Understanding and Diagnosing Repeater "desense"

To detect the amount of desense and trace its source, a signal generator with an "isotee" is used to generate a precise signal level with a dummy load and antenna.

The test is accomplished in two parts--first you determine what sensitivity loss the system has by ambient antenna system noise.
First, connect the dummy load, then note the SINAD level achieved for a given signal level (typically values of 10 to 15dB would be used).
Then, the antenna system is connected, and the signal generator RF output level is raised until the SINAD reading matches that with the dummy load. The difference between the two signal generator level in dB represents the loss of sensitivity caused by ambient site noise.

Next, the test is repeated, but this time transmitting while measuring SINAD (if the device under test is a repeater or full-duplex device). The increase in signal generator level over that required to overcome ambient noise is the amount of desense created by the transmitter.

It should be noted that if you see desense with only the dummy load, assuming your dummy load and test cables present a good impedance and are well shielded, then you may suspect that your duplexer is mistuned, or there is a cabling/connection problem.
If you see no desense with the dummy load, then proceed to test the antenna system. If you see a rise on the antenna system only, you may suspect connections, or lightning protector that's gone bad, or a bad antenna (either because it was a cheap model or otherwise oxidized to create IM products).

Expected results

Ideally, you would have no site noise (rather, site noise below the thermal noise floor set by cabling). This is usually not the case, especially for VHF. You may expect to see some single digit dB degradation due to site noise on VHF, hopefully less than 3dB on UHF and above.
In terms of desense caused by the repeater, it should be below detectable levels. Only in cases where the so-called "flat-pack" notch-only duplexers should you ever see detectable levels of desense from the duplexer (and even with flat-packs it should be 1dB or less unless you have the really crummy radio-telephone type).
If you are using band-pass/band-reject duplexers with quality RG-142 cable and N-connectors, you should find no desense to the 100 Watt+ output level.

W8MSU's desense

I didn't have the time to do this during the semester, so I instead used a fixed weak signal transmitter, and toggled the transmitter on and off repeatedly. This does not give such precise results, but does give a ballpark estimate of if the transmitter is affecting the repeater's receive sensitivity.

In this case, I found the repeater was desensing itself by something on the order of 6-10dB--the level varies and sometimes is up to 15dB. This may indicate an antenna system issue, either alone or in conjunction with a duplexer/connection issue--but a more precise diagnosis would need to be undertaken first.

Example of repeater desense self-interference

In this audio clip, you hear how it's somewhat difficult to hear with TX on, then easy to hear with TX off, and hard again with TX on.

30 seconds of audio, TX on, then TX off, then TX on again