pxie 5673E modulation

[kylelove2902]

hello everyone,

I'm using pxie 5673e and pxie 5663e. What I want to do is making a pulse about 100 MHZ by pxie-5450 (AWG) which modulates with pxie-5652 (LO) about 5 GHZ, , and then using the pxie 5663e to demodulate.

I saw the RFSG Getting Started Single Tone Generation example, so I know how to generate a sine wave. SO my question is: I need to make two  tones which one is controlled by AWG and the other one is controlled by LO, and then use somekind of upconversion vi to modulate them.

Is the step wrong and where can I find upconversion vi?

Thanks!

[kylelove2902]

Hi everyone,

I have another question. I put pxie 5673E and pxie 5663E in the same chassis.  I let pxie 5673E generate a sine wave at 1GHZ, and then I use the eaxmple  RFSA Acquire Continuous IQ.vi to get the IQ which I set carrier frequency at 1GHZ. What i expect is 0 or 1, but the value of Q and I are sine waves about 1 kHZ with 90 degree phase difference. So my question is maybe they don't share the same clock?

Thanks!

[Andy_Hinde]

The generator and analyzer will by default use their own onboard refernece clocks, so even though you are tuning both to 1 GHz, you are measuring a 1 kHz frequency offset between Tx and Rx as a result of not sharing a common timebase.

Lock both the generator and analyzer to the same reference clock (PXI backplane clock) and you will remove any frequency offset between them, and your I and Q waveforms will be DC traces/flat lines vs time.

[kylelove2902]

Thanks, Andy_Hinde

I'll try it later!

Do you have any suggestion about the modulation or any vi example?

What I want to do is letting the Awg signal modulates with local oscillator signal which both are in pxie 5673E.

Thanks!

[Andy_Hinde]

Hi,

Your original question seems to imply that a user has to individually control the LO source (5652), the baseband IQ generator (5450), and the IQ modulator (5611). This is not the case. You use the NI-RFSG driver to control the 5673E as a single instrument.

You create an IQ waveform which represents the pulse timing characteristics you want and upload that IQ data to the RFSG driver, while configuring the output power and RF output frequency (5 GHz). If you want a different output frequency, the baseband waveform stays the same, you just configure a different frequency in the driver and it will (automatically) adjust the LO from the 5652 accordingly.

The 'RFSG Getting Started Single Tone Generation.vi' example makes use of an ease of use mode of the driver called CW mode which hides some of the thigns under the hood. I suggest looking at the 'RFSG Arbitrary Waveform Generation.vi' example to show how you create and upload AWG data to the device so you can generate anything. IF you were to replace the IQ data being uploaded in this example with data where the I data array is all 1s and the Q data array is all 0s, the output would be an RF CW signal at the chosen frequency.

If you want a pulsed output, your waveform would be a similar one (I=1s, Q=0s) during the time you want the pulse active, and then I=0 Q=0 during the time you want the pulse off. You use the number of samples and sample rate to control the timing. There is also a shipping example called 'RFSG Pulsed Data.vi' which shows you exactly how to do this.

[kylelove2902]

Hi,

I tried to follow this link http://digital.ni.com/public.nsf/allkb/F2946625E778BB19862577500072EEB7 to synchronize the NI 5663E and NI 5673E. The difference is the property node which I changed the ClkIn to RefOut 2 because if I chose ClkIn, the IQ results were sine waves. After finishing it, I used the RFSG Getting Started Single Tone Generation example to generate waves and used RFSA Acquire continuous IQ example to get IQ values. I find that every time I stop the program and then start it, the IQ values are different although they are DC curves. Is this because I didn't let RFSG and RFSA initialize VI synchronized, but actually I put these two examples in the same program. Another problem is when making the RFSA frequency difference with the RFSG frequency, the IQ values are all zero, so I'm not sure all the things I did is right.

Thanks!

[Andy_Hinde]

"I find that every time I stop the program and then start it, the IQ values are different although they are DC curves"

This is expected behavior. The I and Q values are the Re and Im components (Cartesian) of the complex data, so their values will change based on the phase relationship between the Tx and Rx changing each time you start a new acquisition. 

"Another problem is when making the RFSA frequency difference with the RFSG frequency, the IQ values are all zero, so I'm not sure all the things I did is right."

If you are tuning the RFSA to a different frequency and receiving all 0s, it is because your acquisition bandwidth no longer captures the signal from the RFSG. Your acquisition bandwidth is (0.8 * RFSA IQ sample rate), so if your sample rate is 1 MSPS, you are acquiring 800 kHz of input signal. If you move the RFSA center frequency more than 400 kHz away from the RFSG frequency in this scenario, your analyzer will no longer see the signal and you will be capturing noise.

[kylelove2902]

Thanks! It helps a lot!

So is there any way to fix the phase relationship between them?

[Andy_Hinde]

The 5663 and 5673 have different architectures, the 5663 is a single stage downconversion architecture where the IF is usually at 187.5 MHz. That means the LO from the 5652 is at either (Fc-187.5M) or (Fc+187.5M) where Fc is the downconverter center frequency.

The 5673 directly upconverts the baseband IQ to RF with no IF stage, so the target Fc upconverter center frequency is the same as the LO frequency coming from the 5673's 5652.

What this means is that when both devices are tuned to the same frequency, the LOs coming from the 5652s are different frequencies, so a single LO cannot be shared between the 5663 and 5673 to create a phase-coherent Tx/Rx system. When Tx and Rx share the same LO, not only is the phase relationship between them fixed, the phase noise of the LO is correlated and common to both and Tx/Rx move together effectively cancelling out all phase noise in the LO PLL loop bandwidth.

In this case, the Tx and Rx have separate LOs, which are synthesized from different PLLs on different 5652 modules. While these PLLs can use a common reference clock to generate LOs of the same frequency, the dividers in the PLL are not synchronized in such a way as to produce the same phase relationship each time. So each time you generate the Tx and Rx LOs out of the 5652, their phase relationship relative to each other changes. Remember of course that the PLLs on the 5652 modules are not generating LOs of the same frequency for Tx and Rx in the case of a 5663 and 5673, so it gets more complicated from there.

What is your goal or application where you are trying to achieve a consistent Tx and Rx phase relationship across device sessions/acquisitions?

[kylelove2902]

In my project, I send a wave into the sample, after passing through it,  the wave will have a phase shift and I need to get the value of the phase shift. In addition, the wave is too weak to perform a single shot, so I need to do lots of times to get the average. If I cannot fix the phase relationship of Tx and Rx , then the phase shift I want will not be correct. 

Thanks!