My Construction of a Fremodyne FM Receiver

The following description is concerned with my last construction of a Fremodyne, in late 1996. I decided on the simpler two RFC circuit as per the "Meck", as experience showed that there was no difference in performance between it, and the three RFC circuit. An audio amplifier was included to form a mantel style FM receiver.

A simple plywood cabinet was constructed for the Fremodyne and its 8" speaker.

The Circuit.


The chassis was made from aluminium; its dimensions being 200 x 110 x 50mm.

Fig 4, below, shows the component layout of the Fremodyne portion of the receiver. As can be seen there is not much to it. Construction is basically a matter of connecting the
components correctly, keeping in mind the necessity of keeping leads straight and short in the VHF portions of the circuit. It is also advisable to use thick (eg. 18 or 20SWG) tinned copper wire for connecting between points due to skin effect. That is, at VHF and higher, current flows on the outside of conductors. Likewise, it is necessary to have a good groundplane. The 12AT7 pin connections assist with layout as the oscillator can be confined to one side of the valve socket, and the super-regenerator to the other, with the heater and earth connections in between.

The centre sheild of the socket should be earthed, and as for the socket itself, a teflon or ceramic one should be used if available, but a bakelite or wafer type seems to work just as so long as it is clean and not leaky. If the socket used is second hand, it would be worthwhile cleaning away the old flux with metho.
With a circuit of this type there will always be the question of parts. Fortunately there are always alternatives and it would be possible to build the receiver with all new parts. The audio
amplifer and power supply are the easiest sections to deal with first as they offer the greatest flexibility.

Audio Amplifier.
My audio stage is unconventional in that a 6AW8 is used with the output pentode triode connected. The pentode of the 6AW8 was designed for video output and although I've never seen data for audio output use, it does work well as a low power output stage. Incidentally, the European version of the 6AW8 , the 6DX8/ECL84 does have ratings for audio output use and performs like it, but it requires more heater current (720mA) than was available. The triode connection was necessary as I used a Jaycar MM1900 PA line transformer for the output transformer. Although the impedance ratio is ideal for this application, the inductance is lower than desired. When used as a pentode output transformer, the sound is shrill and trebly. The output stage is also unstable. Triode connected, all is well. Although triodes in output stages have lower sensitivity and power output than pentodes, for this application the performance is perfectly adequate. I imagine the reader would use something more conventional such as 6AV6, 6AQ5 or 6J7, 6V6 type of circuit as per a typical mantel set or radiogram.
I have included a circuit using a 6BM8 for those who want to use a more easily obtainable valve. It is a well tried circuit in many of my projects, but the M1100 PA line transformer specified is no longer available. Dick Smith used to sell these and they worked well as a valve output transformer. Unfortunately, the Jaycar MM1900, while it has the same impedance ratio, is not directly suitable.
You can try an MM1900 as per the circuit (ultra linear mode), but if there's excessive treble you will need to connect the 6BM8 pentode as a triode. To do this, remove the 2.2K screen resistor, and connect the screen to the plate. The MM1900 only has an 8R secondary and it will be necessary to phase the windings correctly or the amplifier will oscillate. If you have a proper valve output transformer,  then a pentode connection can be used. In this case, connect the screen to the B+ via the 2.2K resistor. If you have a square wave audio oscillator and oscilloscope, you can also optimise the feedback circuit. Feed a square wave in to the input and try different values of capacitor across the 4.7K feedback resistor. The value will be around 1000pF. Use the lowest value that removes the spikes from the square wave as viewed on the secondary when loaded by an 8R resistor.

Of course, one can save time and parts by using an existing set with P.U terminals, or as stated earlier, high impedance headphones.

Power Supply.
The power supply will be largely dictated by the audio stage since the Fremodyne itself only requires 100V at about 4mA. Again, any standard type radio power supply can be used as a basis. Some radiograms even have a socket to deliver B+ to an FM tuner, which could be used via a suitable resistor to provide power, with the P.U terminals providing the audio stages. To cater for those who don't have any valve power transformers, back to back low voltage transformers can be used as shown here.


It will be necessary to check the supply voltages and adjust the filter resistors accordingly with the particular power supply used. It is easiest to adjust the main filter resistor first to get the audio stages working correctly, then adjust the second filter resistor so the Fremodyne is supplied with 100V.

The Receiver.
Now to the actual receiving part. The IF coil is not critical. Something with a former around 1/4" diameter with a ferrite core suitable for 30Mc/s operation is required. I have used a modern Neosid 5mm former and slug with complete success. It may be possible to use a slug tuned 455Kc/s transformer but any ferrite cups or rings will need to be removed. For the 5mm or 1/4" formers, 12 turns of 36 B&S enamelled copper wire is wound on. As there is only one winding it is easy to alter the number of turns if resonance if the IF cannot be obtained.
As for the RF chokes, all three of them consist of 100-120T of 32 B&S E.C.W on a 5/16" former such as part of a ball point pen casing, a Rawl plug, or a high value (>100K) carbon resistor. For those who don't like winding coils, it is possible to use commercially available 15uH chokes. The value of inductance calculated from the choke dimensions turned out to be very close to their measured value and operation of the prewound chokes has been confirmed in practice. A receiver constructed with these minature RFC's is also a lot neater.

Next we come to the aerial and oscillator coils. They are both 7mm inside diameter, wound with 3.5 turns and 3 turns of 18 B&S tinned copper wire, for the aerial and oscillator coils
respectively. The trimmers across the two coils should have a capacitiance range in the region of 2-20pF and are only necessary if not provided for on the tuning gang, which incidentally is likely to be the most obscure item.

Tuning Condenser.
This is a dual gang unit with a capacitance range of 2-15pF. Higher capacitance values are usable but will give a greater tuning range. The gang in my latest receiver was obtained at a HRSA meeting, but there are several other solutions to obtaining this item.
One could take the approach of using separate single tuning condensers as per 1920's technology. A pair of reaction condensers would be a good start but the capacitance is likely to be around 100pF, and therefore series condensers [ try 33pf] will be required to prevent the FM band being crowded together at one end of the tuning range. Unless the two condensers are mechanically ganged together, however, tuning will be difficult, because peaking the aerial condenser once a station is tuned in, will then affect the local oscillator condenser setting. If a dual gang cannot be obtained, it would be better to leave the input broadly tuned, and only tune the local oscillator as shown here and here.

One butcherous method used in some home made sets for VHF reception was to remove plates from a standard broadcast gang. Typically two or three plates would be left on the rotor section. It is not something I recommend these days with vintage components. I have also used with success a modern plastic unit for replacement in AM/FM radios. The FM sections are about 20pF and have trimmers. These units require an extension shaft to allow use of a normal 1/4" knob. This consists of a piece of pot shaft with a 2.5mm screw through the middle attached to the tuner shaft. Panel mounting is by two short 2.5mm screws and care must be taken not to let them protrude inside the case and damage the plates.

Having assembled the receiver it should be turned on and adjustments made to the power supply voltages if necessary, ensuring the Fremodyne part has 100-110V. A rushing sound
should be audible at this stage to indicate the super-regenerator is working. The next step is alignment, which is done the same way as with any other superhet. First, feed an AM signal of 21.75Mc/s into the aerial terminals and adjust the IF coil for a peak in signal, reducing the signal generator output as necessary. If the coil won't peak at 21.75Mc/s this would indicate that turns would have to be added or subtracted from the IF coil. If the coil peaks up at a higher frequency it has insufficient turns and vice versa.
The local oscillator is aligned next. By means of the local oscillator trimmer at the high end of the band (108Mc/s) and by contracting or expanding the turns spacing of the oscillator coil at the 88Mc/s end of the band, ensure that the reciever will tune from 88-108Mc/s. Again, turns may need to be added or subtracted if there is not enough adjustment available.
Finally the aerial coil is set up. With the receiver and signal generator on 88Mc/s, expand or contract the aerial coil for maximum sensitivity. Tune to 108Mc/s and use the aerial trimmer to peak up for sensitivity again. Both the oscillator and aerial stages should be aligned once more due to interactive effects. Note that the aerial tuning is very broad and adjustment is not critical.
In fact, one version of the Fremodyne eliminated the tuning circuit altogether, but this resulted in poor image rejection.
For those without a signal generator, adjustment is still possible. Tune a shortwave receiver to 21.75Mc/s and place it near the Fremodyne. Adjust the IF coil for maximum noise on the shortwave receiver. Then, use off air stations to set the local oscillator. E.g.; for Sydney check that 2RDJ (88.1) is receivable at the low end and 2SER (107.3) is receivable at the top end. Adjust the aerial coil as before.
The easiest way I found to align the aerial and oscillator coils was to use a G.D.O. Simply ensure that the oscillator coils resonates betwen 110-130 Mc/s while at the same time the aerial coil resonates at 88-108Mc/s.

First thing evident with a Fremodyne, or any other superregenerative set on the FM band, is the whistle audible in the background on some stations. This is a result of the stereo and SCA subcarriers beating with the quench frequency. It can be annoying or not noticeable depending on the station. Obviously the problems does not occur when receiving mono stations. It is important to realise that stereo FM did not exist prior to the 1960's and therefore this problem did not exist with the original Fremodynes. Some improvement can be made by increasing the quench frequency, by reducing the 150K resistor, at the expense of sensitivity. Also reducing the .0047uF across the 150K can make a big improvement.  Interestingly, this problem is not evident with Hi-Z phones, probably due to limited bandwidth.
A problem that existed with some stations was the SCA subcarrier at 67Kc/s. Thankfully, most stations that were transmitting this service have since dropped it as the internet has since taken over such things. I recall it being quite problematic at times from when I started experimenting with super-regenerative detectors in 1987. By the early 2000's the problem ceased to exist, and the otimum quench frequency is now only dependent on the stereo subcarrier.

Unfortunately, sensitivity is poor. At least 100uV of signal is required for good reception. The original Hazeltine specification was 200uV. So, this is not a set for DX, but it will bring in all the local stations without any problems. An outdoor FM aerial is essential for peak performance - the receiver performing very well for what it is. I don't recommend VHF TV aerials, unless they are wideband, or cover Ch 3,4, or 5. Even then, the wideband designs, such as log periodics, generally do not have a gain as high as that of a dedicated FM aerial. Some low power community stations are then receivable. I have even tried the set in a car powered off a vibrator inverter, and using the existing broadcast aerial. It certainly worked around Sydney quite well, but I wouldn't recommend replacing the existing car radio with it! An RF amplifier stage could help in regards to sensitivity. Experiments with a VHF TV distribution amplifier seemed to indicate potential here.
Where this set really does lead over ordinary superregenerative receivers on FM, is that the sound quality is superior and remains constant across the band, and with weak and strong signals. Tuning is also less critical.
For its day, the Fremodyne certainly acheived  its aim as a low cost FM receiver of acceptable quality.

[Further notes: Recent super regen experiments seem to indicate that the quench frequency/subcarrier beat is improved by using an active filter with a cut off around
8000c/s. This was used in my mains operated 12AT7 receiver.
There are more stations on the air since writing this article, but due to the Optus satellite,
there is less SCA in use. It appears that at the present, 2SER, 2CBA, and 2WS are the only
Sydney stations using it. The Fremodyne set that this article was written about has been
in use regularly since late 1996 for reception of stations which are about 80km line of sight from my home. Reception is quite good depending on signal strength. Unless the listener had been told, it would not be evident that a slope detection super regenerator was being used.
The SCA and stereo subcarrier beat is far less evident with the Fremodyne than other super regenerative receivers used for FM.]

August 2004: From an internet forum, it appears that someone has done a Spice simulation of the Fremodyne circuit and claims to have increased the sensitivity to 125uV as well as eliminated the SCA/stereo subcarrier problem. The sensitivity improvement has been acheived by using 6CW4 Nuvistors. The subcarrier beat problem was "fixed" by raising the quench frequency, apparently by reducing the 150K to 33K and supplying this resistor from a source of about 25V (same bias current therefore) rather than the 100V supply.
I gave up trying to post the link here; anyway it probably wouldn't have lasted long. I don't have any further info on this design mod unfortunately.
It was very interesting to see mention of this, despite the lack of information. Of course no real working model has been built using the modified I've said elsewhere, home building of electronics, especially RF circuitry. is almost dead. However, it raises some interesting points. For a start, it would be pointless using a 6CW4 for the local oscillator, if this was so done in the Spice simulation (unless you've got a good collection of them). I wonder how much gain is lost by raising the quench frequency and if the gain of the 6CW4 is enough to offset this, as well as provide the extra 75uV of sensitivity. It is a surprise that simulation software can emulate the complexity of such a circuit to start with.
I am keen to try the mods however, but using a 6ES8 twin triode instead of the 6CW4's as this more easily obtainable, has the same pin connections as the 12AT7 (except for the heater) and is about the highest gain VHF triode you can get, being of frame grid construction.

February 2005: I've recently rewired my FM aerial distribution system to provide extra outlets in my kitchen and garage. Previously it was a collection of splitters and an amplifier (15db) to provide four outlets. Now I have a six way splitter with no amplifer. Of course the signal level is less (but still sufficient) at each outlet. Interestingly this has resulted in a considerable reduction of the SCA/stereo subcarrier beat problem. This is what I've also found with the various 12AT7 receivers. So the key to successful Fremodyne operation with stations transmitting stereo and/or SCA is to keep the signal level to the lowest useable level. 2WS was a bit of a problem previously but now very nice to listen to.