How do I measure the phase difference between an acquired waveform and a generated waveform?

[arvmenon]

Hello everyone, I will try to give you an overview of my setup and what I am trying to accomplish and maybe somebody can help me figure out a solution. Sorry if this is too detailed.


In brief this is what I am trying to do. I have a flame inside a combustion chamber. The flame oscillates at a regular frequency. I can get this frequency by way of a photo multiplier tube aimed at the flame, which measures the flame radiation.

My goal is to obtain phase-locked images of the oscillating flame using an ICCD camera, triggered using a pulse generator, which is triggered by the voltage signal from the PMT. I can image the flame at different phase locations by changing the delay on the pulse generator.

Now, the flame is longer than the diameter of the viewport. So to image the entire flame, I have to move the burner down and image the flame in sections.

Using the different images I can recreate the entire flame. But the problem with collecting radiation from different regions in the flame is that the measured signal will have a slight phase difference between different flame regions. This phase difference has to be compensated for on the pulse generator delay times. After the lengthy intro, the question.

How can I measure the phase difference between the two waveforms obtained at different flame regions?

My line of thinking is to generate a signal using Labview at the exact frequency of oscillations and phase zero; and then get the difference in phase between these two waveforms.

I used the tone measurement VI (Extract Single Tone Information 1 Chan) to get the frequency of the incoming signal and used this fr...

[jpaul]

Hi Arvmenon,

Since your flames are oscillating, could you take any offset out of the waveforms and then compare zero crossings of the sinusoids to determine phase difference? Let me know if this is a potential solution and we can investigate it further.

Cheers,

Jonah

Applications Engineer

National Instruments

[johnsold]

How stable are the oscillation frequency and the phase shifts? Is it even meaningful to think about the phase measured at time t1 and position z1 compared to the phase at time t2 and position z2 after you have moved the burner?

To make phase measurements you need to have a timing reference which does not change as the other parameters (the burner position) vary.

Is there any way you could put in a prism or mirror so that the PMT which generates the timing reference always looks at the same part of the flame? Or a fisheye lens in front of the PMT so it always sees all the flame?

Lynn

[arvmenon]

Hi Lynn,
The oscillation frequency is more or less stable. At this point, however, I do not know how stable the phase shifts are.  I did consider using fisheye lens, but I have optical access limitations. And you are right on point when you say that  it might not be meaningful to compare the phase of at time t1 and position z1 to another time and position.

This is what I am thinking. Use two PMTs aimed at different heights in the flame and acquire both signals simultaneously. Now I am using a PCI-6221 which doesnt have simultaneous sampling capabilities. I am considering using a PCI-6143 with an SCXI 1125 (for amplification) on an SCXI 1000 chassis. I think to even compare the phase you have to do the acquisition simultaneously.

What do you think?

[johnsold]

Dual PMTs sounds like a good choice.

You indicated in an earlier post that the oscillation frequency was about 15 Hz. If you sample both channels, non-simultaneously, at 1 kS/s on each channel, then the worst case timing error between the two channels is 1 ms. This represents a phase error of 5.4 degrees at 15 Hz. The actual difference will likely be half that much or less. Thus, within the limits of the DAQ device maximum sampling rate you can determine a limit on your phase error. The 6221 has a maximum aggregate sampling rate of 250 kHz. If you sampled 2 channels at 100 kHz each, your phase error would be lass than 0.06 degrees at 15 Hz. I seriously doubt that your flame is stable enough that phase errors of that magnitude would be an issue. So, I do not think you will need true simultaneous sampling for this application.

Lynn

[arvmenon]

Lynn,
Thank you very much for the suggestion. You are right... for this application I might not need a simultaneous sampling board. But it does pretty much make my board useless for other channels which would also be in use at the time.

I suppose there is no way I can run a data acquisition process on the same board at two different rates is there? 🙂

 Anyway, for the time being I am going to try your suggestion as a proof of concept.

Thanks,
Arvind.

[johnsold]

Arvind,

No, you cannot run two different rates at the same time. The board has a single A/D converter and sample timing clock which are shared across channels.

Typically all channels are scanned at the fastest rate required for any one channel and then the excess data for the other channels is ignored.

So you need to divide the maximum sample rate by the total number of channels and see if that will give you a low enough phase err

You can put more than one board in your system and run the two boards at different rates.

Lynn

[arvmenon]

Lynn,
I was going to use two boards. Incidentally, I have another 6221 in my lab which is currently not being used. I will give that a shot.

Thanks,
Arvind.