This miniature receiver is based on a design I created in 1987; it was the first super regen set that was easy to use and had decent sound quality. It did have a particularly bad shortcoming in that its performance dropped off severely at the low end of the VHF broadcast band. In more recent times, (2002) I wanted a small FM receiver to use as a portable. Remembering the simplicity of my earlier design, I reconstructed it and solved that shortcoming. Additionally, I elaborated on it so that the RF amp would perform as an audio stage to drive headphones; i.e.. reflexing. Current consumption of my new receiver is low with the B+ drawing about 3mA. Heater current is .15A or .3A depending if the 12AT7 heater is wired for 6 or 12V. Battery operation is therefore practical, so I also made up a vibrator supply to run the whole set off a 6V SLA battery. Total consumption at 6V is about 600mA, so dry cell operation could be contemplated for short periods of time. The B+ drain is low enough to give reasonable life out of 15 nine volt batteries in series if you wanted to take that path. In that case the 6V battery would only need to supply 300mA which is about the drain of a torch bulb.

Circuit Description.

To enlarge the circuit diagram, click here.  The heart of the receiver is the right hand side triode. This is the super regen detector. By means of the 15uH cathode choke and the internal grid and cathode capacitances this triode oscillates at whatever frequency the grid circuit is tuned to. This is the same frequency as the station we wish to receive. Grid leak bias is obtained from the 330K and 33pF in the grid circuit. The coil is 4 turns of 18 gauge tinned copper wire, air cored with diameter of 3/8".
For the tuning condenser, a 15pF unit is the best choice, but as I've got a few MW broadcast units with a 100pF oscillator section, I used one of those with a 33pF series condenser to give complete coverage of the 88-108Mc/s band.
  So far, we have a VHF oscillator, but in order to super regenerate, it needs to go in and out of oscillation at a supersonic rate. The grid RC network values are chosen for this so the valve cuts off after a certain time whereupon oscillation recommences. See my Fremodyne article for more info about super regen operation.
  My breakthrough with this receiver design was with the regeneration control. The usual way is to vary the B+ to the detector. For most sensitivity, the detector is adjusted so that it has just started to oscillate. However, I found sensitivity and sound quality to be very poor towards the 88Mc/s end of the band. Examining the quench waveform on the CRO revealed how obvious it was. I suppose more by accident than anything else, I discovered that by making the grid go more negative the performance picked right up and performance was consistent right from 88 to 108 Mc/s. I also noted that the variable B+ type of regeneration control had little effect so it was dispensed with. Instead, making the negative grid supply variable gave far more control.
 The B+ supply was found to be optimum around 157V. Likewise the grid supply needed to go up to about -30V. However, these figures are not carved in stone and will and do vary for other clones of this receiver. For instance in one receiver I found the B+ should be 140V. The regeneration control should be able to cut the receiver off altogether; if not increase the -30V supply or reduce the B+.

Back Bias
Note that the supply shown on the circuit is 180V. This is between the B+ and B- terminals. Back bias is used, meaning there's a 30V drop across the 10K resistor from B- to earth. This provides the -30V supply. The plate supply is still 150V.  (180-30)=150.

RF Stage
   While the aerial can be connected directly to the 4.7pF input capacitor, some problems arise by doing this. Firstly, loading by the aerial can cause the oscillator to be unreliable, stop oscillating, and go off tune. This is a major problem especially for a portable set as you can imagine the aerial is going to be shifted around, moved near other objects etc. Secondly, the oscillation from the super regen detector will be radiated by the aerial. In my opinion that isn't a problem...who cares if other receivers are interfered with...it's the one you're listening to that counts isn't it?
  A grounded grid RF stage is used here. Although it has hardly any gain, it's stable, can be untuned (avoiding annoying two gang tuning condensers and getting them to track), and has excellent isolation.
The cathode and plate loads are 15uH chokes. Aerial signal is fed into the cathode via 1000pF so as not to upset the bias if the aerial has DC continuity to earth. The grid is RF bypassed also by 1000pF.
From the triode plate, the signal proceeds into the above mentioned 4.7pF condenser.
The value of the RF coupling condenser should be as large as possible to transfer maximum signal into the detector without it becoming difficult to oscillate. 4.7pF seemed to be about right.

Reflexing and the Audio stage
   Not mentioned so far, to avoid confusion, is how the RF stage also functions as the audio amplifier.
In the normal way of things, another stage of amplification would be needed to drive headphones from the detector triode. However, we can make dual use of the RF amplifier stage instead! Because audio frequencies are so far apart from VHF we can amplify both and keep their paths separate.
  Let us now examine how this is done in detail. For the audio signals, the left hand triode functions as a normal grounded cathode stage. The audio signal from the 56K detector plate resistor is fed into the 1M volume control in the usual way, with the wiper then feeding the grid on the other half of the 12AT7.
As well as making the grid at earth potential, as far as RF is concerned, the 470pF also bypasses more of the quench signal that is still present in the detected audio, but such a low capacitance does not bypass the audio due to the lower frequency components.
  In series with the plate supply is the output transformer. This has no effect on the DC conditions of the triode functioning as an RF amplifier, but because 1000pF is not a very high value of capacitance at AF, the audio signal can pass through into the transformer and drive the headphones. The 15uH plate choke is such a low inductance at audio frequencies it has no effect.
  Bias is obtained with the 1K cathode resistor. Again the RF choke has no effect at audio due to its low reactance. Cathode bypass is effected by the 1000pF for RF and 25uF for audio. Electrolytic condensers are poor at RF, hence the inclusion of the ceramic bypass.
  The speaker transformer I used is a N.O.S replacement for an AWA P1 TV. As far as I can make out it has an impedance ratio of about 8K to 15R. It is not critical. For more efficiency, the plate winding should be of higher impedance. A 100V PA speaker transformer is a good choice with the 10K or greater tap selected. There is no reason why ordinary high impedance phones cannot be used instead, eliminating the transformer altogether. If you do this, watch the polarity, or the phone magnets will lose their strength, and see that their insulation is up to the 140~180V this receiver operates off. The application of 180V to the listener's head would result in a rather interesting facial expression don't you think?
  If reflexing is not required (e.g.; non portable set with additional audio stages), the RF amp is easily modified. On pin 2 of the 12AT7, remove the 470pF and connect to earth. Remove the 25uF cathode bypass. Connect the 15uH plate choke/1000pF connection straight to B+ with no speaker transformer. This circuit shows the non reflexed version.
The wiper of the 1M pot can then connect to the following audio stages. See the mains operated version for an example of a complete FM radio with speaker using this design.
  Note that for high power output stages driving a speaker, more quench filtering will need to be provided to prevent overload at high volume. I have used a Sallen Key filter, based around a 12AU7 or 12AX7 with a cut off frequency of 8000c/s with complete success. Generally one or two triode stages will be needed to fully drive the output stage. Postscript May 2008: A Sallen Key filter has found to be unnecessary in practice and the simple RC filtering used in my later 12AT7 receivers can be used. However, the SK filter does do a better job and if you can make room for the extra circuitry it is worth considering.
Components
  Anyone who is capable of building this would be able to work out what ratings of the resistors & condensers are. Suffice to say, use ceramics for all the RF bits, and electros for the B+ and audio. The .01 volume control coupling should be polyester or polystyrene or a decent ceramic that doesn't leak...designers of those red topped Ducons take note! To be realistic you will get away with 100V rated ceramics in this circuit even though some might have ~150V across them. Obviously with only 3mA current consumption, all resistors can be 1/4W. The chokes are 15uH things wound on ferrite cores. They are a DSE catalog item at present (2003) but Jaycar and WES also have them. Postscript May 2008: The RF chokes from DSE are no longer available and the Jaycar and WES types are unsuitable. I now recommend home made chokes which work better anyway. See the notes here.

For the tuning condenser you can take the easy way out and use varicap tuning, but you'll need a regulated supply and the current consumption is added to. A good varicap diode is the BB105. The best regulator for varicap tuning is the Philips TAA550 IC which is like a very stable 33V zener diode. See the notes on the Model T Ford car radio if you want to use varicap diodes.
I prefer using a proper mechanical variable condenser as it has a higher Q and has less drift problems.
See my notes on the Fremodyne. The resistor values that shouldn't be departed from are the 56K plate resistor, the 3.3M and 330K grid resistors, and the 1M volume control. For the condensers, don't change the 1000pF bypass on pin 6, or the 33pF on pin 7. Otherwise, near enough is good enough. Having said that, if you go and do dumb things like put a .01uF on pin 2 instead of the recommended 470pF, don't email me and ask why the sound is so muffled.

Vibrator Power Supply

Click on the circuit to see it enlarged.

  For portable operation I have used a conventional non synchronous vibrator inverter. The transformer is a conventional 12.6V CT 300mA to 240V unit which is cheap and easy to get.
The transformer feeds a bridge rectifier to provide the DC for the receiver. The 8.2K drops the voltage to about 180 as well as providing filtering. . Of course you could use a proper vibrator transformer with centre tapped secondary, allowing a synchronous vibrator to be used. However, it will need to be of the split reed type to allow the back bias to be created.
  The vibrator I used is a Plessey 614. For AWA types (such as V5105, V5124, V6606 etc.) you will need to provide 6V to the reed connection as these are series driven unlike the Plessey, Ferrocart, and most U.S types. (You will need to check the B+ to your receiver in case 8.2K is not a suitable value for your particular power supply)
  The buffer condenser suited my particular transformer; you will need to check for yours. It is selected to be the minimum value that removes the overshoot on the square wave as measured on the transformer secondary. If you have no CRO, the next best way is to select the value that gives minimum current consumption at 6V as well as least contact arcing. Note that the voltage rating should be at least 630VDC for this condenser, or 250VAC.
  The .047uF's from each contact to earth were to reduce the RF interference. There is room for experimentation here as not all interference suppression measures will work for one vibrator supply as they do for another.
  I used a pair of 6V 4Ah sealed lead acid batteries. They should give about 10 hours running. In reality, I don't expect anyone to be building this supply, but it's here for interest.

Charger
  Included was a circuit to charge the 6V battery off a 12V supply (I use a solar panel). It is important to charge lead acid batteries correctly or they will be ruined. It is also important not to discharge them either or they will be ruined...permanently. It's awfully sad to see batteries sitting on a shelf discharging as their owner plans to 'use them for something someday'. Once discharged, they're damaged for good! If you have lead acid batteries not being used, do keep them charged.
  Unlike Ni Cads, lead acid batteries are constant voltage charged. This means connecting a current limited voltage regulated supply to the battery until the voltage comes up to 13.8-13.9V for a 12V battery or 6.9-6.95V for a 6V battery. Don't let a 12V battery get below 10.8V (6V down to 5.4V) and charge as soon as possible after use.
  To prove the point I've had batteries last for over 10 years simply by looking after them this way.
  After that sermon, let's get back to my little receiver's charger. Voltage regulation is achieved by an LM317 regulator. This is connected to the battery via a diode to prevent discharge back into the regulator. By means of the resistors around the regulator, the voltage is set to required charge voltage (6.9V). To check this, with no load on the regulator (i.e.. radio off and battery disconnected), measure the voltage on the cathode end of the diode. If more than 6.9V, reduce the value of the 560R; if less than 6.9V, increase the 120R//2.2K. Due to component tolerances I can't guarantee my choice of values for other chargers.
  Current limiting is performed by the 18W lamps. I used festoon style car bulbs soldered between a pair of three lug tagstrips. Reverse polarity is performed by the shunt diode. If wrong polarity is applied this conducts and the lamps light at full brilliance. Note that the LM317 can only provide 1.5A max. so it is undesirable to reduce the current limiting by the lamps as this will not charge the batteries any quicker. In any case you should check the battery rating as to what the max. charge current can be; usually 2A is pretty safe.
  NiCad cells can also power this receiver ( you will need five in series), but don't use this charger.
To be of any practical use, you need an Ah rating of at least 4Ah for each cell. Charging NiCads is a real pest and their reliability can be poor (that seems to vary a lot as to who made them...no name ones are often very poor quality). You either have to make a complicated charger that senses the dip in voltage when they've reached full charge, or a timed 14hr constant current (.1 x Ah capacity) charger. Trickle charging can be done by permanently feeding in a constant current of .01 x the Ah capacity. NiCads are also a bit funny about being connected in parallel so I don't recommend doing that to increase the Ah rating.

Other Power Supplies and the Regeneration control.

  As shown, the original circuit requires 6.3V@300mA (or 12.6V@150ma with heater rewiring) for the A+ or heater supply. The B+ requires 150V@3mA and the C- or bias supply needs -30V@60uA.
However, providing a negative 30V supply is not always convenient with certain types of power supply, such as a synchronous vibrator type or where a separate power supply is used that has it's B- terminal earthed. So, to make the receiver work with only positive voltages, I redesigned the regeneration control circuit to allow this.
  Basically, by using cathode bias, we can eliminate the need for a negative supply. By taking the cathode positive, with the grid at earth potential, we get the same result. Cathode positive with respect to grid is the same as grid negative with respect to cathode.
  So, by varying the cathode voltage in a positive direction we can adjust the regeneration. The range required is small. 0 to 3V is sufficient. So why then did we need 0 to -30V before? That's because the bias was fed into the grid via a 3.3M resistor. This with the 330K grid resistor acted as a 10:1 divider.
Here's the first regeneration control I tried in the cathode circuit:

Preferred method of regeneration control. Replace the 25uF with 1uF, and if necessary change the 5K pot to 10K if control range is insufficient.

A 5K pot is used to vary the bias in the usual way. The 25uF is to bypass the audio and quench component while the 1000pF bypasses the RF. The bottom end of the RF choke is therefore earthed except for DC which is variable from 0 to 3V. As far as receiver performance goes, there is no deterioration in sensitivity or sound quality from modifying the control thus. What did become evident was a slight time lag from when the regeneration control was adjusted to when it took effect. It's only a fraction of a second but it's noticeable. If that is not of concern, then by all means use this method of control. My guess is that it's the time constant created by the 5K pot and 25uF that cause this characteristic. This circuit has the advantage of eliminating the C- supply and does not add to the current consumption. Postscript May 2008: After work on the Model T Ford car radio, it became clear that this is the preferred kind of regeneration control. As it turned out, the 25uF was unnecessarily high and 1uF is sufficient. There is no time lag problem then. It was also found a 10K pot may be necessary, possibly shunted with another resistor to spread out the adjustment over the pot's full rotation. The next circuit is now superfluous but shows another method:

To improve upon the "time lag" effect, it would seem necessary to reduce the cathode resistance or decrease the bypass. The latter is not acceptable as the performance could be severely compromised. So, we use a lower value of pot, and put extra current through it to get the 0 to 3V voltage range:

No longer recommended

The maximum resistance now seen by the cathode is 1K. However, the current consumption is now increased by the 47K/1K voltage divider, which consumes 3.1mA. So, the total receiver consumption is now about 6mA. This is not a problem for receivers operating off the mains or a car battery, but it is a waste of current as far as dry battery operation for the B+ is concerned.  The control works a lot better however in that the time lag is less noticeable.
Given a choice, the original method of control with the -30V C- supply does give the smoothest control. In any case, the receiver performance is not ruined by any of the three circuits so far described. Postscript May 2008:  After developing the later tuners it has become obvious that this circuit is unnecessary. While it works perfectly well it adds to the current drain and uses one unnecessary component (the 47K).

Dry Battery Operation
  With cheap no name 9V batteries now available (or badly misspelt copies of known brands so they don't get sued) from $2 shops, this is a possible option.

  The heater (or A supply) can come from 4 D cells or a 6V lantern battery. This will give about the same life as it would powering a torch. More batteries in parallel will increase the life before replacement necessary. The standard carbon zinc 6V battery (e.g. type 509) has about 5Ah, so expect
about 10 hours in the real world. Of course, if Ni Cad cells are used, you'll need five in series to get 6V.
  More efficiency would be gained by wiring the 12AT7 heater to 12.6V and using 8 D cells (or two lantern batteries)  in series. This means only 150mA heater current so the internal resistance of the battery has less effect.
  Looking at the B+ supply for this circuit, there are 15 x 9V batteries in series, giving 135V. By connecting the negative end of this battery bank to the A+, we get another 6V for free, so the total B+ is 141V. You may need to add or remove a 9V battery to get the correct B+ for your particular receiver. If the regeneration control cannot take the super regenerator to cut off, reduce the B+ until it does. You shouldn't have to go much below 140V.
  The regeneration control uses a 1K pot as per the third method, but instead of providing it with 3V from the B+ using a 47K resistor (and thus wasting 3mA), we get the 3V from the 6V supply via a 1K resistor instead. The 3mA drain from the A+ battery is insignificant.
  Note that the power on/off switch is only in series with the heater battery. A second switch is not needed for the B+ battery as no plate current will flow with the heater not energised. However, if you don't like the effect of the radio slowly dying as you turn it off, then use a double pole switch and cut off the B+ as well. Users of switch pots for volume controls don't need a double pole switch as volume will automatically be at a minimum when the set is turned off.

Performance
  I haven't measured the sensitivity but I can say it is more than the Fremodyne. I'm using my set with two telescopic aerials of the TV kind. From my home in the Blue Mts, I can easily get all the Sydney, Wollongong, and Gosford mainstream stations with minimal noise. Quite a few community stations are receivable but with noise. I can get 2ST from the Southern Highlands with quite a bit of noise as well as some Newcastle stations. Taking it in the car from the Blue Mts. to Bendigo, I was able to receive stations all the way. Shepparton's 3SR was receivable well over the border into NSW.
  How the receiver is tuned and the regeneration control is operated has a huge bearing on the maximum sensitivity. Remember that slope detection is being used for FM, so the receiver can never be tuned to the centre of the carrier where max. sensitivity and least noise occurs (one disadvantage of the super regen approach on FM). This is why the receiver will work much better on AM.
  However, to get every last microvolt out of this set, tune to the carrier centre as close as you can without intolerable distortion and adjust the regen control to the point where oscillation almost cuts off.
At this point the two controls will interact slightly so repeat the operation.
  Sound quality can be quite good (despite what the textbooks say, a correctly designed slope detector can give Hi Fi quality), especially on stronger stations with no noise. As usual  there's the dreaded SCA & stereo subcarrier beat evident on some stations. You won't see that mentioned with most articles describing super regen sets for broadcast FM. Again, see my Fremodyne article about this. With this receiver, the regeneration control can be used to great advantage as it also controls quench frequency. This means it's often possible to find a point where the beat is least annoying or gone altogether. I find it particularly effective with 2WS.

So you want to modify the circuit??

  First it needs to be made clear that super regenerative receivers, and VHF circuitry in general is very critical. There has been a considerable amount of work to get this design to work right. If you insist on changing the critical parts of the design, I suggest you build the original circuit first so you know how it performs, and can judge the performance of an alterations from that. I do not recommend other valves. Although there's plenty of other VHF valves such as 6ES8, 6DJ8 and 6AQ8, the 12AT7 is what this circuit was designed for.
  As to construction, forget how you build MW sets with long wires all over the place and built on a piece of wood. A proper groundplane is essential and connecting wires must be short. We are dealing with frequencies 100 times greater than the MW band and things become very critical. If you can't make an aluminium chassis, then use one of the commercially made aluminium boxes to build the receiver in. If you don't mount the tuning condenser rigidly, you will find the receiver a real pain to tune. You might care to do some calculations, working out the inductance of a few cm of wire and then see what the reactance is at 100Mc/s...you'll soon give up poor construction techniques.
   Unfortunately I can't redesign the circuit to suit certain components you may wish to use that differ from the original but I can answer any questions on the design and how the receiver could be modified to suit your requirements, but you'll have to build it and optimise it yourself...I can only answer at a theoretical level.
  As for the commonly asked question as to how to change the frequency range of this receiver,  it's easy; as with any LC tuned circuit, reducing inductance in a tuned circuit raises the frequency, and vice versa.
Practically speaking, if you want to make the receiver tune higher than 108Mc/s, take turns off the coil.
To make it tune lower, add turns. With my original 12AT7 receiver I designed back in 1987, I did get up to about 210Mc/s. However, having said that I can't guarantee the performance; you might have to optimise some of the component values by trial and error.

 Mains Powered 12AT7 Receiver
   This is a 240V operated version of the above receiver, but with such refinements as a Sallen Key low pass filter, and of course a power amplifier. As the pics show, the chassis is getting a bit grotty; this receiver has spent a few years being used in a garden shed. The vacant hole in the chassis was for an 0B2 regulator used in a previous version of this receiver which used a 6ES8 valve. In case you're wondering, the 6ES8 version was nowhere near as good.
   The Sallen Key filter has a cut off frequency of about 8000c/s and is a very effective method of removing the quench frequency. The output stage can thus be fully driven without being overloaded.
A 6DX8 (ECL84) is used for the audio amplifier. This is a triode pentode with the pentode being of the frame grid type intended for video output use. Europeans would know this valve better in it's series heater form; the PCL84 or 15DQ8.
  The power supply is conventional with a 5Y3 rectifier and back bias is used to supply the -30V for the regeneration control.
  As a study into using this tuner in a car radio for my Model T Ford, I have established that although the B+ and C- (the negative supply) need to be selected for each receiver, it is unnecessary to use a regulated supply. My mains powered version (pics below) of this receiver operates consistently from 200 to 240V.


The reason is that as back bias is used to get the C-, it tracks with the B+. I would expect that in the 5~7V environment of the Model T the receiver should work without having to regulate its supply.
  For the circuit diagram of this version click here.

Further Experiments July 2004

6ES8 I thought I'd just try this twin triode frame grid valve given it's higher gain over the 12AT7. It's pin compatible except for the heater connections; the 6ES8 only being suitable for 6.3V. Performance was poor compared to the 12AT7, with low sensitivity, and the ratio of quench waveform to audio signal was very poor. Having said that, I did not modify the 12AT7 circuit except for changing the heater connections. However, this is not the first time I've used a 6ES8 in a super regenerative circuit. Previous attempts were not very good when using it in a self quenched circuit. I have had more success with the 6ES8 in a separately quenched receiver. The 6ES8 also performs well in a standard regenerative VHF receiver.

Voltage I've found that the specified 140-150V may not actually be necessary. With the 12AT7, performance appeared to be satisfactory down to about 80V. With the 6ES8, down to about 40 volts still allowed the receiver to operate. So, this could be good news for those who want to make the battery powered version.
 
 

                                          back to homepage
 

Email me: cablehack at yahoo dot com