Paul Sherwin wrote:
On Tue, 08 Jun 2004 16:11:02 GMT, "Frank Dresser"
wrote:
Getting wideband IF transformers will be a real problem. I don't know of
any NOS sources for them.
You can easily reduce the frequency selectivity of IF transformers by
adding resistors in parallel, though this will reduce sensitivity.
The typical impedance of an undamped 455 kHz undamped
IFT is between 20k and 50k at 455 kHz.
Adding some R to both coils reduces the load seen by the tube,
hence its gain drops because pentode IF amp tubes have a high Ra,
and gain varies with load.
So the gain of the IF amp drops maybe 6 dB with R loads to both LC circuits in
IFT2,
and gain drops the same amount in IFT1, powered by the F converter tube
The nose of the selectivity broadens, ie, the Q of the circuit reduces,
ie, the bandwidth passed by the IFT is broadened out,
but 50 kHz away from resonance the attenuation rolls off at 12 bD/octave.
The roll off of a typical single tuned LC IF circuit away from the pointy nose
shape of
the the curve is only 6 dB per octave. The profiles of typical response curves
for RF and IF
coils are illustrated plentifully in all the good old radio books.
So with damping R, and two IFTs, the amount of attenuation of signals only 50
kHz away from
the wanted station is reduced by at least 12 dB.
This may perhaps be enough to allow a station 50 Khz away to be heard in the
backgound of a wanted station,
especially if its one thats putting out 5,000 watts and the wanted station is
putting out only 300 watts,
and they are both within 10 miles of the receiver.
Therefore its important to have some selectivity, although quite broad,
ahead of the converter tube.
I use two low Q LC circuits in cascade which are slightly tuned apart
at the low end of the BCB so effectively broadening the RF bandwidth, but
enabling a steeper roll off away from the pass band.
At the top end of the BCB, the two input RF LC circuits are very nearly tuned
at the same F,
and since the Q is still low, but the Fo is higher, the pass band does not
cause
side band cutting and a reduction of RF bw which would then limit the audio
after another 4 tuned circuits in the IF stage.
To use TRF to do the same thing would be almost impossible, and
I would need at least 6 tuned circuits tuned in the same way, and a six gang
tuning cap,
along with a seventh gang to tune the oscillator. There would have to be two
low gain
IF amps, which could be cascoded triodes instead of pentodes.
Its a hell of a lot easier to do it all with a superhet.
Not many NOS IFTs.
The old ones seldon suffer from spending 50 years in an old radio set,
and they are actually fairly ruggedly made, with brass tuning shafts for the
ferrite cores,
and in cans which have kept out the pollution failrly well.
The coils are often pie wound coils of litz wire.
The distance between the coils determines the amount of magnetic coupling, and
most IFTs
have just the right distance to cause critical coupling which gives the flat
topped
bandpass characteristic so you get about 10kHz of BW from a typical 455 kHz
IFT.
This allows 5 kHz of audio.
Two IFTs of the same response will give 7 kHz of BW, which allows
3.5 kHz of audio BW.
Now the minute one cuts the single tube the IFT coils are mounted on and
moves them closer together, say by 5 mm, the magnetic coupling increases,
and the response usually widens, but not greatly, but the shape of the
response
becomes twin peaked either side of Fo.
If you have a twin peaked IF response it means the audio BW will be also
peaked up at say 4 kHz, before rolling off even more sharply than it did
before when the
response was flat.
But sometimes the first IFT1 is deliberately slightly overcoupled to give the
twin peaked response,
which then is compensated back to being flat by the following normally single
peaked response of
IF2.
But tuning could be strange, with a tuning indicator having to be set to the
slight
null between two peaks.
Alignment of the IFTs becomes more difficult.
This is why I suggest that an IF of 2 MHz be used instead of 455 kHz, because
for the same Q the pass band of say 3 normally critically coupled IFTs
would be nicely flat topped, but still have an overall wider bandpass than
two 455 kHz IFTs.
The would have to be two IF amps instead of 3, but their gain need only be
low,
so cacoded triodes come to mind.
The cascoded triode has an effectively very high
Ra looking into the anode of the top tube, and a 12AT7
would have Ra' = 1 Mohm. If RL was 20k, gain would be about 60.
12AU7 would also be OK with Ra' = 200k,
and gain about = 29 with cathodes fully bypassed.
But pentodes could be used, with 6BA6 as IFamp1, with AVC applied,
and 6AU6 as IF2, with no AVC applied to keep the final IF amplification
as linear as possible.
Distortion of the IF envelope shape will all be detected as audio distortion
to
the shape of the recovered audio at the diode detector stage.
It would be possible to perhaps simply remove turns from a 455 kHz
IF coil and halve the existing capacitors to raise the Fo to 2 Mhz.
This all has to be done carefully, so that after halving the cap size,
just the right no of turns are removed to get the IFT to tune to 2 MHz with
its tuning
slug in the middle of its travel range.
I have never done this, so perhaps its just easier to wind ones own
new IF coils, but large sized old ones with cans of 35mm dia are plentiful.
The tiny IFTs which became prevalent in radio sets in the 1960s are a PITA
to modify.
The use of 2MHz IFTs requires strict adherence to using shortest leads
from tubes to IFT connections, because the higher the F, the greater the
likelyhood
of oscillation and IF amp instability.
So the IFT and tube line up will be in a neat straight line, with small 7 pin
tubes being able
to be close as possible to the IFT cans, and perhaps with additional grounded
sheet metal shields up off the tube sockets.
Patrick Turner.
Best regards, Paul
--
Paul Sherwin Consulting http://paulsherwin.co.uk