How to compensate mixer self-mixing in Zero-IF?

Self-mixing creates a DC offset, and it is not a constant value, question is : how to remove this DC offset by circuits design?





Thanks

coul'd you be more explicit? Explain better please

Hi

If your mixer works in a TDMA system, you can use idle times to compensate LO leakage.

Regards

You can compensate it in digital domain, by calibrating with the weakest signal. By measuring the DC offset when there is no signal, then substract the same offset from the real time signal. Some solution calibrate once some everytime.

If the design is diferential then the DC and bias is substract. If you need differential outputs that you are O.K. But if yiu need a single port then substact the outputs.

Use digital domain to compensate the DC offset problem .
But you need add a feedback chain and downconvert to baseband ADC( this chain you can only use traditional receiver structure, because of avoiding another IQ error in feedback part)

In generel the old paper from Razavi is still a good introduction to the topic; "Design Considerations for Direct Conversion Receivers", which can be found here; http://www.ee.ucla.edu/~razavi/jp13pg8.pdf.
As mentioned you can do a lot in the digital domain, but if your baseband chain exhibits just a recent gain the dc-offset might be amplified to a degree where it will saturate the output stages (so you won't get it into the digital domain). So you have to use a DC-offset compensation circuit/architecture. An example of such one could be as the one depicted below (taken from RFIC2004, "A Low-Power 10Gb/s AGC Optical Postamplifier in SiGe", by Daniel Kucharski and Kevin T. Kornegay)

i think analog feedback loop can remove the dcoffset.
for example, @ the mixer load ,you can add to a low-filter analog feedback loop.

Self mixing DC offset is a inherent problem of circuit itself. With DC correction
loop, this effect can be seen as "gone" in received analog/digital signal view of point. Since DC offset direct hit the RX C/N performance with varying high/low
input signal amplitude, usually we should make calibration before normal TDMA
receive burst working. Without good correction applied, RX sensitivity, 3MHz blocker, and even co-channel interfer rejection would be affected much.

With Regards,