That is correct. Another thing that can cause that (2 X 2nd I.F.) image is
2nd LO leakage coming back into the input of the 1st mixer. Once the concept
of mixer theory is understood, things become clear. I know that many, if not
most of the folks on this group understand this, but for the new folks, I
will repeat it.
A mixer is a three port device. It doesn't matter if it is used as an SSB
detector or as the first stage in a superheterodyne receiver; the concepts
are the same. In the first stage of a receiver, the antenna input is applied
to the RF port of the mixer, the tuning oscillator is applied to the LO
port, and and the resultant signal is taken from the IF output port. Here is
where it gets a little tricky (not really!). If you could see the IF output
of the mixer on a spectrum analyzer, you would see three distinct signals.
You would see a lower sideband signal, the LO signal (suppressed by a
certain amount, depending on the LO feedthrough), and an upper sideband
signal. Both the lower sideband and the upper sideband will be separated
from the LO feedthrough signal by the intermidiate frequency. This means
that if you have a 455kHz I.F., the lower sideband will be 455kHz BELOW the
LO signal, while the upper sideband will be 455kHz ABOVE the LO signal. If
you have now preselection ahead of the mixer, the receiver will respond
equally well to both of those sidebands. This is why if you are tuned to
580kHz, for example, you can also receive a signal on 1490kHz. This is not a
defect of the receiver; instead, it is a characteristic of this type of
receiving system.
On lower frequency receivers, high-side LO injection is used. This means
that the lower sideband is the desired sideband. In this case, the upper
sideband is the image.
Now, in the case of the DX394, the 1st I.F. is 45MHz. The 1st LO operates
from 45 to 75MHz. This means that the receiver can conceivably respond to
two bands, depending which sideband (or band of frequencies) is selected by
the preselector that is ahead of the 1st mixer. The desired band of
frequencies is 0 to 30 MHz, but the upper sideband range of frequencies is
90 to 120MHz. This explains why this receiver can hear some FM band
responses in the 19 Meter band (15MHz LSB.................105MHz upper
sideband). This can be cured by additional low-pass filtering ahead of the
front end, of better shielding of the receiver from external signals.
In the case of an SSB detector, the 455kHz (or 450kHz in some cases) I.F.
signal is applied to the RF port of a mixer, the BFO signal is applied to
the LO port, while the audio is recovered from the IF port. You still have
an upper sideband and a lower sideband signal, except these signals are now
at an audio frequency. There are three ways to select the desired signal in
this case (there may be others).
The first way is to use a 455kHz BFO signal, and use ceramic, crystal, LC,
or mechanical filters ahead of the SSB detector. With a 2.3kHz bandwidth,
the LSB filter will typically be centered at 453.5kHz, while the USB filter
will be centered at 456.4kHZ.
The second way (and the more common way) is to use a filter centered at
455kHz, and use two different BFO frequencies of 453.6kHz and (you guessed
it) 456.4kHz. This is done either by using two different resonators in the
BFO circuit or by using one resonator and warping it with either a varactor
diode or a trimmer capacitor.
The third way (less common) involves using the phasing method to eliminate
the undesired sideband.
Finally, concerning that 1st mixer..................as far as I know,
there are only two ways to eliminate that primary image. The first one
involves using either a lowpass (when an upconversion scheme is used), a
bandpass, or a combination of the two.
The bandpass can either be a tracking type or a fix tuned type.
The second way involves using a Quadrature image-reject mixer, the way that
Drake does it with the R8 receiver. This technology has been around for a
long time, being used by Hallicrafters in their SSB transmitters (the
phasing method of sideband generation).
I hope all of this LONG explaination helps to clarify at least some of the
questions you might have.
Remember, just because a receiver is dual conversion..........it doesn't
mean that it will totally reject that 910kHz image. If you look at the
crystal filter specs, a 910kHz rejection factor is usually given. For a
single monolithic two-pole filter at 45MHz, this factor is usually
about -50dB, while a 4-pole set is usually about 60dB, and an 8-pole filter
is around 80dB. This "roofing" filter is what usually determines the
close-in IP3 rating, since the 2nd mixer characteristics are the dominating
factor here.
If this isn't enought, feel free to e-mail me directly, and I can explain
this stuff in more detail!
You're all a great bunch a folks!

Pete