inaccurate phase in multi-tone generation

[Al-rawachy]

Dear all,

I am using the ni example  for generating multitone arbitrary spacing waveform. I am generating 4-tones at the same time with different IF frequency (Intermediate frequency) spacing 0, 1 kHz, 100 kHz and 1 MHz. These tones have fixed magnitude and phase generating multiple times (36 times) in a for loop (as you taught me how to generate or do for example a frequency sweep). then the result is converted to frequency domain using FFT. 

When I measure the magnitude and phase of the signals I found that the magnitude is correct but the phase is wrong. By the way, I am using only a cable to connect the generator ( NI PXIe-5673E) and receiver (NI PXIe-5663E), in other words, no test set of filters and couplers are applied, thus, no calibration is used.Any idea? Thanks in advance.

[Bob_Schor]

I think I (don't) see a missing wire in the VI that you (didn't) attach.  However, my powers of Extra-Sensory Perception are a little weak today, so please forgive me for asking you to attach the VI that "doesn't work" (my colleagues might also have their ESP not fully functional ...).

If we assume that you generated the waveform correctly, and also assume that you analyzed it correctly, it is possible (gasp!) that the answer you receive is "correct", namely that the phase you record is accurate.  What causes a phase shift?  Why, a Time Delay.  What could cause a Time Delay?  If you are actually generating and receiving signals, I would expect that they are not "instantaneous" and able to "predict" what the next point would be (i.e. there is an inherent phase lag associated with digital sampling).  There may also be shifts in the synchrony of your transmitter and receiver, and of the code(s) that drive them.

What are the phase differences?

Bob Schor

[Al-rawachy]

Dear Bob,

Thanks for your reply. Actually I sort out the problem by referencing the phase to the main signal. 

[Al-rawachy]

Hello Bob,

a New phase issue is discovered. As long as increasing the phase spacing from the main-tone. the phase start distorted. Have you noticed like this before? In the attached post. you can see comparison between the measured and the setting for the plotting circle with different frequency spacing and how the phase distorted when the frequency be far away form the main-tone. Any idea? by the way, all the setting and the information are the same as in the above post. Thanks in advance

[Bob_Schor]

I can certainly see that you are producing interesting pictures.  "What", I wonder, "is being plotted?"  "Where", I ask, "are the data points coming from?".

I like the first plot the best, as the red and blue points seem nicely distributed around the circle.  Whatever code you used to produce that, I recommend you keep using for the other plots.

Bob Schor

[Really, what are we supposed to do with the scant information, and complete lack of code, that you present to us?]

[Al-rawachy]

Bob,

Thank you for your quick feedback,

.The plotting represents a circle by fix a magnitude at a certain value and sweep the phase from(0-360) with 25 points. The data comes from the VSG generator (NI PXIe-5673) through a cable to be received by VSA (NI PXIe-5663). As I mentioned once the frequency spacing get increasesd the phase distorted. 

I used RFSG multitone generation as mentioned in the snapshot of my code.

I am ready for any further questions.  Thanks in advance

[Bob_Schor]

Sigh.  I refuse to look at pictures of code.  I can't manipulate it, edit it, try to run it, clean it up, ...

BS

[Henrik_Volkers]

I assume some samplerate/samplenumber/frequency/phase were coerced by the hardware drivers or the tone generator but not taken into account by the following signal analyses?

Reading the help of the used vis/hardware discover such coersions 🙂  

Some generators need a samplenumber in fractions of 4 sample ... samplerates couldn't be arbitary ...

Just because you can enter arbitary values into the vi, it's not for shure that these values can be used.

Say 2 samples for the generation are missing or added (at a fixed samplerate). No problem at lower frequencies with a lot of samples  but more serious at higher frequencies.. mainly for the phase ...

How does it look like if you use the pure analytical signals? (no hardware/driver involved )

[Al-rawachy]

Dear Henrik,

Thanks for your help.

Regarding the sampling rate. I did know that the multitone generation vi is restrict about the frequency spacing of other tones whereas the (sample rate / no.of samples) should be an integer number and you can use it to generate the fine frequency steps of multitone depending on this number.

Regarding the vi itself . It works fine because I checked the mag & phase of the generated tones and they look accurate. but in term of hardware, there is something didn't be taken in the account. Thanks again