Phase Locking an NCO

Hi Matt,

--- Quote Start ---

it should in theory be easier to just use the phase-mod input of the NCO. Anyone have any ideas they would be willing to share?

--- Quote End ---

Your idea cannot work.

If you shift the phase of the NCO, then you shift the phase of the transmitted sine as well. So your dot-product of the NCO with the received signal will still be the same, as the NCO phase you added is in common to both paths.

You would need an NCO at the receive end if you wanted to align that NCO to your received signal. However, what would be the point? The measured phase tells you something about your propagation delay. It might be useful to you.

Note that this is what frequency response analyzers exploit; you inject a sinusoid into your device under test, and then measure the magnitude and phase of the output sine, and then use that to construct a Bode plot for the device response.

Cheers,

Dave

Sorry about that, I have no idea what I was thinking when I typed that.

Basically, im trying to get rid of the delay that I have when Im running my code and transmitting through a small patch cable. I shouldn't have a big delay but I do and I'm not sure how to get rid of it. It throws off all my future calculations.

Additionally, the delay depends on the frequency of my NCO. I assume this is because a 5 clock delay is not much for a low frequency sine wave, but 5 clock cycles end up inducing a rather large phase shift for high-frequency signals.

Matt

--- Quote Start ---

Sorry about that, I have no idea what I was thinking when I typed that.

--- Quote End ---

No worries :)

--- Quote Start ---

Basically, im trying to get rid of the delay that I have when Im running my code and transmitting through a small patch cable. I shouldn't have a big delay but I do and I'm not sure how to get rid of it. It throws off all my future calculations.

Additionally, the delay depends on the frequency of my NCO. I assume this is because a 5 clock delay is not much for a low frequency sine wave, but 5 clock cycles end up inducing a rather large phase shift for high-frequency signals.

--- Quote End ---

You're going to get delays no matter what, so why not just incorporate them into your analysis?

A frequency response analyzer gets around this by measuring the injected signal and the output signal, and then you compare those.

A vector network analyzer gets around it by you having to first short the injection and sensing cables together and then zeroing the response, and then connecting your device under test between the calibrated cable lengths.

Perhaps either of those options can work for you.

Cheers,

Dave

If I understand the question right, you want to derive a phase shifted second signal from the original NCO signal rather than phase locking the NCO. At least you didn't mention which signal the NCO should be locked to. Phase locking of a NCO would basically work with a PLL, but apparently doesn't apply here.

To generate derived phase shifted signal, you have two options:

- use a dual output (quadrature) NCO and generate the phase shifted output by linear operations

- use a multi channel NCO that has independent phase and frequency settings for each channel

Thanks everyone. Sometimes I can be a little confusing....

Here is what I was trying to do:

I have a delay in my system that I am trying to get rid of. I use a Costa's loop for carrier recovery and have a rather large delay when measuring during calibration. I was thinking since I know the delay, I can somehow remove it by phase locking my NCO somehow. But as Dave said, it is much easier to just use some post processing in Matlab to compensate, which is what I did and it seems to work.

Thanks again

Matt