Zeros within I &Q data at low power level

[thoalfukar]

Dear All,

I am trying to check the voltage samples that VSA measures at the high and low power level.

the number of samples per measurement is 10000 samples. I found the distribution of the data is very good as shown on the attached pictures but I can't find an explanation for the absolute zeros I am getting within the I&Q data at the very low power level. 

Actually, I am PhD student and I am trying to find an explanation for that. 

Please if you need further explanation am ready to provide, and I will be appreciated for your help.

Thank you in advance

Zulfikar

[thoalfukar]

I wish if there is someone can help with this, or at least tell me where I might information about that. 

[Henrik_Volkers]

It's not clear to me what you did.

Sample the data and made a histogram?

When you say IQ signals .. one channel??

My first guess is a DC offset in the DAQ...

[thoalfukar]

Thanks Henrik for reply, 

actully I am doing very simple analizing to the measured samples using VSG. 

I am doing measurement for two power level once at high power level for example (5 dBm) and the other one at the noise floor for example ( -100 dBm) 

Then I check the I and Q samples ( in this case I measured 10K samples at sampling rate 10MHz/sec)

for the high power level The I and Q samples are fine as you see in the attached picture

for the low power level I found there are many samples goes to zero, and I am not sure if that some thing normal or there is a problem.

Regrding the histogram, I did it just to show me the frequency of each voltage and which one is the most common voltage within the samples.

I wish that make it clear for you what I am doing, and please tell me if you need further information. I realy need to understand this case and the right explination for it.

Many thanks  

[Henrik_Volkers]

I still didn't understand what you want to measure...

If both histograms are the amplitude distribution of 10k samples (1ms) you measured, I would say both show about the same noise distribution (about the same standard deviation) .. one with a -6mV offset... maybe the 1ms right at a line hum minima...

How does the signal look like (in time domain) if you sample 20ms (or one line periode)?

How does it look ike in the frequency domain?

[thoalfukar]

OK let me explain in more detail what I am doing and why I am doing this observation.

I am working on developing a new measurement system. This system configured by two chassis. The signal is generated by a VSG (PXIe-5673E) and the received by VSA (PXIe-5663E). I use the List mode for signal generation.

By using the List mode configuration, the acquisition data for more than one measurement would be a little bit critical as there might be an overlap between the samples of the successive measurements as a result of the imperfect synchronization between the two chassis.

For that reason, I had to check the measured samples (I and Q), and as you know if there is any overlap between measurement would be reflected on the distribution of samples which luckily, we don’t get it as the synchronization very good.

While I am doing this observation and plotting the samples, I noticed the too many zeros at the very low power level. If you look to the Q-data for example at the low power level, the number of the zeros is about 1000 zero and nearly the same for the I-Data

I don’t see any clear explanation for those zeros (absolute zeros), so I asked here whether that normal thing or there is a problem. Knowing that the signal has got the 1GHz frequency and the sampling rate being used is 10MHz/sec.

Many Thanks

Zulfikar 

[Henrik_Volkers]

Still no idea what you want to measure.

OK you show the magnitude distribution histogram... what does it tell you ?

My interpretation of such a histogram distributions is explained above. 

How does the signal look like (in time domain) if you sample 20ms (or one line periode)?

How does it look like in the frequency domain?

If you look at an AC signal you will always have zeros unless you have a DC offset 🙂  

A distribution like yours (nealy perfect gauss distribution) usually is created by noise... (with DC offset ... )