One way of building a stable transmitter(in terms of frequency
stability) is to mix the signal for a VFO operating at a relatively low
frequency like 8MHZ or less with the signal from a crystal oscillator
at VHF. This allows setting to any frequency and the use of FM without
access to PLL chips.

The problem of course with this apporach is the tendency to radiate
unwanted mixer products such as the image or the high oscillator
signal. The mixer MUST be fed at least 10 db more high than low
oscillator signal or harmonics of the low oscillator will leak
out-almost always with one or more harmonics too close to the carrier
to stop. Furtunately this is a mixer problem people long learned to
minimize by limiting the level of the signal whose harmonics must be
stopped.

The use of a balanced mixer can shut up the high oscillator leakage,
but no mixer can in itself stop harmonics of the oscillator from
beating with harmonics of the high oscillator. In addition, if your
crystals won't oscillate on the 7th or 9th overtone and you have to use
a multiplier its unwanted harmonics can leak. You can stop even
harmonics with a push-pull multiplier, or odd ones witha push
push(input in push-pull, output in parallel) multiplier, but some
always wants to leak.

How do you solve this? THE SAME WAY OLD TUBE RADIOS GET I.F.
SELECTIVITY-the doubly tuned, critically-coupled circuit. This is
MUCH simpler to build than you might think.
It consists of two tuned circuits, inductively coupled to each other.
Due to a detuning effect when overcoupled, it is necessary to reduce
the coupling between them to the amount that just transfers full energy
from the primary to the secondary. In my experience with transistor
circuits, coils placed side by side and just touching work well.

Coils should have no more than 3 or 4 turns in the VHF range, and you
can squeeze or spread the turns to get exact resonance, or change caps
to if its really off. You tap the secondary coil at one turn for a
match to a bipolar, or couple a FET to the top of the secondary coil.
No trimmers or slugs needed, though if you don't mind buying a sack of
trimmers they make alignment easier at the price of additional board
space. Still, if I used trimmers for every circuit on one of my Gen X
exciter boards, there would be more than 10 on the board and it would
be much larger.

To align: first resonate the primary with the secondary disconnected.
Then connect the secondary cap and resonate teh secondary coil. Now
push the secondary further from and closer to the primary until you
find the point that just barely couples full energye into the
secondary, checking resonance at each step.

Three or four transister amps including the mixer output stage with
this arrangement will cut in-close spurs to such a low level that they
will barely move the s-meter on a receiver 15 feet from an antenna
energized with 60 watts of transmitter output power onb the carrier
frequency.