Verizon News wrote:
Keepthis in mind:
+7dbm (the recommended RMS input for a level 7 mixer) is about 1.5Vpp sine.
This holds up to reason, as a DBM has two diode-drops that the LO must
overcome in order to switch the diodes properly. Therefore, if you use a
square wave, you also need to supply about 1.5Vpp, or the diodes won't
switch.

There are two common misconceptions many people hold about these devices.
One is that they are inherently 50 ohm devices. In reality you will only see
50 ohms on a given port if the other two ports are also terminated in 50
ohms. You could run the device just as well at 75 ohms if you keep it
balanced properly. Also, many people think that the LO input is power
dependent. This only partially true. It's really voltage dependent,
requiring sufficient drive voltage to turn on the diodes and pump an amount
of current through them that is large compared to the expected input RF
signal.

There is no magic in these things. They're just diode DPDT switches that
reverse the polarity of the RF signal at a rate determined by the LO
frequency. This is essentially the same as multiplying the RF signal by +/-1
at the LO rate. Simple math shows that this multiplication process yields
the sum and difference frequency products. The more symmetrical the LO
switching waveform, the more perfect the multiplication. This is why square
wave LO drive is preferred.

Joe
W3JDR



Thank you all for your help, suggestions and especially for the simple
explanation that the mixer diode pairs need enough voltage for
switching. I will remeasure with higher voltages and use buffers that
supply enough current to drive the mixer.

Andreas