[RFI] advice wanted: finding the source of mixing products

[Jim Brown]

On 2/24/2020 12:32 PM, Matt NQ6N wrote:
> Can you recommend a test procedure I could use to determine how much of
> this broadband noise is originating in my transmitter (so that I can
> determine whether it is within spec)?

A good start is to study ARRL Lab tests for the rig(s) you're using, 
paying attention to plots of measured Phase Noise. Some rigs are far 
cleaner than others. About five years ago, inspired by a list of rigs 
used at that year's WRTC and complaints about some of them being pretty 
dirty, and by Rob Sherwood's work on RX performance, I coordinated with 
ARRL Labs to produce a document with their data for 15 different rigs 
plotted on the same scale, allowing meaningful comparisons. The report 
is here.

http://k9yc.com/TXNoise.pdf

Yaesu took the crown for the dirtiest of the higher priced rigs tested, 
with ICOM a not-too-distant second. Elecraft's was the cleanest, with 
Kenwood closer to Elecraft than ICOM. All of the el-cheapo JA rigs were 
dirty. A few months after I leaked a preliminary copy of the report to 
someone who I suspected would pass it along to Yaesu, the company issued 
a firmware update that significantly improved it's CW bandwidth, 
bringing the rigs to which it applied roughly in line with ICOM.

At the time of my work, it would be a year or two before the excellent 
Flex 6500-6700 rigs were tested, and when test results were published, 
their keying was quite broad. Within a year, new firmware cleaned it up 
a lot, putting the radios in a class with Elecraft for CW bandwidth.

Nowadays, pretty good spectrum analyzers are available to any ham with 
the technical skills to know how to use them. Some are listed beginning 
with slide #31 in this deck.

http://k9yc.com/KillingRXNoiseVisalia.pdf

In general, you'll get decent data if 1) you don't overload the A/D 
conversion/sampling process; 2) you use display width settings that 
don't produce aliasing; and 3) your sense antenna is relatively flat in 
its sensitivity.

As an alternative to #3 input to the analyzer could be obtained from a 
simple voltage divider across the coax at the transmitter output, with 
the output resistor being 50 ohms. The resistors used should have flat 
response vs frequency; modern SMT resistors have that characteristic, 
and it's pretty easy to solder leads to those of the larger form factors.

73, Jim K9YC