TheRadioBoard

I know that you have three different kinds of regenerative circuits:

* The Armstrong type with a tickler coil at the anode/drain circuitry, or a tickler coil in the cathode/source circuitry;
* The Hartley type with a tapped coil at the cathode/source circuitry;
* The Collpits type with a capacitive voltage divider tap at the tank circuit.

The regeneration can be adjusted by:

* Varying the supply voltage at the anode/drain;
* Adjusting the screen grid voltage or the G2 voltage of a Dual Gate-MOSFET (G2 of to JFETS Cascode circuit);
* Adjusting the feedback current in the feedback circuit of an Armstrong, Hartley, or Collpits circuit by means of a potmeter or a throttle capacitor

In al these circuits a separate Infinitive Impedance Detector is assumed, consisting of a drain bend JFET, which is connected at the top of the tank circuit.

Which of these are the best, regarding the smoothness of regeneration and the frequency coverage? I know that the loop gain of the active element in such a receiver must not be too high, otherwise the adjusting of the regeneration will not be smooth, but will suffer from the nasty backlash.

The regeneration can also be adjusted by changing the coupling of the feedback.

also I have used VERY high gain devises (MPSA18 bipolar to name one) and had no problem as long as the feedback is correctly designed.
to see what I am talking about just think of the the gain of a FET and the gain of a tube, difference is staggering, yet some tube circuits that use tube having low gain is hard to control while a FET circuit having ten to thousand times that can be so smooth that it is almost hard to find the "sweet spot" where regeneration gives highest gain without distortion, this is mostly down to circuit, both layout and component values aswell as design of coil/s...

just one question, why do you need a separate detector? I have used a single J309 and matching transformer to catch 20M SSB and CW not to mention broadcast stations.

Don't forget that there are other topographies, too, such as the Franklin circuit. Plus, there are variations on the three circuits you cited, such as the Vackar configuration.

Dave

Yes, I also know the Franklin circuit. Such as this regenerative receiver with a separate Q-Multiplier and Audion detector at

.

The Franklin oscillator in this schematic is in fact a differential amplifier in which you adjust the regeneration level by adjusting the current through it.

@OErjan:


It is usually better to use as separate detector and a separate Q-Multiplier. In this case the Q-multiplier can be optimized for optimal regeneration control without backlash while the detector can be optimized for detecting the audiosignals. At http://www.epemag3.com/index.php?option=com_docman&task=doc_download&gid=192&Itemid=38 you can download the schematic and building instructions of such a receiver with the detector and Q-Multiplier separated. In this case the Q-multiplier is a Hartley oscillator and a JFET as a drain bend detector.

DrM,

In the hartley, and Colpitts and thier variations, You can adjust the feedback(statically) by moving the "tap".

Move the tap "down"(less feedback) for "stability, and move it "up"(more feedback) for more Gain(output).

As far as I know this principal holds true for ALL the various oscillator circuits. I.E. Less Positive feedback(Regeneration), for stability.

Basically that is what a "regen detector" is: An oscillator who's loop gain is <1 .

I call this a "static" adjustment because once set/built it cant be changed.

Then you adjust the "dynamic" loop gain with some kind of regen control.

Throttle Cap, Screen grid voltage Pot, Tickler Variometer, etc.

Moving the "tap" down can help with that Hysteresis effect.

Also the throttle cap method seems to be less susceptible to it.

As to what method of dynamic regen control is BEST, I dont have an opinion on that!

73
kb0lxy

Everybody who has built a number of regenerative sets will have their own favorite ways of controlling regeneration. What my favorite method is, may be different from your favorite method. Nothing wrong with that at all. About the only method I do NOT like is the pot across the tickler winding method. The pot will be hot with RF, as it is directly in the oscillatory circuit and it will suffer hand capacity affects. Also it is placing a resistance across the windings, so you can imagine what that does to the Q of the circuit. However the Q is not that important in a regenerative set due to the losses being made up with the regeneration. But why waste the high Q of the coils to begin with?
Curt

_________________
Curt, N7AH
CW forever
Connoisseur of the cold 807

How much does resistance across the tickler winding affect the Q of the tank circuit? Guess it depends on how tightly coupled they are, Given the turns ratio, though, I wouldn't think the effect would be that great. Not something I'd do, but with a triode-tickler your options are more limited and a pot is cheaper and easier to find than a variable cap. Except in this house, where variable caps outnumber pots. Guess I need to adjust my hamfest buying tendencies!

73,

_________________

http://kr1s.kearman.com/
http://qrp.kearman.com/

No influence whatsoever because the feedback in regenerative receivers compensates for all losses. At the point of oscillation all losses are compensated, by the gain of the tube, and Q is infinite no matter what loads are on the tickler or tank.

There are only three basic ways to control regeneration:
1) the amount of feedback (Tickler, capacitive, inductive)
2) the amount of tube gain (anode voltage, G2 voltage)
3) the amount of losses (potmeter on ticker or tank)

Corné

@Corne:


The Q of a coil is indeed important in regenerative receivers. If the Q is low, you will suffer from a poor skirt selectivity, while the nose selectivity is very well due to the regeneration. While receiving a rather weak station, the stronger stations on the flanks of the tank circuit's response will disturb the reception of the desired station at the sharp nose of the tank circuit's response.

Hello Drm,


AFAIK poor skirt selectivity is the result of an antenna, which is coupled too strongly to the LC circuit.

Yesterday I did some experiments to see if selectivity changes with Q of the LC.
I used two of my tube regenerative radios, the one in my avatar and the EL3 regen.

In both radios I tuned to 1458kHz (Sunrise radio) without an antenna connected. In both cases I was able to receiver Sunrise without antenna.
I used a RF generator (RFG) with 400Hz modulation as the strong local station. To check the selectivity I tuned the RFG slowly to 1458kHz, approaching from both a lower and higher frequency, and noted when the 400Hz became audible. I did this measurement with both radios. Then I put a 56k resistor across the LC, which effectively destroys the Q and did the measurements again.
Before I did the measurements I first tuned the regeneration setting just below the point of oscillation. With the 56k resistor paralleling the LC I had to open up the throttle in both radios sets close to maximum setting. I compared the results of the selectivity measurements and all measurements had identical selectivity.

So low Q doesn't affect skirt selectivity.


In the circuit below I've drawn a LC circuit. All losses in a LC circuit can be represented by a resistor (Rloss) paralleling the LC circuit.

In a regenerative radios the feedback compensates for losses in a LC circuit. In fact feedback behaves like a negative resistance, -R in the above circuit. The Q-multiplier and lambda-diode are negative resistance devices that can also be used. At the point of oscillation the negative resistance -R equals Rloss so only LC remains with infinite Q.


Corné

fascinating Corné. i remember you spoke about this before. what happens with progressively lower resistance across LC? is there a minimal Q for regen and does it relate to tube characteristics, say plate current?

regards,
phil

I don't know about all those types but my experience has been that putting the feedback coil on the drain / collector (the Armstrong type regen) does not yield as smooth regen across the spectrum as putting the feedback coil on the source / emitter.

I suspect that the variation in a/c impedance (thus the a/c gain) across the tunning band is too much for the drain / collector (high output impedance), but not much for the lower impedance source / emitter output.

Hello Phil,

Interesting questions.
I didn’t test lower than 56K but in theory nothing would change as long -R matches Rloss. With progressively lower Rloss more and more power is dissipated in Rloss this means -R must supply more and more current to compensate for Rloss. Assuming a constant AC voltage on the LC circuit the increasing current means the -R (tube, transistor, etc) must supply more power. This in turn means the amplifying device, used to make -R, must have larger power gain.
Effects on the three regeneration controls:
1) more positive feedback
2) increased transconductance
3) less losses
In case 3) the amplifying device must be able to supply more power than required. The excess power is dissipated in the potmeter paralleling the tank or tickler.

The theoretical limit is a Q of zero. At Q=0, Rloss would also be zero (supper conductor) and no power can be dissipated in Rloss (ideal short circuit). The practical minimum Q is determined by the amount of power that can be dissipated in Rloss. Any power dissipated in Rloss means that L and C heat up and there's a point at which L, C or the connections between them will melt.

Corne

Corné, very interesting! thank you for your explanation.

regards,
phil

I think that some regen topologies are more suited for regens intended for a small band coverage and other topologies are suited for a wider coverage.

Collpits and to a lesser extend hartley/armstrong are more suited for a smaller coverage, while Buttler and Franklin topologies are more suited for a wider coverage.

Two examples of a regen with a buttler oscillator: