[1000mp] Test IMD vs actual IMD

[Tom Rauch]

>    In fact the IMD products that are created by a voice signal going
>    through the
> transmitter are far worse than the industry standard two tone test.
> This is because there are many simultaneous tones within the voice
> spectrum. All of component tones are capable of producing intermod
> products with every other one.

The major reason I have found voice signals are worse is because they 
vary at a syllabic rate. The very low-frequency components change 
power supply loading of bias and main power sources, and make the 
supplies "waddle" all around.

Any test that steadily loads the supply with a high frequency load, 
such as two tones spaced 2 kHz or a notch-noise test, misses most 
dynamic power supply problems because bypass capacitors can hold up 
the supply voltage through their energy storage capabilities and mask 
low frequency dynamic regulation problems.

Another misconception is the idea reducing power output always 
improves the IMD when expressed in dB down from mean output. That 
simply is not true. Many systems have worse IM performance at very 
low power than they do at full power, mostly these problems come from 
from biasing errors.

  73, Tom W8JI
W8JI@contesting.com