This receiver was described in Radio News, August 1951,
by John F. Clemens, W9ERN.
Apparently, Mr Clemens thought he'd see how the Hazeltine Fremodyne
circuit could be
simplified without a reduction in performance. The result was a rather
cute looking FM reciever
constructed in a small steel box. Of course I had to try out the modified
circuit, and found not all
was as the article described.
The simplification involved removing a number of parts
associated with the quench waveform
control circuit, the three RFC's and using an untuned RF stage, thus
requiring only a single gang
tuning condenser. The super regenerative IF stage was changed to a
Hartley type oscillator.
Looking at the circuit we can see a resemblance to the
original Fremodyne. Firstly, the local
oscillator is using a 6C4, which is equivalent to one 12AT7 triode.
To eliminate the cathode RFC,
the oscillator has a cathode tap on the coil. Unlike the Fremodyne,
the oscillator runs on the low side.
The aerial coil is broadly tuned to the centre of the band and left
at that, so an extra tuning gang
is eliminated, and so are the hassles of tracking alignment. The received
signal is fed into the following superregenerative IF stage, along with
the local oscillator signal, and as in the original circuit, the two are
mixed to produce the IF at 21.75Mc/s. In the modified design, the super
regenerator is nothing fancy at all; just a simple self quenching circuit.
Quench frequency is dependant on the 500pF (C4) and the 560K (R1).
Like my own Fremodyne receiver, this one also included
a loudspeaking stage, using the other
half of the 12AT7 as an audio amp with a 50B5 output pentode.
Armed with enthusiam and curiosity, I set about building
this receiver with a few mods. Firstly, the AC/DC power supply was not
acceptable. The chassis could easily become live if C17 broke down, and
it's just too easy for a strand of wire, etc. to become lodged between
a component and the
chassis also rendering it live. This sort of dangerous design was very
common in the US where
the population seems to think nothing of exposed mains connections,
or the dreadful practice of connecting headphones to live chassis sets
without any isolating transformer.
Besides, 150mA valves like 35W4 and 50B5 are not standard in Australia.
There would also be
the question of operating off 240V mains, which increases the
shock hazard as well as
requiring a heat dissipating power resistor to drop the heater volts.
Obviously a transformer power supply was the way to go.
For the audio output, I had limited B+ current, and heater
current, so chose a 6AU6. I didn't
require much audio power and the additional gain of this valve would
also be useful.
Otherwise, I constructed the receiver to the circuit as
shown. For the tuning condenser I used a
modern plastic AM/FM gang with a 1/4" shaft attached.
Here's the original circuit:
Turn on time...and time to correct the poor design! First,
as usual inadequate quench filtering
had been provided. A 470pF from the 6AU6 grid to earth improved that.
The IF coil was the next problem. I was sceptical when
I saw that 24 turns was required. Experience taught me that for ~22Mc/s,
12 turns should be nearer the mark. But I wound the coil as per specs anyway
just to prove the point. Sure enough it wouldn't resonate anywhere near
22Mc/s, so I rewound with 12 turns and that one was fixed.
Having got this far, it was plainly obvious that the quench
frequency was way too low. It was clearly audible. I had to reduce the
500pF (C4) to 150pF and also do a major modification in the grid circuit.
Here, the 560K grid resistor had to be reduced to 27K! In view of this
drastic reduction of resistance, the grid resistor needed to be fed from
a voltage much lower than the B+; therefore I used an adjustable
voltage divider to set the optimum voltage (about 7V), so the SCA/stereo
subcarrier beat could be minimised while retaining as much sensitivity
as possible.
In view of the higher quench frequency used to minimise
the beat effects, I raised the IF to 28Mc/s.
The 3rd and 4th harmonics are 84Mc/s and 112Mc/s which are outside
of the FM band. The reason for choosing a higher IF is that there is an
optimum ratio between detector oscillation frequency and quench frequency.
The local oscillator wouldn't work as specified. I had
to set the cathode tap at 2 turns instead of one.
Ultimately, my version of the receiver ended up with this circuit:
By now the receiver was working quite well, and a sensitivity
test was performed. On AM, 5uV was receivable, but 25uV was required for
a reasonably clear signal. So, it appears to have slightly higher sensitivity
than the original Fremodyne. I didn't test it on FM, but the sensitivity
would be less.
Sound quality is not bad either; but it does not have that unique,
slightly mushy, sound the original Fremodyne puts out.
What really became obvious with this receiver once I was
using it, is the effect of an untuned RF input. Stations at the top end
of the band ~100Mc/s were being also received at the low end of the band
~88Mc/s. This appears to be due to local oscillator harmonics. When I was
using a Rhode & Schwarz SMS signal generator instead of the 6C4 local
oscillator, the effect was not there. It is possible that running the 6C4
oscillator on the high side will eliminate the problem. In any case, the
broadly tuned aerial coil does not seem to affect sensitivity.
Front panel, made of Marviplate. Speaker cloth and a calibrated dial
would improve the looks. A wooden cabinet is also on the agenda.
Chassis view. Layout is much the same as the Radio News design, except
of course mine has a power transformer.
Underneath shows a very uncluttered circuit, despite the small chassis.
Conclusion:
The concept certainly works, but I have to question the original design
as submitted by John Clemens.
I fail to see how it could have worked properly.
The IF coil: How on earth could the original receiver
have had a 21.75Mc/s IF? There is just no way that 24 turns of #24 ECW
on a 3/8" former can resonate at such a high frequency. Mr Clemens must
have had a faulty GDO if that's what he used to determine resonance. However,
the receiver would still have worked at the lower IF, but image rejection
would be worse.
The Quench: I think Mr Clemens must have had incredibly
poor hearing if the shrill squeal emanating from the speaker went unnoticed.
Given the huge change in component values required to get it in the supersonic
range, I doubt it was just a matter of component tolerances. It is true
that in 1951, with no SCA or stereo signals that the quench frequency
would not have to be as high as I made it, but that's no excuse for having
it clearly audible at all times.
The 6C4 oscillator: The original must have had an unusually
high gain 6C4. I certainly couldn't make mine oscillate until I increased
the feedback by changing the cathode tapping on the coil.
The Image rejection: This one can be excused as there
were probably a lot less FM stations where Mr Clemens lived, than there
are in Sydney now, so the problem might not have been evident.
Having said that, the design is one worth building if you're into simple VHF receivers, but you will need to make a few mods.