> If I can measure the baseline (the exciter) first, the additional IMD
> generated in the amp shows up as an increase from that baseline.
> Subtract the baseline and I have a number that represents the amps
> distortion.
Distortion is phase coherent. Because of phase and level
differences through a system where each stage adds distortion,
distortion can add, subtract or do anything in between. Measuring
white noise is much different than measuring distortion.
This is the same mistake an amplifier manufacturer made when he
said he was going to patent an input circuit that "cleaned up
exciters". He used a broadband toroid, which added distortion that
just happened by dumb luck to correct distortion from the exciter
he was using. Put it on a different exciter, and it could just as
easily add IM distortion!
This is why everyone either uses ultra-low distortion test generators
or two separate isolated exciters operating on CW in tests. Not
because they want to complicate the system, but because they
want to have a meaningful result.
> The above procedure is widely used in analytical labs to subtract out
> interfering spectra. It works there.
It probably does, because it is an entirely different situation!
> I don't need numbers rigorous enough to submit to FCC on an
> application for approval.
There is no point measuring something if we don't know at all what
we are measuring.
> >You certainly can't measure the IMD on anything but a good
> >selective level meter or spectrum analyzer...looking at the audio of
> >a receiver is a total waste of time.
>
> Analyzer2000 - DSP spectral anaysis on the PC soundcard. Looks pretty
> good. Seen it?
Receiver bandwidth, assuming you are sweeping with the software,
would have to be flat and of the entire sweep range of the analyzer
program. You also could have a DSB response if you aren't careful,
and that might cause several dB error. The receiver would have to
be significantly better that the linear amplifier in IM DR, including
the audio stages of the receiver, as would the sound card (and
some are pretty poor). Not a good idea.
You could turn the AGC off, use a narrow filter, and look at
monotone levels from the audio port on a meter. In that case you
would be using the receiver as a narrow selective level meter, and if
the receiver was linear enough you could obtain some useful
data....but there is no reason to use an analyzer program at all.
As a matter of fact, it would be worse than just using a meter on
the audio output and an accurate front end attenuator
pad...assuming the receiver was good enough (which I doubt).
When I want to make wide DR measurements, I sometimes use a
homebrew receiver that has a 120+ dB blocking and 110+ IM
dynamic range. It's better than my spectrum analyzers.
My FT1000D, with mods to correct some design errors at Yaesu,
is one of the better amateur grade receivers and is around the 80+
dB blocking and IM DR area, but that is ***NOT** using it as a
tunable broad IF where even detector and audio distortion can
create garbage. That is using 250Hz IF filters.
73, Tom W8JI
W8JI@contesting.com
--
FAQ on WWW: http://www.contesting.com/FAQ/amps
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com
|