Re: Filter problem

On Thu, 12 Jun 2025 17:02:59 +0100, liz@poppyrecords.invalid.invalid
(Liz Tuddenham) wrote:

john larkin <jl@glen--canyon.com> wrote:
>

On Thu, 12 Jun 2025 10:06:26 +0100, liz@poppyrecords.invalid.invalid
(Liz Tuddenham) wrote:
 

I am using a common-cathode double triode as a balanced mixer to
generate a frequency in the range 144 to 146 Mc/s fo a 2-metre
transmitter.
>
The inputs are a 150 Mc/s signal from a crystal oscillator and 6 to 4
Mc/s from a VFO.  (The crystal oscillator is modulated with narrow-band
FM by injecting a 90-degrees phase-shifted current from a reactance
valve.)
>
Both inputs are applied to the two grids in push-pull with the anodes
connected in parallel.  The150 Mc/s input can be accurately balanced so
that very little 150 Mc/s signal appears in the output.  The VFO signal
frequency is so far removed from the output frequency that accurate
balancing isn't needed.
>
Coming out of the anodes we have:
144 to 146 Mc/s wanted signal
150 Mc/s unwanted but at a low level
156 to 154 Mc/s unwanted, at the same level as the wanted signal.
>
I need to select for the 144-146 signal and reduce the 156-154 signal by
about 60dB.  Some of this selection will take place in subsequent tuned
stages but it really needs a filter to reduce the unwanted signals
sufficiently.  Some reduction at 150 Mc/s would also be desirable.
>
There are four possibilities which I have considered so far:
>
1)  Use a sharply-tuned circuit to select a single wanted frequency and
re-tune it every time the VFO is altered.  This means an extra operating
burden unless the two controls can be ganged, which is going to be a lot
of trouble to get right.
>
2)  Use a band-pass filter to select 144 - 146 Mc/s.
>
3)  Use a band-stop filter to remove 156 - 154 Mc/s (with possibly a
sharp rejector circuit to attenuate the residual 150 Mc/s).
>
4)  Use a low-pass filter, the 'skirts' of  which may also reduce the
residual 150 Mc/s sufficiently .
>
The second question concerns the physical form of the filter.  It could
be a ladder network  of coils and trimming capacitors in a die-cast box
or it could be made up of resonant lines or lengths of co-ax.  I don't
know of a resonant-line low-pass filter but someone might have come
across one.  There might be room in the enclosure for loosely coiled-up
co-axial cable resonators but trough-lines might be a bit too long
unless they are heavily capacitively loaded.
>
I have some ferrite toroids that could be used to match the valve output
impedance to the filter characteristic impedance.
>
>
Does anyone with experience of filter design have any recommendtions
that don't involve custom-made components or semiconductors?

 
What's the power level?

>
Milliwatts at high impedance.
>
>

You might make a bandpass filter out of commercial inductors and caps
and a few padders or trimmer caps. Build it on dremel'ed FR4 and bolt
it into the box for good grounding. One narrow deep movable notch
might help a lot.
 
https://www.dropbox.com/scl/fi/d7x21xc720gq0ztl8am5e/Z452.JPG?rlkey=v3lr01
34wjl5dliqdh2aowlpd&raw=1
 
Just scale that up 100,000:1

>
That's only one section, I am going to need a lot more than that because
the wanted and unwanted frequencies are only 5.3% apart.
>
>

The arithmetic is tedious.  My NORMA program might help.

>
I have found some calculations in the RSGB Handbook but my big worry at
these frequencies is unwanted 'components' in the form of wires (or box
sides) with inductance and stray capacitance everywhere.  If I make a
simple tuned circuit with an air-cored coil (or helix) across the width
fo a small die-cast box, the return currents from the grounding of the
capacitor at one end will flow back through the box wall to the
grounding point of the coil at the other end..  These currents will
spread out and may interact with a similar circuit layout at the other
end of the box to give unwanted coupling.
>
I had this happen many years ago with something that worked at 100 Kc/s,
so the problem could be much worse at 150 Mc/s.  (It also caught out the
designers of the RA17, hence the hacksaw slot most of the way through
the chassis.)
>

Heck, 150 MHz is almost DC.

>

If you are running low power, you could build a bp filter on one of
these:
 
https://www.dropbox.com/scl/fi/pym7yn95rqopxlkt7x6wk/Z368_BP_Filters.jpg?r
lkey=vijxk9kcqw0ve5iplw7p9h390&raw=1

>
I am avoiding the use of printed circuits, it is all being built on
tagstrips and standoff pillars - and a lot of the circuit can be
supported off the valveholder tags (but not the filters).

Retro look.