Automatic Regeneration Control for Regenerative Receivers.

 This regenerative receiver adjusts regeneration automatically and thus can be used by anyone; the user only needing to adjust tuning and volume.

Regenerative receivers have the advantage of simplicity with good performance, and for the last 25 or so years have been a large part of my means for listening to medium wave transmissions. And, they are probably one of my most built projects. However, while this is all very well with someone who understands the principles of regeneration, and more importantly, how to control it, the problem is with non technical operators.
Briefly,  the concept of a regenerative receiver is to cause a controlled amount of positive feedback in a tuned circuit which increases gain and Q considerably. The tuned circuit may be part of an RF amplifier or the detector. Regeneration makes it possible with just one tuned circuit to obtain sensitivity and selectivity similar to a superhet. Having said that, there are plenty of poor designs around that give regenerative receivers a bad name.
The key to obtaining optimum performance is in the amount of feedback. Too little, and sensitivity is poor; too much and oscillation occurs which obliterates the signal. The level of feedback must be adjusted to the point just before oscillation occurs. So, regenerative receivers have a control for this adjustment.
Because the performance of the tuned circuit varies across the band, it means that the regeneration has to be adjusted each time another station is tuned in.
These variations are caused by such things as aerial loading and the LC ratio of the tuned circuit. Furthermore, supply voltage variations will affect the level of feedback.
Experience has shown it is impossible for non technical users to optimally adjust regeneration controls, with the results invariably being frustrating squeals emanating from the receiver.
This is one reason regenerative receivers were quickly replaced with TRF's and superhets in the late 1920's  and early 1930's. After this, they were relegated to the realms of the technically minded constructor, with many magazine projects right through to the end of the valve era, by which time building radios at home was a declining hobby.

Regenerative receiver improvements.
First, it needs to be pointed out that the basic regenerative receiver can be improved to lessen the amount of regeneration adjustment required.
To eliminate aerial loading effects, a simple RF amplifier can be placed between the aerial and tuned circuit where feedback takes place. The B+ supply can be regulated to eliminate changes in battery or mains voltage affecting regeneration. However, regeneration still has to be adjusted as the receiver is tuned from one end of the band to the other because of the LC ratio. My thoughts were, if a human can know where to set the control, then why not some kind of electronic circuit doing the adjustment instead?

Fortuitous construction mistake.
The first regenerative receiver I built was during 1984, based on an Electronics Australia design (Three Band Three). It used a 6BL8 pentode as a regenerative detector, the triode being a following audio amplifier to drive a speaker or headphones. Beyond this point, I deviated from their circuit and used a 6BM8 for audio output instead of a 6AQ5. I didn't realise it until many years after, but an accidental misreading of a capacitor value during construction made the circuit perform slightly different to that, had I used the correct value.
And, it was because of this I could see how regeneration could be made automatic.
Instead of using 100pF for the grid capacitor, I had used 1000pF. In as far as general performance went, it didn't make a lot of difference. Regeneration was as smooth as it should be, the sound was good, and sensitivity down to the noise level. What happened when the regeneration control was advanced further than it should be, was another thing. In normal regenerative receivers, RF oscillation occurs and a beat is heard against station carriers, as the receiver is tuned across the band. Between stations, not a lot is heard. If the receiver is oscillating and tuned right on the carrier, a form of direct conversion occurs, and it isn't always evident the detector is oscillating.
Because of the increase in grid capacitor value in my set, and that a high gain TV valve was being used as the detector, what happened as well was a fierce audio oscillation. This was simply because of the grid time constant increased to the point where the detector was squegging at an audio rate.
So, to sum up, a loud whistle was heard if the regeneration was advanced beyond the optimum point, whether a station was tuned or not.

The concept of automatic regeneration control.
The whistle was far louder than the ordinary program level, so it seemed to me that if one was to rectify it, then a DC voltage would be produced in proportion. As the whistle got louder, then the DC would increase, and vice versa. The increasing level of DC could then be used to back off the regeneration.
In early 1985 I started experimenting. To get DC proportional to the whistle level just entailed a germanium diode and filter capacitor, taking signal from the audio amp before the volume control. How to control the regeneration with DC? Well, one could do as in any voltage controlled amplifier and feed the negative going DC into the grid of the detector, via a high value resistor so as not to shunt the RF to earth. A control of sorts was indeed possible, but it wasn't as good as I'd have liked. Looking back I can easily see why - the grid to cathode circuit is used as a diode in a grid leak detector. By negatively biassing the grid, while stage gain is reduced, the characteristics of the 'diode' are altered. In other words, distortion occurs because only the peaks of the signal are detected.
Regeneration is normally controlled by adjusting the screen grid of the 6BL8 pentode. Taking it towards earth reduces regeneration. We could take it to earth via a transistor of which the base is fed with positive current dependent on the audio level. And that's what I did. Results were very good and I used the circuit for quite some time.
One reason for developing the ARC was because I was developing a car radio using the 6BL8 regenerative circuit, and I wanted to eliminate the necessity for regeneration adjustment by the non technical users. Certainly, the prototype receiver worked brilliantly for mobile use. In fact, the ARC operating also as a form of audio derived AGC meant that volume was fairly constant as the car was driven through areas of fluctuating signal strength.

Taken from my original regenerative receiver, this circuit shows the concept fairly clearly. Excess audio levels (such as when the receiver has gone into oscillation) cause the transistor to conduct and reduce the screen voltage. The valve type is 6BL8/ECF80. Note that although it worked, operating conditions are not optimal. It can be seen that if the ARC pot is wound all the way up that the postive going audio will be clipped. Also, the time constant is rather crude relying on transistor bias current draw to discharge the 470uF capacitor. For mobile use this receiver was fitted with an RF amplifier and a vibrator power supply.

Fast Forward to 2012.
While the original ARC circuit worked, there were three things I didn't like. 1) It used a transistor, 2) It could cause audio distortion because of asymmetrical loading of the AF amplifier, and, 3) It was only good for reasonably strong signals.
With a lot of analysis and thought, plus some experimentation, I've come up with quite a good performing all valve circuit. A previously built regenerative receiver became the 'victim' for the new circuit, which has worked so well that it is now a permanent part of it.

The latest regenerative receiver with ARC. (The 6BX6 plate load resistor is 100K).

Controlling Regeneration with DC.
Where a triode is used for the detector, the plate supply can be varied, or in the case of a pentode, the screen (g2) voltage. Previously, this was done by a BC548 transistor. To use a valve is rather problematic though. The problem is that the voltage needs to be taken quite low - around 15 to 20V is typical. It is difficult to get the plate voltage of a controlling valve that low using normal valves. One could take the cathode to a negative voltage so the controlling valve still can have a fairly high plate to cathode voltage, but it means a negative supply has to be available.
As stated before, the control grid (g1) cannot be used for control as it functions as the detector diode. What about the suppressor grid (g3)? That would not affect the diode performance at all, and by taking it negative should reduce the gain of the detector valve. Taking it negative would also mean no current would be required, just voltage. This was all theory at this point.
So, out with the seldom used regen receiver I built in the mid 90's which I'd never really liked anyway. Because it had a 300mA series heater circuit, and I needed a high gain pentode with the g3 connection available, in went a 6BX6 for the detector. I found the idea certainly worked very well, although the negative voltage required was higher than I expected (about -20V to take it out of oscillation). The suppressor is the furthest from the cathode, so this is not surprising.
Of course the grid leak capacitor had to be 1000pF to provide the audio oscillation required. And, previous experimenting had shown that a high gain valve was required. 6J7 and the like don't oscillate in a regenerative detector circuit fiercly enough to get the squegging.
At this point, I had smooth control, and nothing of performance had been sacrificed.

Obtaining the control voltage.
Wanting to keep things simple, my idea was to use a 6AV6 and the diodes within to obtain a negative voltage proportional to that fed in. Briefly, the audio from the 6BX6 plate was fed into the the 6AV6 triode in the usual way, but the triode plate then fed the diodes via a capacitor as a shunt rectifier. Doing this provided two benefits; 1) it was obvious that the diodes would have to fed with a voltage higher than what came from the 6BX6 plate, so the triode could provide the extra drive, and, 2) to prevent the diode loading causing distortion at the 6BX6 plate. The idea worked, and for local stations we had automatic control.
A word on the time constant, which is a 10M resistor and 1uF capacitor. Obviously, if it is too short, the audio gain of the receiver will be reduced because negative feedback will occur at an audio rate. Conversely, if it is too long the receiver becomes difficult to tune because it takes too long for the gain to come up when tuning between stations.

Further improvements.
First thing was to use the 6AV6 diodes in conjunction with another diode as a voltage doubler; I used a 1N914 silicon diode to test the theory. As expected, more control voltage was developed. A further test revealed that if the input to the 6AV6 grid was higher, so would be the resultant DC output. Here I used a 12AU7 to do two things; act as a preamp for the 6AV6, and the second triode could take place of the 1N914, thus making the ARC circuit all valve.

The New Receiver.
This receiver has a few aspects of its design which should be covered first. The power supply is a demonstration of how to use modern low voltage transformers to power valve circuits. A 240V to 36V transformer powers the whole receiver. The B+ of around 185V is obtained via two half wave voltage doublers which are then series connected. This is more efficient than using a simple Cockroft-Walton multiplier which would provide inferior regulation, as well as not asymmetrically loading the transformer secondary on each half cycle. This receiver orginally used a 12AX7 for the detector, and a variant of the transistor ARC circuit, but with a 6AL5 instead of the germanium diode. Partly because the 12AX7 triode does not provide the squegging effect, the circuit never worked properly.
To make most efficient use of the 36V AC, the heaters are in a series circuit; they add up to 25.2V, so a dropper resistor is included. Of course, the heaters must all be the same current; in this case 300mA.
Note that the heater chain floats at about 95V above earth as one side of the secondary is where the two doublers connect. This is below the usual 100V heater to cathode voltage allowable for most valves. Note also the order of heaters. The detector valve is first in the chain (in terms of AC earthing) to have the least hum.
The choice of audio valve will seem strange to those not familiar with 1950's English TV sets. It's a PCL83 which has a 12.6V 300mA heater. For the output transformer a 240 to 12.6V 150mA type 2851 was used because it was to hand at the time. There appears to be no ill effect from the 20 odd mA DC flowing through the primary winding.

The detector and ARC circuit.
Signal via the tuned circuit feeds a 6BX6/EF80 pentode wired in Hartley configuration to provide feedback. Manual regeneration is controlled by varying the screen voltage. Note the grid leak components; the values are important for the ARC to work. Following the detector, the audio signal is amplified by a 12AU7 triode. It then splits two ways; to the PCL83 audio amplifier which is conventional - no more need be said about it. The other path is to a preset pot which controls the input to the 6AV6, again wired as a normal triode amplifier. So the cathode can be earthed (in view of the diodes functioning correctly), contact bias is used. The 10M grid resistor allows sufficient electrons to build up on the grid to cause it to go negative by the required amount.
The amplified signal now proceeds to a half wave voltage doubler, using the 6AV6 diode, and the other half of the 12AU7 connected as a diode.
The diode load and filter capacitor are formed by the 10M and 1uF which gives a suitable time constant.
From here, the negative DC proceeds to the suppressor grid of the 6BX6. When manual control is desired, the switch shorts the control voltage to earth, as well as the supressor, so as to return the 6BX6 to normal operating conditions.
As a guide to operating conditions, I tested the receiver at 1503Kc/s with 80% modulation and a level of 67dBu. The following voltages were obtained:

Setting Up.
With the ARC switch in the manual position, the receiver is used in the normal way with tuning and regeneration controls. To select automatic regeneration control, the regeneration control is set to the minimum point where oscillation occurs right across the band. The ARC preset pot is set first to minimum, and the switch opened. With the receiver tuned to a station, the ARC pot is slowly turned until oscillation ceases. The weakest station is then tuned and the control touched up if need be. From now on the control is automatic, and only the tuning and volume controls are used.
At this point, a further improvement would be to have an additional preset regeneration pot switched in with the auto/manual switch so the panel control does not have to be reset each time the ARC is used. If the receiver was to be built for a non technical person then the regeneration pot should be a preset control and not on the panel as a user adjustable control.
One may wonder why not just set the ARC for full on. The problem is that on strong stations the regeneration will be at an excessively low level, providing poorer selectivity and lower audio.

It should be obvious by now that the ARC is operating on the audio signal and not the actual RF signal strength. This is not normally a problem as radio stations do not usually cease modulation for any length of time. This is done for commercial reasons, because if there is no audio, listeners may tune past the station without realising it was there. But, for reception on other bands where there may only be sporadic voice communications it is unsuitable.
Because the circuit looks for the audio level, it does mean that stations which are so weak as to be barely above the noise level will not provide enough control. But then, non technical listeners don't usual go seeking such reception.
In practice it works very well and makes it possible for a non technical user to tune such a receiver as easily as a superhet or TRF. For fading signals such as in a car, or from distant stations, the circuit also provides AGC.

home page