect a multimeter as close to the sensor as possible and see if that is giving you the same readings that the software is? Thanks!

[rmichels]

Using a multimeter directly at the NI9172 shows a great, constant value even when the software (both labview and Measurement & Automation explorer) show a different value.  For example, we have an unwavering 0.4V signal directly into the device,which is what we are expecting, but the software shows 0.4 for about a minute, then drops to 0.39 and stays there with occasional spikes back up to 0.4. 

I'm sorry, but I don't actually know what "exciting my sensors" means.  Might as well be honest.

Thanks,

Richard