PXIe 5673 PXIe 5663E phase locked

[kylemonmon]

Hi, everyone

I'm using the PXIe 5673E to generate a pulse to DUT, and do the arctan (I/Q) to get the phase changing versus time by PXIe 5663E. However, due to bad signal to noise, I need to do lots of average. SO what I want to do is repeating the pulse and averaging the IQ data and then I can get the phase information versus time. What I have done is that I repeat the single-shot experiment by using the for loop, but the problem is that because PXIe 5673E and PXIe 5663E are not phase-locked, every single-shot data shows different phase response. 

I'm curious that is there any way to fix the phase-locked or is there any better way to do the repeating measurement. If not, then does it mean that I cannot use them to do the measurement at bad signal to noise?

Any suggestion is appreciated!

[muhammad.kamran.ayub]

Hello kylemonmon,

I highly doubt this can be done upto a sophisticated level. "Phase locked" means:

• Triggering both the devices together (can be done using PXIe back place triggers and NI RFSA/RFSG drivers)
• Sharing LO (cos (theta) and sine(theta) multiplies with signal, we  need to be sure that this is consistent for VSA and VSG)
• Sharing ADC/DAC Ref clocks (PXIe backplane clock can be used, or in case of onboard clock, it can be exported and given to secondary module)
• Equalize for any mismatch in phase because of electrical length of  RF/LO sharing wires (normally done using constant group delay filter)

As a VSA and VSG can have different IF and thus different LO values for up/down coversion, sharing the LO is a question mark. Though, LO can be exported out of most NI devices. Is it useful in your case? I do not think. Though, still you can read about the device architectures in depth. One thing that I can think of is to use same ref clock to derive PLL which generates LO. This can help a bit.

Sharing ADC and DAC ref clocks can be done. 

Equalization step comes after once LOs are shared. In the case there is (mostly) a constant phase offset whenever you recieve, we use equalization. Though, until LO sharing isn't done, phase will stay random and I don't believe this step can be applied.

My recommendation would be to look for algorithmic fix. 

If this is the answer you are looking for, please mark it as solution. Otherwise, please let me know the further queries. Thank you!

Best Regards,

Muhammad Kamran Ayub
Staff Technical Support Engineer | Platforms & Systems

Certified LabVIEW Architect (CLA) | Certified TestStand Developer (CTD)

National Instruments UK & Ireland

[Andy_Hinde]

Hello kylemonon,

You can do mostly what you want to do with a 5663E and 5673E. To start, here is a link to an RF Multichannel NI Community group which has background info and LV code. This is mainly speaking to using multiple 5663Es to create a multichannel Rx system, or multiple 5673Es to create a multichannel Tx system, as opposed to using a 5673E with a 5663E to create a phase-coherent Tx/Rx system. Nevertheless, many concepts apply.

https://forums.ni.com/t5/RF-Multi-Channel/gp-p/5218?view=overview

You cannot share an LO between the 5663E and 5673E unfortunately, as the 5663E is an RF-IF architecture, and the 5673E is a Baseband-RF architecture, so when the 5663E and 5673E are tuned to the same frequency, the LOs are different. The best you can do is lock the two instruments to the same 10 MHz reference clock (presumably the PXI backplane clock) to remove any frequency offset between them (i.e. frequency lock). The instruments cannot share LOs so you cannot achieve a phase lock, the phase noise from each instruments' LO PLLs will cause some random phase movement between the Tx and Rx. You'll have to determine if this adds too much uncertainty to your measurement.

The behavior you are seeing could be due to not locking the Tx and Rx to the same 10 MHz reference, and the resulting frequency offset appears as a constantly rotating/changing phase at the Rx.

The behavior could also be due to the Onboard Signal Processor (OSP) of the 5663E resetting its digital LO for every single-shot acquisition. The 5663E has an OSP which functions as a digital downconverter to digitally downconvert the digitzed data from IF to IQ (baseband). This digital downconverter has a digital LO signal called a numericaly controlled oscillator (NCO). The NCO is reset to 0* phase every time a new single shot acquisition is performed. If you want to make repeated acquisitions all using a common phase reference, the NCO cannot be reset between acquisitions, so you need to perform a multirecord acquisition, which will allow you to take multiple acquisitions without resetting the 5663E NCO. 

Try starting with the NI-RFSA shipping example RFSA Getting Started Multi Record IQ.vi and make sure you are locking the 5663E and the 5673E to the same ref clock source. Multirecord will need you to trigger each individual record/acquisition, you can either export a trigger from the 5673E to trigger the 5663E or you can use the IQ Power Edge trigger on the 5663E which will trigger on the rising magnitude of the input signal when the pulse starts.

Regards,

Andy Hinde, MBA

Principal Systems R&D Engineer, Semiconductor

National Instruments