NI 9234 phase

[BillyM]

I have an NI 9174 compact DAQ with an NI 9234 and an NI 9269.

I am measuring two accelerometers simultaneously with the NI 9234. I am receiving proper waveforms and RMS values, but I am having trouble measuring a proper phase.

I am not looking for measuing a phase difference! I can do that already, and that's not an issue (and not really valuable information). 

I am looking to measure the phase for one (or each) of the waveforms provided. The phase of an accelerometer is important.

While I can read a phase measurement (I have tried a couple of ways), I have not been able to figure out how to obtain a reliable phase measurement. The NI Calibration documentation says to use the "Extract Single Tone Information.vi", but this puts my phase all over the place. The waveform is rolling, and so every measurement is varied. Using an FFT block provides a more constant value, but it is similarly varied based upon where it thinks the waveform is being graphed.

The NI 9234 is an IEPE Module without built in analog triggering capabilities, so I am getting a little confused on this one. Any thoughts/examples?

Thanks,

Billy

[Chris-G · Accepted solution]

Hi Billy,

I'm not sure exactly what you're looking for, but there are a few VIs designed for phase measurements.

Do you have the Sound and Vibration Toolkit? If so, this One-Shot Gain, Phase, & Distortion Measurement might be what you are looking for. The SVT Gain and Phase VI or Amplitude and Phase Spectrum VI may do it. Again, you may not have these though depending on the toolkits you have installed.

Hopefully this helps,

Chris G

[johnsold]

Billy,

Your question does not make much sense, unless you have omitted some key information.

Phase is always measured againt some reference time point.  You say you do not want to measure a phase difference, so what is your reference?  How is the acceleration produced? Is the frequency fixed or variable?

Extract Single Tone uses FFT methods internally, so it will not differ much from doing the FFT yourself.

Please provide enough details that we may be able to suggest something which can work for you.

Lynn

[BillyM]

Hey all, I solved this yesterday afternoon. Sorry for not getting back to everybody.

I'll explain what my original thinking was more clearly. I had an accelerometer and a reference. But as I measured continuously, the data (i.e. the acquired waveforms) were rollling across the graph plot. More specifically, The rolling sine wave at 100Hz made the Single Tone phase value roll (as well as the basic FFT phase block phase value). All it measures is the phase on the waveform plotted at that specific time. This is why a reference is needed, you are absolutely correct. Even still, I found the results not easy to compare when the values are constantly rolling. I wanted a better visual way with more favorable numbers to work with. Namely, if I had an accelerometer reference with a velocity sensor, I should see a 90 degree phase difference. (Then, 180 actual phase difference)

I was trying to remove the reference, and trigger the result waveform so that I could measure the initial phase difference. The downside of tracking the initial waveforms generated is that the signal sent to a shaker/sensor output may not be settled, even with a buffer.

Soon after posting my phase question I realized that the triggered single waveform would always be at a phase of 0 - it's starting point, which has been pointed out here as well. Without a reference to compare with, there is no telling how far off/close my phase is. So, I abondoned that original post, realizing it was pretty silly. But I do appreciate the help. If I checked back earler, I'd have received most of my answer.

And while that Single Tone vi does perform an FFT analysis, it doesn't use a preferred window for my analysis. There are specific windows that have the ability to better benefit results pertaining to either amplitude (Flat Top), frequency (Hanning), or Frequency Response (Force Exponential).

Anyways, to solve the rolling phase issue I used a provided software trigger I found in NI's examples. The link is here:

http://zone.ni.com/devzone/cda/epd/p/id/3438

If you're wondering, I didn't require the Sound and Vibrations toolkit like it mentions there. It works just fine with the LabVIEW 2011 Developer Suite.

The important vi's of that is the overall SW Analog Triggering_DAQmx.vi and Find Threshhold Crossing.vi (subvi of that).

This let me trigger my analog reference signal's 0 crossing. Then I could get the proper phase of the accelerometer/velocity sensor in question.

At first thought, I was concerned that this doesn't differ against which point of the sinusoidal waveform is crossing 0. I was thinking of tracking the up swing vs. down swing. But, as I soon thought to myself, it doesn't matter, because now I can just worry about the phase difference, which was already working fine. Well, mostly fine. If somebody out there is interested in a solution for this, check out this posting here:
http://forums.ni.com/t5/LabVIEW/looking-for-zero-crossing-point/td-p/688282

While acquiring the single tone phase values and subtracting did work, I had the Sound and Vibrations toolkit (an old version - 3.0) at my disposal. I ended up using SVT Gain and Phase.vi. I found this to be mildly more accurate than subtracting the two values. If you delve deep enough in the subvi's, you'll find that it uses a 7 Term B-Harris window. While Hanning is prefferred, it will still filter the results well, and I will receive my phase pretty accurately.

Thanks for taking the time to provide some input.

Best Regards,

Billy M.

[BillyM]

If people are like me, they like diagrams more. Here's the part of my code that deals with Phase and RMS Amplitude output

.

[Henrik_Volkers]

The extract single tone vi work with a windowed FFT, however if you take more than 20 periods to measure the main source affecting the uncertainty is the noise of your signal. Some slight improvements can be done by using SAM (Sinus approximation with LMSE) .. However differences will be smaller than your over all uncertainty 😉

I posted a vi to play with phase measurements including noise here

PS: Still don't understand why you used the trigger..... unless a stable visible graph to detect visually if something went wrong.

PPS: If you use charge amplifier, what type?

[BillyM]

As to why I use it, you hit the nail right on the head. When looking at the two signals, I like the triggered plot more. I can quickly verify if the test I'm measuring is reasonable, and plan how to best average my RMS values based upon how much is captured in my sample rate. While I just tend to average the entire waveform, I could just use a specific amount if need be.

The trigger is also nice in general for debugging purposes. Is my waveform clean? How is my sample rate? Are my values as expected? Is it really acquiring the signals simultaneously? Did I remember to set the channel excitation on? basic, quick glance items.

We have a couple different charge amplifiers here, but overall - I don't need them unless I'm going to measure the charge of a sensing element. The piezos are conditioned with a PCB, so all I need is a BNC port to provide the 2mA excitation power and measure the output directly. I'm using sensors from CTC. I am very happy with them.

One charge amplifier we have is a Bouche Labs 2113F. They are a small company with a shaker that's great for finding resonance. Unfortunately, their website is pretty abysmal and out of date. I find their charge amplifier to be pretty good, but it has a lot of output noise compared to other models. Really, I just use this with their Shaker.

Another is B&K 2635. This one is not nearly as simple to use, but it provides a cleaner signal that I highly appreciate. This is the one I use the most.

- Billy

EDIT: Thanks for that link Henrik, I'll take a gander at that after lunch! 🙂

[Henrik_Volkers]

I like the 2635 too 🙂 

I wonder why you stick with RMS calibration. RMS will include all the noise and hum ... ( something nice for the old analog meters)

If you can sample fast and accurate enough why not use SAM or FFT tone detection not only for phase but also for amplitude  ?

OK might be customer request 😉

[BillyM]

It's really due to the calibration data I have on hand. I have nominal sensitivities (in mV/g) at 100 Hz as a base.

I suppose when I start delving into noise and uncertainties, I'll be agreeing with you pretty quickly.

- Billy