Hi all,

I'm trying to do on-wafer measurements of a mixer i designed. It is a fully differential quadrature mixer. The 2.45 GHz LO input is split into 4 phases by an off-wafer balun and an on-wafer polyphase filter. The 2 MHz mixer outputs (I and Q) and connected to differential to single ended buffers and can be read individually. Also, a polyphase filter was added to the I and Q paths on wafer to test the image rejection. So a third output is available. The input actually comes from an LNA (which has been silicon verified) with a 2.4-2.5 GHz RF input.

Anyway, the results were not at all what i expected. I used two sig gens and a spectrum analyzer. For the individual I and Q paths, the gain was 0 dB when it should have been around 10 dB including cable loss. However, the combined output showed between +4 to +12 dB gain depending on the buffer current (the buffer has a maximum gain of 0 dB unless theres some unknown problem with the buffer). Unfortunately, there was no image rejection at all. Not only that, there was a nice big noise-like signal at around 8 MHz.

Anyway, I don't expect anyone to be able to pinpoint the problems, but does anybody have advice on troubleshooting? How can i go about finding the problem?

I was able to observe the LO leakage at the mixer output (-53 dBm), and if the RF input was high enough, it was also observable at the output (-70 dBm with RFin of -30 dBm).

thanks,
Aaron