I have only been able to estimate the 90Hz signal by injecting a signal into the input from a sig gen at a high amplitude to override the spurious signal and see where it sits on the waveform. I actually suspect that the spurious signal is nothing like 90Hz but a signal close to the LO signal or a harmonic of it.
Regarding the mixer, I have replaced the IC and the balance trim pot to no effect.
I totally agree that the best way to fix it may be another 3850. I have for some time been on the lookout for a HP8556 plug in for my HP141T that would solve all the problems.