One question; why the predominance of a multitude of bandpass
filters on the output of TXs? Why not a tuned circuit, along the
lines of valve PA stages, once the impedance level were to be raised?

gareth wrote:
One question; why the predominance of a multitude of bandpass
filters on the output of TXs? Why not a tuned circuit, along the
lines of valve PA stages, once the impedance level were to be raised?

A bandpass filter IS a tuned ciruit and there is usually more than
one since a single, effective filter that covers 160 to 10 meters
is very difficult, if not impossible, with LC circuits to achieve.



--
Jim Pennino

In article ,
gareth wrote:

One question; why the predominance of a multitude of bandpass
filters on the output of TXs? Why not a tuned circuit, along the
lines of valve PA stages, once the impedance level were to be raised?

Practical answer: designers and manufacturers have probably determined
that this results in transmitters which are smaller, less expensive,
more reliable, and easier to use, than would be the case if tuned
circuits were used.

Fleshing this out a bit:

- In order to meet current FCC emission requirements, most
transmitters require multi-stage low-pass filters (5- and 7-element
seem to be fairly common). With a tunable filter, you'd need a
large number of wide-range adjustable reactances to tune all
stages properly.

- Tunable capacitors and inductors capable of handling 100 watts
or more of RF aren't small, even if you look only at the voltages
and currents present when they're properly tuned. Fixed-value
components (e.g. RF doorknob or chip caps, toroids wound on iron
cores) pack more value into less space, and I'd guess that even
when you need N set of them to handle N bands it's still a win,
space-wise.

- In order to avoid arcing if the transmitter is operated with
the output stage mis-tuned (and high voltages appear in the filter)
you'd need wider vane spacing in the air-variable caps... which
means you need larger (or more) plates and stators to achieve the
required maximum capacitance for operation on the lower-frequency
bands.

- A multi-stage tunable filter requires numerous moving contacts,
which are integral to the components (e.g. ground wipers on
capacitor rotors, rolling wheels on roller inductors). These
typically aren't sealed, get dirty, and can require cleaning.

Bank-switched fixed loss-pass filters usually use relays, which
are sealed and which are easier and less expensive to replace if
the contacts ever fail.

- You can bandswitch in an instant with a relay-switched fixed
filter. Tunable filters with continuously-variable reactances have
to have the caps and inductors "rotated" in some fashion to the
correct value - this takes time. You could build a tunable filter
with banks of switchable reactances in e.g. 1:2:4:8 combinations,
and switch these quickly, but that would involve even more relays
than a "one fixed filter per band" arrangement.

- As somebody else pointed out, it's not easy to build a single
tunable filter which works well down on 80 and 160 meters (needs
physically-large L and C to achieve the necessary values) and up at
6 and 10 meters (where you need small minimum values for these
components, and where wiring parasitics become a big issue). A
naive design for a tunable filter for this wide frequency range may
end up having some nasty parasitic resonances which will can let
the magic smoke out of your transmitter finals.

On 21/11/14 00:27, David Platt wrote:
In article ,
gareth wrote:

One question; why the predominance of a multitude of bandpass
filters on the output of TXs? Why not a tuned circuit, along the
lines of valve PA stages, once the impedance level were to be raised?

Practical answer: designers and manufacturers have probably determined
that this results in transmitters which are smaller, less expensive,
more reliable, and easier to use, than would be the case if tuned
circuits were used.

Fleshing this out a bit:

- In order to meet current FCC emission requirements, most
transmitters require multi-stage low-pass filters (5- and 7-element
seem to be fairly common). With a tunable filter, you'd need a
large number of wide-range adjustable reactances to tune all
stages properly.

- Tunable capacitors and inductors capable of handling 100 watts
or more of RF aren't small, even if you look only at the voltages
and currents present when they're properly tuned. Fixed-value
components (e.g. RF doorknob or chip caps, toroids wound on iron
cores) pack more value into less space, and I'd guess that even
when you need N set of them to handle N bands it's still a win,
space-wise.

- In order to avoid arcing if the transmitter is operated with
the output stage mis-tuned (and high voltages appear in the filter)
you'd need wider vane spacing in the air-variable caps... which
means you need larger (or more) plates and stators to achieve the
required maximum capacitance for operation on the lower-frequency
bands.

- A multi-stage tunable filter requires numerous moving contacts,
which are integral to the components (e.g. ground wipers on
capacitor rotors, rolling wheels on roller inductors). These
typically aren't sealed, get dirty, and can require cleaning.

Bank-switched fixed loss-pass filters usually use relays, which
are sealed and which are easier and less expensive to replace if
the contacts ever fail.

- You can bandswitch in an instant with a relay-switched fixed
filter. Tunable filters with continuously-variable reactances have
to have the caps and inductors "rotated" in some fashion to the
correct value - this takes time. You could build a tunable filter
with banks of switchable reactances in e.g. 1:2:4:8 combinations,
and switch these quickly, but that would involve even more relays
than a "one fixed filter per band" arrangement.

- As somebody else pointed out, it's not easy to build a single
tunable filter which works well down on 80 and 160 meters (needs
physically-large L and C to achieve the necessary values) and up at
6 and 10 meters (where you need small minimum values for these
components, and where wiring parasitics become a big issue). A
naive design for a tunable filter for this wide frequency range may
end up having some nasty parasitic resonances which will can let
the magic smoke out of your transmitter finals.







Question : Would a PI filter configuration involving an inductor and (a
combination of fixed and ) variable caps be OK for a single band
transmitter with solid state output device(s) ?

This question is posed because of the construction of a low power 3.580
MHz ARDF transmitter with a short vertical wire antenna. The PI filter
would be adjusted for max antenna current at the operating location.

Frank , GM0CSZ / KN6WH in IO87AT

In article ,
highlandham wrote:

Question : Would a PI filter configuration involving an inductor and (a
combination of fixed and ) variable caps be OK for a single band
transmitter with solid state output device(s) ?

This question is posed because of the construction of a low power 3.580
MHz ARDF transmitter with a short vertical wire antenna. The PI filter
would be adjusted for max antenna current at the operating location.

What you seem to need here, is actually two different functions -
impedance matching (the load is a "short monopole", probably
presenting a low radiation resistance and a high capacitive reactance)
and low-pass filtering.

Pi sections like this (or even the simpler L section) are often used
to perform the matching, in cases like this... a series inductor, and
a shunt capacitor on the transmitter side, would be what you'd want
here, I think. Having two variable caps (one on each leg of the pi)
would eliminate the need to have a variable inductor. You can
probably guesstimate the required value of the inductor fairly well
based on the calculated feedpoint impedance of a short vertical of the
length you'll be using.

However: although this is a low-pass network, my guess is that a
single pi section might not provide adequate low-pass filtering of a
typical Class B or Class C transistor final, to meet FCC
harmonic-and-spurious-emission standards. You'll have to ensure that
any spurious emission is at least 43 dB below the power of the
fundamental (current standard for new transmitters at that
frequency).

You can comply with this limit in any number of ways. The brute-force
way is to use multiple low-pass filter sections, sufficient to get
your worse harmonic (whatever it is) down below this limit. You can
use a more sophisticated filter topology which introduces notches at
one or more of the harmonic frequencies. Or, you can use a "cleaner"
transmitter design - e.g. use a clean sine-wave oscillator, and do all
of your amplification using high-bias (e.g. Class A) stages which
don't introduce much harmonic content.

In practice, you could breadboard your initial design, test it very
briefly, and look at the RF output with a spectrum analyzer to see
whether you need additional attenuation.

Probably the most straightforward way to add additional attenuation
would be to add one or more additional series-L and shunt-C stages
prior to your variable pi section... these could probably be
fixed-value components, if you figure that you'll have a fairly
predictable impedance looking into the pi section once it's properly
tuned up for the antenna.

On 21/11/14 18:58, David Platt wrote:
In article ,
highlandham wrote:

Question : Would a PI filter configuration involving an inductor and (a
combination of fixed and ) variable caps be OK for a single band
transmitter with solid state output device(s) ?

This question is posed because of the construction of a low power 3.580
MHz ARDF transmitter with a short vertical wire antenna. The PI filter
would be adjusted for max antenna current at the operating location.

What you seem to need here, is actually two different functions -
impedance matching (the load is a "short monopole", probably
presenting a low radiation resistance and a high capacitive reactance)
and low-pass filtering.

Pi sections like this (or even the simpler L section) are often used
to perform the matching, in cases like this... a series inductor, and
a shunt capacitor on the transmitter side, would be what you'd want
here, I think. Having two variable caps (one on each leg of the pi)
would eliminate the need to have a variable inductor. You can
probably guesstimate the required value of the inductor fairly well
based on the calculated feedpoint impedance of a short vertical of the
length you'll be using.

However: although this is a low-pass network, my guess is that a
single pi section might not provide adequate low-pass filtering of a
typical Class B or Class C transistor final, to meet FCC
harmonic-and-spurious-emission standards. You'll have to ensure that
any spurious emission is at least 43 dB below the power of the
fundamental (current standard for new transmitters at that
frequency).

You can comply with this limit in any number of ways. The brute-force
way is to use multiple low-pass filter sections, sufficient to get
your worse harmonic (whatever it is) down below this limit. You can
use a more sophisticated filter topology which introduces notches at
one or more of the harmonic frequencies. Or, you can use a "cleaner"
transmitter design - e.g. use a clean sine-wave oscillator, and do all
of your amplification using high-bias (e.g. Class A) stages which
don't introduce much harmonic content.

In practice, you could breadboard your initial design, test it very
briefly, and look at the RF output with a spectrum analyzer to see
whether you need additional attenuation.

Probably the most straightforward way to add additional attenuation
would be to add one or more additional series-L and shunt-C stages
prior to your variable pi section... these could probably be
fixed-value components, if you figure that you'll have a fairly
predictable impedance looking into the pi section once it's properly
tuned up for the antenna.

===========
David , Tnx for your explanation/advice , much appreciated.
I'll follow that up

Frank , GM0CSZ / KN6WH in IO87AT