Agilent makes a bunch of possible detector diodes. For example, the
HSMS-2850 series of diodes suggests a sensitivity of "up to 35mV/uW"
at 2.45GHz. So with a 20dB coupler and 10mW (+10dBm) excitation, 40dB
return loss would give you about .3mV out of the detector, plenty to
see on a typical 3.5 digit digital voltmeter on a 200mV range. You'd
be lucky to get your coupler to have "40dB directivity" (that is,
calibrated that accurately to 50 ohms) anyway, unless you have a good
50 ohm precision load to check with. Be careful with such diodes:
they have a very low reverse voltage breakdown, so you shouldn't put
more than about +10dBm directly into a detector built with one of
them.

Ok I looked through the Agilent website and found what I think is the
proper diode. An HSMS-2852 two diode in series pack 3 pin surface
mount. However I noticed that the specs say it is best below 1.5ghz
and recomends the HSMS-282X Series for below 4ghz and above -20dBm.

In either case I can't find a place to purchase any of them in
quantities of less than 100. I can't seem to cross reference them to
anything that digikey carries. Quantities of 10 or 20 wouldn't be bad
but 100 at $.90/per is a little much.


I also found these other possibilities
HSCH-3486
ASI-3486
MA4E928B-54