I would have to agree, that it is the reference sidebands of the LO that you
are seeing................it sounds like they are using a relatively wide
loop filter in the synthesizer, in order to achieve a fast settling time.

Pete

Richard Hosking wrote in message
. au...
Tom
What is the spectrum going into the mixer at the output of the crystal
filter? If this is OK, then it must be the second mixer that is at fault.
Look at the spectrum of the LO into the first mixer - does it have 5KHz
reference sidebands, and at what level?
What is the 1st mixer and post mixer amp? Are they able to cope with the
strong broadcast signal? If they are not something substantial like a ring
diode mixer and a strong post mix amp, then the IM might be happening
there.
A typical bipolar mixer might only be able to cope with -20dBm without
overload. If the RF gain doesnt help, then maybe it is the RF amp that is
at
fault, though I wouldnt expect all this to get through the crystal filter.
(though even crystal filters can be nonlinear if driven hard enough...)

Richard

Tom Holden wrote in message
.. .
I was startled when I hooked up a DX-394 radio (RX2) to the output of
the
2nd mixer of another DX-394 (RX1) to find that the strong MW station RX1
was
tuned to showed up not only at the 2nd IF of 455 kHz but at every 5kHz
from
around 200kHz to 30MHz! The strength varied with frequency, probably in
part with the gain vs frequency curve of RX2. There were stronger
signals
at
the 2nd and 3rd harmonics of 455kHz. Had to reduce the RX1's RF gain to
the
point where its AGC was inactive and the receiver was operating just
above
its threshold of detection to diminish the forest to just a few trees.

The 2nd mixer is preceded by a crystal filter at the 1st IF of 45MHz
+/-2.5kHz so I expected to see spectrum of about 15-20kHz width across
455kHz and some 3rd and higher order products like 910 and 1365kHz. The
5kHz
interval is suspiciously the same as the steps in the 1st LO VCO
frequency
(45.150 to 74.995 MHz), controlled by a digital PLL. The 2nd LO at
44.545MHz
+/-2.5khz is a varactor tuned VCO.

To generate products every 5kHz, surely we need a 5kHz signal or two
very
strong signals that are 5kHz apart going into the mixer or some
non-linear
stage after the 1st IF filter. Am I seeing something unusual or is this
what
I should see? If it's an anomaly, can anyone suggest what the cause
might
be? I'm thinking this is a fault or a design weakness that with more
complex
signals gives rise to a lot of interference to good reception. Could it
be
possible that this is the result of the method of observation and is not
really occurring in normal practise?

73 Tom