01-24-2012 10:21 AM
"SMT" in the missing VIs stands for Spectral Measurments Toolkit. The NI Spectral Measurments Toolkit should cover the missing VIs.
<Joel Khan | Applications Engineering | National Instruments | Rice University BSEE>
01-24-2012 10:29 AM - edited 01-24-2012 10:30 AM
The Acquisition example uses functions from the Spectral Measurements Toolkit (SMT), which comes with NI's vector signal analyzer products.
Modulation Toolkit is another software dependency for the acquisition example, and the Mopdulation Toolkit comes with both our vector signal analyzer and vector signal generator products.
Here is a complete list of dependencies for these examples as listed on the software download page:
RF Systems Engineer
01-25-2012 09:28 AM - edited 01-25-2012 09:30 AM
Hello! Thank you for your response! I have successfully generated and acquired signals, particulary arbitrary signals. However, their I and Q waveforms are constantly 'moving'/changing amplitude. How can I generate I and Q waveforms with constant amplitude? I would also like to verify what you said in one of your replies in this thread that the phase offset in the code does not affect the I and Q data/waveform? I'm still confused Sorry for repeating my question Thank you so much for your time!
01-27-2012 05:12 PM
Hello, Sir! This is about the 'NI RF Phase-Coherent Multi-Channel' Generation VI. I would just like to ask if the phase offset button is the only way to compensate for the delay of the other I/Q vector modulator since the LO is connected in a daisy chain manner. I mean, the first I/Q vector modulator would be generating/operating ‘earlier’ than the second I/Q vector modulator so that there is an inherent phase delay/difference between the two. Thank you!
01-27-2012 06:04 PM - edited 01-27-2012 06:07 PM
What you are seeing is expected behavior, and requires a little explanation.
To start, assume you have a single RF generator and a single RF analyzer that are both locked to the same reference clock source. Locking Tx and Rx to the same ref clock source removes any frequency offset between the generator and analyzer. This is a requirement because frequency is a measure of how fast phase is changing - frequency is the first derivative with respect to time of phase. A frequency offset is the same as a constantly changing phase, so the first thing to do is make sure you are frequency locked.
Next, imagine the analyzer operating in a continuous acquisition mode, where it is continuously acquiring 'gap-free'data. In this situation, you are acquiring a sine wave with the analyzer that has no frequency offset relative to the generator, and therefore, the acquired data will show no phase change as a function of time. The I vs time and Q vs time plots would be unchanging horizontal lines, and the I vs Q plot would show a point with a constant magnitude and phase angle.
One important point to stop and think about though, is that this phase angle is not going to be 0 degrees (unless by chance it happens to end up at 0). Phase is a relative metric, and in a single channel Tx/Rx case, ask yourself what you are measuring when you measure the phase of a sine wave. The phase of the sine wave is constantly changing, so measuring the phase of a sine wave depends on when you start your acquisition. There is no such thing as absolute phase, phase is a relative metric.
Next, imagine the same setup except instead of continuous IQ data, you are acquiring multiple single-shot acquisitions over and over. Each acquisition will display the phase of the sine wave depending on when you start the acquisition. Every time you acquire a new set of data, the measured sine wave will be at a different phase and this is what will be displayed. This is why you are seeing movement in your I and Q waveforms vs Time. Each new acquisition results in each channel starting its acquisition at a new point in time of the sine wave. This is why you always still see horizontal lines on the I and Q vs Time plots, but over time these horizontal lines are changing the Y axis point at which they have 0 slope. If there were some kind of frequency offset, you would see some slope or sinusoidal change in the I and Q vs Time plots, and some curvature (like an arc) in the I vs Q plot.
You can try this out for yourself. If you set your IQ Rate to 10 kSPS, and the samples to acquire to 1,000, you will acquire 100 ms records. While running with these settings, you will see horizontal lines in I and Q vs Time (that move around from acquisition to acquisition) and a point in the I vs Q plot that has a constant magnitude but changing phase angle. Next, change the sample to acquire to 10,000, increasing the record length by 10X to 1 second. Notice there is no change in the slope of the I and Q vs time traces, and no increase in any arc length of the I vs Q point. Increase it again to 100,000 for a 10 second acquisition and see the same thing. If there were a frequency offset, you would see essentially a circle plotted in the I vs Q plot, and sinusoidal traces for both I and Q in the I and Q vs Time plots. By the way, any arc you see in the I vs Q plot in this case is showing you phase noise...
So, you will only measure the same phase value in a single channel situation if you synchronize the start of your acquisitions with the same phase of the incoming sine wave (resulting in non-moving I and Q vs time plots, and a point in the I vs Q trace that doesn't move).
Since we're talking phase-coherency here with multi-channel use cases, what is important is the phase relationship between each channel. This should stay stable in a phase coherent system. If you plot phase delta vs time, or a combined mag/phase delta on a polar I vs Q plot, these will stay stable since even though each channel is acquiring the incoming sine waves at a different phase, the changes are all the same across channels and the phase deltas stay constant.
RF Systems Engineer
01-31-2012 09:33 PM
Hello! I would just like to ask if I am missing something in my settings because the generator and analyzer are still out-of-phase.
The I vs time and Q vs time plots still show frequency offset (changing horizontal lines).
The master NI PXIe-5601/PXIe-5611 has a baseband module and LO module associated with it, with the remaining slave having a baseband module associated with it and LO source set to external.
For the master's reference clock source, I set it to 'PXI_CLK' in the front panel for both the generator and analyzer.
Thank you for your time!
02-03-2012 04:07 PM - edited 02-03-2012 04:09 PM
The goal of my previous post was to explain why what you are seeing is expected behavior. The only way to get the types of behavior you are expecting to see is to either look at the I and Q vs time plots for phase offset data (instead of the individual channel data), or to switch from repeated single shot acquisitions to a continuous, gap-free acquisition.
Perhaps some simple examples may help illustrate this.
I have attached a simple NI-RFSG shipping example that generates a CW signal and lets the user set the reference clock source, The default settings are a 1 GHz sine wave @ 0 dBm with the PXI backplane set as the reference clock source. This is RFSG Reference Clock.VI.
I have also attached RFSA Getting Started IQ.vi. This example acquires repeated single-shot acquisitions and plots the data on a I vs Q plot. The default settings are for 1 GHz center frequency, a 0 dBm reference level, and the PXI backplane as the reference clock source. The assumption is that the signal generator and analyzer are both locking to the same PXI backplane clock (meaning they are in the same chassis).
You are expecting a constant phase angle for the IQ data, and the RFSA Getting Started IQ.vi shows that this angle will move. Again as explained in the previous post, each new NI-RFSA acquisition will occur at some different starting point of the incoming CW sine wave. Since there is no such thing as absolute phase, the phase reference used is changing and shows up as a rotating point on the I vs Q plot.
I have also posted RFSA Acquire Continuous IQ.vi to illustrate this further. If I am performing a gap-free, continuous acquisition, there is only one acquisition instead of multiple, separate acqusiitions each with a different starting reference phase. You can see that with the continuous gap-free acquisition, the phase angle is constant.
Again, there is no difference in how the hardware is operating except for the simple difference of one case being multiple, separate acquisitions with slight breaks in time between the end of one acquisition and the start of the next acquisition, and the other case being a single acquisition with multiple fetches and no gaps in the data, meaning the starting reference phase is always the same and you don't see any change in phase angle.
In both cases, if you were to plot phase offsets between additional channels and a master, they will always have constant phase angles assuming the generator and analyzers are locked to the same ref clock.
RF Systems Engineer
02-07-2012 09:15 AM
Hello, Sir! Thank you very much for your last reply (and all others ). I would like to ask if the NI RF Phase-Coherent Multi-Channel acquisition code (main VI) is configured for repeated single-shot acquisitions. This is because when I choose CW in the generation front panel, I get changing horizontal lines for the I and Q vs t waveforms. On the other hand, I get 2 points (since I deal with 2 channels) which both move in circles for the I vs Q plot.
May I also ask if you have any suggestions on how I can fix my I and Q vs t waveforms when I choose Arb in the generation front panel. The resulting I and Q vs t waveforms are changing sinusoids which I thought to indicate frequency offset between the generator and analyzer. However, I already had my generator and analyzer both locked to the same reference clock source (they are in the same chassis and both set to PXI_CLK reference clock source in the front panel of generation and analyzer VIs).
Attached is a summary of the results I obtained.
Thank you very much for all the help!