03-15-2012 11:26 AM
Hello Sir Andy!
I think we are having a misunderstanding here. I know for a fact that I won’t be getting a sinusoidal pattern representing the values of the instantaneous voltage of a sine wave.
Maybe what I’m trying to consult is how many IQ samples (from the Excel file) are needed to properly represent a 915 MHz CW signal?
03-15-2012 11:49 AM
Each IQ sampe represents the magnitude and phase of the input signal, so technically, the answer to your question is one sample if you know the input is CW/sine wave. A 915 MHz sine wave will look just the same as a 10 GHz sine wave when represented by IQ data, so the frequency doesn't matter.
Be advised that there is a transient in the beginning of the IQ data that is being caused by digital filter initialization. This just means you should ignore the first few samples of a recorded data file. This is not a bug or unexpected behavior of the hardware, but one of the changes I plan to make to this software in the near future is to truncate these samples to remove the transient from the recorded file.
If you didn't know your input signal was CW/sine wave already, you would need to look over your IQ data to see if the mean mag and phase is changing over time. If the phase or the magnitude of the sine wave is changing, which could indicate some type of modulation or frequency offset, your mean I and Q values would change, and the rate at which they change would depend on this potential modulation/freq offset.
In summary, there is no direct answer to your question, and the indirect answer to your question is 'it depends'. After ignoring the first few samples while the transient dies away, each IQ sample represents the RMS magnitude and phase of the 915 MHz sine wave, and what you should be measuring/looking for is changes over time. You can use statitics to measure the mean I and Q values over time and see if there is any change.
Hope this helps.
RF Systems Engineer
03-16-2012 12:35 PM
Thank you for answering my question.
I understand that the phase offset I am compensating through the front panel adjustments is caused by the unmatched cables.
When I correct this phase offset, the IQ values for the two channels become equal.
Since the phase of a sine wave is arctan(Q/I), then I will be getting equal phases for the two channels and that I will have 0 phase difference?
03-16-2012 01:15 PM
The different channels of phase-coherent acquisition will not automatically be aligned themselves, meaning that there is an inherent phase offset between the two or more channels, due mainly to the phase shift of the LO between channels that are daisy-chained. Differences in cable lengths can add to (or subtract from) this phase shift.
To answer your question, arctan(Q/I) would give you the individual channels' phases, and they should be about the same after you do your alignment, resulting in a phase difference of very close 0.
03-16-2012 01:48 PM
Thank you for always enlightening me.
About the NI-RFSA 2.5, I have successfully installed it. However, when I tried playing back, the phase offset appeared to be doubled. That is, if the original phase offset is 7 degrees, this would appear to be 14 degrees when the playback mode is used.
03-19-2012 01:04 AM
I've already figured out what is happening. Everytime I regenerate the recorded signal (through the playback mode), the inherent phase offset of the system adds up to the actual phase offset of the recorded signal.