Zero IF Local Oscillator bleed compensation questions

[gbugh]

If I have a switching quadrature demodulator with a differential input and I purposely set the H input at a slightly higher DC offset and the L input at a slightly lower DC offset then anytime my L.O.'s I and Q switching inputs are not each exactly 50% duty cycle it will cause my I and Q zero IF outputs to have a DC offset. If I integrate those offsets over a long time period, like 100 seconds (0.01Hz), and feed them into circuits that correct the duty cycles of the I and the Q square waves then I ought to be able to keep the duty cycles very close to 50%.

But if I have any LO bleed into the RF input then that would cause a DC offset in the output also and so the offset will drive the duty cycle correction slightly off 50% untill it nulls the LO bleed effects.

So assuming I have a small amount of LO bleed and the correction circuit nulled it out but as a consequence the duty cycle is now 49.95%/50.05%, then what adverse effects will that have on received RF signals? Is it like no big deal? I'm trying to visualize how it would effect any other incoming signals over an output bandwidth of like 2.5KHz or less for the I and the Q outputs and I can't think of how it would be a problem.

Is it correct that usually the whole reason a non 50% duty cycle is a problem is that it allows the LO or its harmonics to bleed through? So if I nulled out that bleed through then everything else should still work well, yes?

If I zero beat on a strong signal it still ought to have some output higher than the 0.01Hz correction response so it shouldn't cause the correction circuit to null it out, I think.

any feedback would be appreciated, thanks,

I could just not worry about a slight DC offset except I feel challeged to see if I can succeed at designing a receiver with a large gain and still have all the stages DC coupled, or at least with a lower frequency limit of 0.01Hz. Maybe it would be of some use to someone later. I'm already using zero drift op amps with no 1/f noise corner frequency after the mixer output.

73
AF5IE

 

[gbugh]

I see tons of research papers about DC offset in DC receivers but I don't see anyone trying to manipulate the LO duty cycle to null DC offsets.

If the L.O. I and Q square waves are slightly off from 50% so as to null L.O. leakage into the RF input does that let strong signals of near by frequencies come through?

It still seems like that would be only a small problem if the duty cycle is only slightly off. I'm not good enough at the math to figure out all the equations involved to determine what the effect would be.

George

 

[kh4n]

You can easily obtain 70-80dB cancellation by injecting DC offset to IF ports of mixer but, it will get harder to obtain that levels of suppression if you have a variable LO frequency, wide temp range, etc.