Phase changing

It sounds like the chip might be power cycled during your testing. The 180 degree difference between channels is likely due to the LO frequency divider initial condition being random at chip startup. You may need to keep track of the outputs at startup with a defined RF input to see if there is any 180 shift. Your other option is to use a demodulator without the frequency divide by 2 in the LO path.

First of all, thank you for your answer. We kept track of the outputs at startup, and we can see a 180 shift sometimes, but it only happens during the startup as you said. The problem we are having is that after the startup, once the outputs are stablished, there is not a 180 shift unless our equipment start to transmit. If we decrease the power output of our equipment, the 180 shift ratio decreases, even stops. 

We have tried to power the chip externally, with a DC power supply of high power output, and the 180 shift keeps hapenning. Also, we have isolated the RF input of our I/O and put an isolated input directly to the demodulating logarithmic. The phase shift keeps happening. As well, we isolated the output and it keeps happening... We believe that the problem is that there is a radiated emission in our PCB that conduces to the chip and make it shift the phase, but we dont know across which pins of the chip it could be...we suspect of the OL input, but we put an external OL as well, and the phase keep changing also.

Do you know if you have a demodulator without the frequency divide by 2, with the same layout? I mean, that we could easily change it without redesigning our PCB.

Thank you very much!