It sounds like there is a single pole (second order) filter/PLL system. The
whole system would have to have a loop cutoff of 1/20th to 1/50th of the
reference freq (100-250Hz) to get reasonable sideband suppression. You could
try looking at the waveform on the VCO control line - if the 5KHz signal is
more than few mV then there is probably something wrong. Try also putting an
attenuator in front of the RF amp - if the problem suddenly disappears with
increasing attenuation, then the problem is overload

Richard

The DX-394 comprises a 3SK195 RF amp, double 3SK195's for the 1st mixer
differentially driven by the r.f. amp and in common mode by the 1st LO,
45MHz crystal filter, 2SC27140 1st IF amp, 2SK210Y 2nd mixer, diode
switches for choice of two filters. The 1st LO PLL is a Fujitsu
MB87014APF-G-BND referenced to a 33MHz crystal all under a soldered
shield.
From the schematic, (I'm no expert at reading these) pin 12 (designated as
the output for an active type low pass filter) feeds what looks to be a
control for voltage applied to a varactor diode with a feedback path
consisting of a 33k in parallel with a 0.01 (M - mylar?) cap both in
series
with a 0.033 (M) cap. That would be the low pass filter and would control
settling time, right?

73, Tom