45 Hz noise

[caz]

I trying to make an acoustic (pressure) measurement using the PXI 4496 and I don't fully understand why my signal looks the way it does. Here is a graph of the time signal with 3 channels and also the power spectrum of the pressure channel.

http://forums.ni.com/ni/attachments/ni/100/1989/1/acoustic_channels_test_FP.png

the white trace is the pressure channel, the red is a IEPE accel. and the green is a disconnected channel and both tranducers have excitation to them. However, neither transducer is seeing any changing phenomena. That is, the sensors are just sitting there and you can see a large ~45 Hz signal on the pressure channel. For this measurement, I'm using a kulite XTE-190 transducer that has a FSO of 100 mV and you can see that I'm getting signal at 25% of the FSO without doing anything. My wiring configuration goes like this. A 4 conductor shielded cable brings the 10V floating DC excitation to the transducer using 2 conductors and the bridge output of the transducer is brought back to my PXI 4496 using the other 2 conductors. Right now I don't have the shielding tied to ground anywhere.

Message Edited by caz on 02-26-2009 01:55 PM

[caz]

Heres the pics from above. I got timed out in editting my post. Anyways, any comments or suggestions would be helpful. Also, I wonder about the 10 V DC power supply and floating its output. Should this be ground referenced, or is the 4496 ground referenced somewhere. The accel. channel looks fine and that one uses a standard 50 ohm BNC and constant current excitation provided by the 4496. So my pressure channel either has issues with the transmission line or ground loops or something else I might be missing.

 

Message Edited by caz on 02-26-2009 02:15 PM

[caz]

In case anyone is wondering why the signal looks the way it does, its due to a noisy DC power supply providing the 10 V excitation to the transducer. Since the 4496 is pseudodifferential and not truly differential, this common-mode voltage is not canceled. The following figure shows the difference between differential and pseudodifferential measurements. The visible frequency in the black trace is the signal of interest which is a lot easier to see in this mock differential measurement versus the pseudo measurements.

Message Edited by caz on 02-27-2009 05:09 PM

[Jason_D]

Caz,

Have you verified that these components work correctly sepereately in the Test Panels of MAX?

I would try grounding that signal, I don't expect that to be the entire cause of the problem, but I don't think it can hurt.

[caz]

Yes I verified these components in MAX.

I have a cDAQ chassis with 9205 board that I tested this transducer output with. Setting the measurement type to differential and sampling at 125 kS/s, I get a signal that looks like the black trace in the figure below. This black trace was captured by measuring the + side of my transducer with a channel 0 of the 4496 and the - side with channel 1 and then subtracting the 2 channels in software. It appears that the problem is related someway to the pseudodifferential nature of the 4496 and not canceling the AC portion of the common mode voltage.

Message Edited by caz on 03-02-2009 08:14 AM

[dousley]

Hi Caz,

As Jason mentioned, it's a good idea to ground your shield. But only ground it in one place: at the transducer, or at the device. If the shield is not tied to anything then it is floating and essentially does not function as a shield anymore.

You have a point about common mode rejection. It does appear that the 4496 is not rejecting the common mode signals in the excitation signal. I'm curious about the excitation signal you are using. You said it's a noisy DC source. Is it stepping? The AC-coupling filter on the 4496 has a settling time in response to DC steps. If you were to look at an acquisition on a 4496 during a step it would look like a spike with an exponential decay curve afterward, kind of like what you have in your acquisitions. It would be great if you could get a hold of a DC-coupled instrument to monitor your excitation and your bridge output and post that so we could take a look at it. 

I'm also curioius about your configuration.

1) Is your excitation source floating?

2) Could you draw a diagram showing where you are connecting signals from your bridge 

    a) When you have the huge noise in the measurements of the 4496

    b) When you are doing the differential measurement with the 4496

3) What is the DC voltage present on the negative (shell) of your 4496 inputs during these acquisitions?

Daniel

Conditioned Measurements Hardware

[caz]

Hi Daniel and thanks for the message.

I think I've figured out my problems. The main thing to note is the 4496 and these unamplified Kulites don't play nice together. That is, the output of the kulite is purely differential and the 4496 is single-ended. Line conditioning is needed. What I've done to improve the situation is put an analog devices instrumentation amplifier ahead of my 4496 to convert from differential to single ended and amplify the small signals coming from the full bridge transducer.

Also, some of that business in the plots I showed was due to ground loops. Taking all my grounds and tieing them together at a single location helped some. The power supply is little noisy but not as noisy as the plots.

I did look at this single on a DC coupled device (NI 9205) but it was in differential mode so the signal looked nice and clean. 

Thanks,

caz