Follows, my longer winded explanation than was
Bill's, W4ZV, just posted ahead of mine, hi.
> Enlighten da masses wot don't understand - wot is "3rd
> order IMD"
IMD = Intermodulation distortion.
In this case, distortion = signals heard in your rcvr audio/whatever
output that are not part of the original RF input signal; that is,
they are "false" signals. In a crowded band, they cause all that
racket you hear when trying to tune/copy a particular signal.
3rd order products are the simple mixing together of two close, in-
band signals which create these audible signals/noises that
really are not part of the incoming signal "spectrum". 3rd
order products are MUCH more troublesome than are 2nd
order. Why? They will lie in frequency very close to the
frequency in which you are interested. So, "close in frequency"
dynamic range problems.
Because, they can always fit inside whatever filter bandwidth you
might be using. Because they are simply added signals generated
in the non-linear devices within your rcvr's RF and IF chains; they
are "error" signals resulting from the non-linear mixing action and
occur within semiconductor junctions, or the non-linear transfer
characteristic of any "active" device: tube, transistor, diode, etc.
With two in-band signals (within the band pass of your rcvr RF/IF
chain), say one signal at frequency f1, the other at f2. Lets
let f1 = 14195; and f2 = 14200, the first from the dx station,
the other from someone responding on the frequency "5 up"
when the dx is listening up. These two signals can mix in system
non-linearity's and "create" added signals which will then pass
on through the rcvr at frequencies (2f1 - f2), and at (2f2 - f1).
These are the 3rd order intermodulation product frequencies.
They are then heard on your rig dial at frequencies: 14190 and
14205. These may or may not be a problem, except you might
be led to believe, if you heard the "intermod" at 14205, to
believe that was the frequency to which the dx were listening
rather than at 14200 where he really was listening!
Do the arithmetic, and you will see that as the signals get closer,
as during contest or dx pile-up conditions (up in the pile-up area)
there would be many more "false" signals even closer together.
Typical pile-up has many stations competing within the same few
kHz. Try the math again with two stations only 500 Hz apart; say,
at 14201 and 14201.5 in the pile. IM's are "created" at frequencies
of 14200.5 and at 14202 !! Add to those other spurious, false
signals being generated by all the other piled up real signals in
the band from 200 to 205, and you see a real mess develops!!
Same thing happens when operating CW, and many CW signals
are close together on the band.
During contest conditions, where there are signals up and down
and all about the band, and close together, there will be absolutely
hundreds of spurious signals developed!! They cause all that
stuff you hear on the bands between and atop the weaker
signals you want to copy to gain points (if you are in to contesting
seriously, I'm not, hi).
The solution, or at least a way to reduce this sort of problem:
a rig with a very high 3rd order Intercept Point specification.
The higher the power level of the "3rd Order Intercept Point" spec
of the rig being considered or used, the WEAKER in amplitude will
these generated spurious signals be and therefore , the less
trouble they become. The 3rd order IMD range has as its'
bottom limit that power level where these troublesome false
signals are just equal to the noise level under the conditions
specified. If the false signals are at, or below the noise
level, they will NOT be troublesome to the rig operator. The
power of two input real signals that mix and generate these
false signals, just down and at the noise level, that upper
power level is the "top" of the 3rd order IMD operating power
range. At what level above the noise these false signals
become a real problem, depends upon the skill and experience
of the operator, of course.
If the 3rd order IMD range is said to be 60 dB say; this means that
two mixing signals at a power level 60 db above the band noise
level at that time, just can generated false signals down at the
noise level. As these real input signal increase in power, those
generated spurious signals will increase in amplitude 3x's faster
that the real signals; that's why their slope on a graph is steeper,
and why the power levels eventually "intercept"; that is the output
power of the false, spurious signals has risen so fast that they
are at the SAME output power as the two mixing real signals.
During SSB DX operating, a typical operating bandwidth might
be, say 2.4kHz. Just now, on 20 meters at my QTH, the noise
level at 14200 is just about -115 dBm with that bandwidth selected
for the DSP IF strip (Ten Tec RX-340 rcvr.). And down on 40 meters,
where the band is not yet open, the noise level at 7088 kHz,
same bandwidth, is at only about -123 or so dBm.
So, on 20 meters I would not be troubled until two inband incoming
signals are up at an input amplitude 60 dB above that noise
level, or up at about -55 dBm, or at 6 dB per S unit, at around
S9 + 15 dB or so at which time the spurs would just begin to
rise up out of the noise.
Hope that explanation helps; maybe too many words, hi. BTW,
Ten Tec specs the RX-340 3rd order intercept point to be up
at +30 dBm, typical, + 25 dBm minimum.
73, Jim KH7M
|