View Single Post
  #6   Report Post  
Henry Kolesnik
 
Posts: n/a
Default

John

I'm from a school that taught me that Q was a physical characteristic of a
coil and I still believe that.
Q = {2 * pi * f)/R Selecting the optimum sizes of wire, number of turns,
and diameter to maximize Q is well documented. However in this Miller TRF
circuit I see no mechanism for changing any physical characterisic of the
coils. There are many ways of reducing Q and I see none in the circuit.

I assume your neat little one stage mock up works and it's interesting that
L1/T2 and L2 are in seperate shielded cans but T1 is open. I'm curious as
to how RF is coupled from L1/T2 to L2 by T1. I've struggled to try to
simplify this simple circuit further but so far no cigar. Is it possible
for you to tune this to a known stations and then measure the variable
capacitor so that the Ls can be worked out from f = 1/(2 * pi * (LC)^0.5)

tnx

--
73
Hank WD5JFR

"John Byrns" wrote in message
...
In article , "Henry
Kolesnik" wrote:

Got any idea how it maintains constant BW as BW is a function of Q, a
relative constant and frequency which varies? Also I don't understand

your
notation "12uh centertapped" (3uh persection).
tnx


I have moved this response from "alt.binaries.pictures.radio" to
"rec.antiques.radio+phono" so that it will not be quickly deleted by the
server.

The following is my take on how these circuits work, if you don't like the
explanation consider that you got exactly what you paid for, as I thought
this explanation up all by myself, I did not find it in the RDH4, nor is
it handed down to me from the ancients.

I believe there are two ideas incorporated in this circuit. The first is
the idea of a tunable tank circuit whose Q, and hence bandwidth is
proportional to frequency, and the second idea is coupling two such
circuits, such that the coupling coefficient is inversely proportional to
frequency, to take advantage of the better shape factor that double tuned
circuits provide. If this could be done in practice we would have a
bandpass tuning circuit that would maintain constant bandwidth and
selectivity across the entire broadcast band.

Theoretically if we had perfect Ls and Cs with infinite Q, and if we
eliminated all shunt losses like diode detectors, antenna source
resistance, and coils with frequency dependent losses, we could build the
required tank circuits. A variable capacitor tuned tank circuit using a
coil of infinite Q, with the loaded Q controlled by a small series
resistance in the tank circuit will have the desired Q that is
proportional to frequency. At this point we could build a traditional TRF
type receiver using these constant bandwidth tank circuits alternated in
the traditional way with RF amplifier stages, making sure that we don't
load the tank circuits with any significant shunt resistance like a diode
detector, or an RF amplifier tube with a high input conductance. For the
detector we would use something like an anode bend detector, or reflex
detector to minimize the grid conductance. Of course in a practical radio
such a circuit is impossible, and can only be approximated, but we try to
do the best we can, accepting some broadening of the bandwidth at the
upper end of the band due to the inevitable shunt losses.

Since the response curve of each tank circuit is rounded, and when we
cascade several single tuned tank circuits the rounding and response roll
off increases, we realize that it would be a nice idea if we could couple
the tank circuits in pairs as is commonly done with the IF transformers in
superhetrodyne receivers to provide a better shape factor. For this to
work we need the coupling coefficient of the two coils to vary inversely
with frequency so that the product of "k" and "Q" remains constant vs.
frequency. Normal mutual inductance coupling as is typically used in IF
transformers won't work here because with mutual inductance coupling the
coupling coefficient remains constant with frequency. In a variable
capacitor tuned circuit what we need is a coupling reactance that is
independent of frequency, which will then cause the coupling coefficient
to vary inversely with frequency. There is not a real component that has
a fixed reactance vs. frequency, but we can simulate one to quite a good
degree of accuracy across the MW broadcast band by using an ordinary
capacitor in series with a negative inductor. The negative inductor acts
like a capacitor whose reactance increases with frequency, and when the
decreasing reactance of an ordinary capacitor is added to this decreasing
reactance, the result is a relatively constant coupling reactance across
the MW broadcast band, thus providing the desired decrease in "k" or
coupling coefficient vs. frequency. It should be noted that the reactance
of both a capacitor and a negative inductor have the same sign, which is
negative. Now the only problem is where to find the mythical "negative
inductor"? In the context of coupled circuits the effect of a negative
inductor is easily simulated by using a center tapped inductor where the
two halves of the inductor are closely coupled with k = 1, and connecting
the two tuned circuits to opposite ends of the tapped inductor, the
capacitor then goes in series with the tap, and we have the desired
result.

Now in the real world we find that we can't really build our perfect
series loaded tank circuits, and some shunt losses intrude, causing the
tank Q to not increase as much as we would like at the high frequencies,
which results in a somewhat wider bandwidth at the top of the dial. I
suspect that the designers of these sets made an effort to compensate
somewhat for this effect, by choosing Qs that made the bandwidth slightly
narrower than optimal at the low end of the band, and then tweaking the
values of the coupling reactances, the capacitance and negative
inductance, so that the circuit becomes slightly under coupled at the high
end of the band, tending to narrow the bandwidth, although making the
response more rounded, and causing the circuit to be slightly over coupled
at the low end of the band widening the compromise bandwidth a little at
the expense of a slightly humpbacked response curve.

That's just my take on how these sets were designed, and obviously there
are a lot of moving parts which probably were adjusted in different ways
by different designers with different tastes in design.

I await Patrick's take on how these so called "band pass" double tuned TRF
circuits actually work.


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/