How can keep two channels phase difference constant vs. SNR from 85dB to 5dB?

https://www.edaboard.com/threads/221992/
I have a two channels system, as I have posted in the foresaid thread.
Now the two channels phase difference can keep constant (Error less than 1 deg) vs. SNR from 85dB to 30dB. I want to improve it, who can give some recommendations?
I have the following ideas:
1. Replace the normal RF cable with better cable? who can give advices on better RF cables? especially on constant phase performance despite the shape of the cable? My freq is about 5.6GHz.
2. Does the phase shift of the receivers vary vs. SNR from 85dB to 5dB? I have two receiver channels with same configuration.

If the gain in your receiver is varied using a VGA (active device), most probably you get phase variations.

If the gain in the receiver is varied using passive attenuators (or PIN diodes) there are chances to get much less phase variations.

Can you pass both signals thru the same receiver? i.e. offset one up in frequency, and offset one down in frequency, using the same offset oscillator, and then send them both thru the exact same receiver. Maybe use 100 MHz as the offset LO. The phases should track a lot better that way. You can digitally process the two signals at the end, or you can reconvert them back in frequency with another mix process using that same offset LO.

1.) An RF cable at 5.6GHz will definitely have besides heavy attenuation around 20-40dB/100ft can definitely produce phase variation. You can look for soil, braided cables, not the flex ones their performance is not as good yet. Constant phase difference performance is possible but to get a phase match for two different signals over a distance is rather difficult.

You could measure for a constant phase difference (use a phase pre-compensation circuit) or better use a PLL and then lock both together instead.


2.) Phase shift is more a practical phenomena than a theoretical aspect. Theoretically SNR should not affect it since it is only a relative shift in the signal position. But practically SNR drop is caused by factors such as jitter and timing uncertainity, so as the SNR drops contribution of smaller noise factors starts to creep in.

thank all of you.

If the gain in the receiver is varied using passive attenuators (or PIN diodes) there are chances to get much less phase variations.

Click to expand...

Yes, we use Aeroflex 7bit digital controller attenuators (0~103dB) at 5.6GHz , I guess it is PIN diodes, I will check the datasheet.

Can you pass both signals thru the same receiver?...

Click to expand...

That sounds a good idea. But that need a lot of work to do for our weather radar systems.

use a phase pre-compensation circuit

Click to expand...

Could you give more details about that?

then it sounds like you need to sit a tech down in front of a network analyzer, and gain/phase match pairs of components from a big lot. Then you install those matched pairs when assembling the two receiver channels, and the two channels stay married for life.

thank you very much, biff44.

All the digital attenuators from Aeroflex mentioned in the datasheet that, "phase-free" (phase invariant) is an option for them.
Did you ask Aeroflex for this option?

From my experience even a very low phase variation digital attenuator using PIN diodes, will have something like +/- 2 deg variation, and this at lower attenuation range. When attenuation is higher, the phase shift increase.
Beware that the phase shift also varies with frequency, and when I say frequency, I mean very small frequency variations.

hi,

I would stick with finding a pair of matching cables with a tolerable phase difference as suggested.

Nevertheless, a more extended route would be using the phase precompensation circuit, though for high frequencies like that you would need a PLL for it to work (if it was low I would deal with it using a simple RC), then you can offset any phase loss that you experience at the end of the cable. This could be viable if your going to mass produce or it is a precision equipment. Will work for any pair ofcables.

btw: I am presuming that you have both the receiver channels operating from the same oscillator circuit.

Yes, both channels operate from the same freq source.
BUT we don't need that the phase shift of the two channels should be same, we need the difference of the phase shift of the two channels be constant in the full range of the input power.