Thanks Colin. I wonder, too, whether the image noise would be a
problem at the 2nd IF, as it is at the 1st IF. I'll have to see if I
can find any info on that somewhere!
---
The image frequency is only a problem in the first mixer since it
translates to the first IF, being the carrier frequency plus twice
the IF.
For example, consider a superheterodyne AM broadcast receiver with a
455kHz IF tuned to a 1000kHz carrier.
Since it's a _super_heterodyne receiver, the local oscillator will
be tuned to the carrier plus the IF and will be oscillating at:
LO = RF + IF = 1000kHz + 455kHz = 1455kHz
This signal will combine with the carrier, in the mixer, and the
output from the mixer will contain, in addition to the carrier and
the LO, the sum and difference frequencies between the carrier and
the LO. That is, LO + RF at 2455kHz and LO - RF at 455 kHz.
Neatly sidestepping a discussion of balanced and doubly-balanced
mixers, suffice to say that everything but the 455kHz signal will be
filtered out by the IF strip, with the result that the only signal
presented for conversion to the second IF will be the amplified and
filtered 455kHz signal.
But, back to the image signal...
If we call the carrier f1 and the local oscillator f2, then the
outputs from the mixer, for the case described above, will be f1,
f2, f1 + f2, and f2 - f1, like this:
SIGNAL FREQ
---------|----------|
f1 1000kHz
f2 1455kHz
f1 + f2 2455kHz
f2 - f1 455kHz
Since everything but the 455kHz signal will be filtered out by the
IF strip, one would think that would be the end of it.
Unfortunately, it isn't because if, at the same time the 1MHz signal
is being received, a 1910kHz signal (the image) is also present in
the mixer, it will beat with the LO and produce:
SIGNAL FREQ
---------|----------|
f1 1910kHz
f2 1455kHz
f1 + f2 3365kHz
f1 - f2 455kHz
Notice that f1 - f2 in this case will also produce 455kHz, which
will go sailing through the IF strip along with the desired 455kHz
signal produced by the 1000kHz carrier!
Once in the IF strip there's no getting rid of it, so the solution
is to have enough selectivity in the radio's front end in order to
attenuate it to the point where its products coming out of the mixer
are insignificant.
I've been away from radio for a while, so it may be that digitally
tuned "image traps" can now be implemented which will take some of
the load off of the front end in terms of selectivity.
In any case, to answer your question, since there will only be one
signal exiting the first IF it would be impossible to have an image
at the second IF.