[Amps] Measuring IMD

[Manfred Mornhinweg]

Hello all,

my approach to IMD measuring is this:

We need basically two things: A dual tone drive source, and some sort of 
spectrum analyzer. I will comment them separately.

The old style of dual tone exciter would comprise two low distortion 
audio oscillators, an adder, and feeding the resulting signal into an HF 
transmitter (or transceiver).

This method is obsolete, because nowadays there is free software 
available that can be used to create the same dual tone signal on the 
computer, with much higher quality. Distortion of -80dB, and perfectly 
matched levels of the two tones, are easy to obtain.

The signal can either be directly fed into the radio, perhaps using an 
existing "soundcard interface", which doesn't need to be much more than 
a few straight wires, or you could create an audio file with the dual 
tone signal, load it into a portable MP3 player, which are cheap and 
easy to get, and there you have your portable, battery powered low 
distortion dual tone generator. Just put the player in loop mode, to 
play the same file continuously, and use WAV format instead of MP3.

Any of the above approaches is strictly limited in how low an IMD it can 
measure, because common SSB transmitters produce lots of IMD themselves. 
  You cannot measure the IMD of a high quality amp, when driving it from 
a poor quality transceiver, because the IMD from the tranceiver will 
mask that of the amp.

To get around this, you can use two HF radios, if you have them, and a 
combiner, to drive the amp under test. The combiner can be built from 
some ferrite cores, some wire, and perhaps a resistor. In this case, no 
dual tone generator (hardware nor software) is required. The 
transmitters don't even need to be SSB. All you need is CW. Just set 
them 1kHz apart, and transmit a steady carrier from each. Match their 
power levels.
The combiner should have as good a cross attenuation as possible. This 
is both to protect one radio from the other, and to keep the 
transmitters from generating IMD by mixing each other's signal.

To play it safe, when using transceivers, use this sequence: Set power 
level to zero on both, key them both up, increase power settings to 
whatever drive level you need, matching them. To stop transmission, do 
the same in reverse. So you don't allow one radio to transmit anything 
into the receiver of the other radio, through a leaky combiner.

Now the spectrum analyzer: I'm using the free program by George R. 
Steber, with the very inexpensive mixer front end, and a still 
reasonably inexpensive DDS signal source. It works pretty well.

But if you want an "entirely free" solution, and you have yet another 
radio (receiver), hopefully one with a rather wide bandwidth, you can 
use that. Just set the bandwith as wide as possible, the BFO to one edge 
of that bandwidth, and feed the receiver output into the computer, using 
the free Spectrum Lab software. Tune the receiver so that its bandwidth 
is centered on the signal center. Depending on the separation of your 
two tones, and the bandwidth of the receiver, you will be able to see 
and measure at least 3rd order IMD, probably 5th order too, possibly 
even more.
If you can shut off the AGC and manually control the RF gain, you can 
tune the receiver off the center of the signal, to look at far-away IMD 
products. In that case even a standard, narrow receiver can be used to 
fully measure IMD, but it's a bit more laborious than if you can see the 
full spectral display at once.

In short: If you have a SSB transmitter, a separate wide SSB receiver, 
and a computer, you can already do some meaningful IMD testing without 
spending any significant amount of money. With an additional 
transmitter, which could be an old AM rig, you can improve the measuring 
range. And with a cheap mixer and some frequency source instead of the 
receiver above, you can measure IMD up to the upteenth order product.

The accuracy of the results will depend on the equipment used. Be aware 
of receiver filters changing relative levels of signals by several dB, 
even well within their passband. They all have passband ripple. The 
wider and smoother the receiver's response, the better. But even with 
pretty basic means, good results can be obtained. Certainly many, many 
times better than putting an amp on the air and asking fellow hams about 
the perceived amount of splatter! Although that on-air test should also 
eventually be used, as a reality check.

Manfred


========================
Visit my hobby homepage!
http://ludens.cl
========================