Regen III: A Journey of BJT Regen

Make up a Regen

2020/11

Based on JFET version, I reckon this  regen gonna works well, plus, the schematic is cute. 

This a BJT replica of JFET version regen, which is kind of famou based on N1TEV introduction Regenerative.

I built a CNC milled PCB, and test it.

That's messed up, surprisingly.

Tuned capacitor could not control the frequency, Regen Adjust could not provide damping,  nothing seems working.

This is not because the gain is too high. Even I tear down tickle winding( series with 2N2222's collector), oscillation sill present.  This is a clue, there is another feedback way which is not wanted.  So this is the key problem. I try to heal that by remove the emitter bypass capacitor.  yeah, it does works, more or less.

Problems still there, just be partly eliminated.  I use a bad  1mH RFC, which provided another possible oscillation mode.  I summarized this 3 oscillator together as following. 

But why these guy's works

So, why JFET version works well? 

There are good reasons.  JFET is much better on the junction capacitance, J301 only 2.5 pF and MPF102 is a little bit worse, but much better than 2N2222 or 2N3904.  Even worse, I deploy a 3DG130C, probably have 100 pF or so capacitance, godness. 

Then, how about regen on QST(1957/10 page 30)? 

Well.  The capacitance Cbe is directory to ground, which eliminated  2 other feedback altogether .  It's a clever design.

And, I built this once.

BJT is configuration as ground base. so Cbe, Ccb is grounded.  Ccb became part of tuned tank.  Cbe grounded, bypass some feedback signal, not a problem.

What can I do about this Regen Monster

Base on knowledge and practice, I recon following configuration could work.  0.47uf  emitter bypass capacitor effectively kill the parasitic feedback. Left only one RED feedback path.

Guess what, Almost nothing changed.  The Regen Adjust does could control the oscillation, but if Regen Adjust capacitor big enough(>200pf), there would emerged group of oscillation, from 1 Mhz extended to 50Mhz.

I even short the emitter to ground, It won't help either. This indicate there is another feed-back path hide there ! After days trying and crying, I confirmed this.  but, Where?  Which?  How?

I had to work out that by ltspice.  It's powerful: all symptom is identical to actually circuits,  that's means it's not because the messed fly wire nor bad RFC. There does existing another feedback path. 

Another experiment day, there is nothing left to try. I wonder how the regen adjust control the feedback, and is the gain achieved via the impedance of RFC?  why not try to connect the regen adjust capacitor directly to collector?

Both experiment and Ltspice simulation confirm this did successfully eliminated the Regen Adjust problem, now it's won't pop out frequency everywhere in whole HF band anymore. How surprisingly!

Simulation confirm, while the regen adjust capacitance increase,  the oscillation stop, but continue increase this capacitor, a much more high frequency  oscillation excited.  And, this time, the tickler winding and the tapped winding(to base of BJT) form a new oscillation tank.(confirmed by damping those tank  by resistor)

Bad Regen

Even though the oscillation control is work now, the gain is so poor while the oscillation is stop, no station at all.  High gain region is too narrow to tune on a station.

Insert R2 series with regen adjust capacitor C1 would smooth the control.   (R4 is nothing but simulation a imperfect RFC. )  RFC here is not make the gain increase, but decrease,  what is a surprise. 

Tried everything I could come up with: change quiescent current/change feedback tap position/ change tickle windings.  Nothing seems work..  no station at all,  but obviously the oscillation get controlled. 

Cross compare all kind regen receiver, they all use a very very low working point: normally less than 100uA.  My version is obviously too high, before i realized this, the quiescent current  was almost 1-5 mA.   By inserting a 2k resistor to emmit, and use 39k collector loading resistor, a uA level quiescent current achieved. 

low quiescent current ensure a better regen control

Now, there is few station.  It get better a little bit,  but again, It's so hard to tune, only one or 2 stations received, totally ruined reputation of  Regen receiver.

There must be something  I missed.

Deeper Dive Regen Detector

What save me is a point contact diode.  I don't know this diode is needed at all. Because almost every regen schematic i founded,  this diode is not needed.

This diode save everything.  Lots of stations pop out,  regen control is decent and I feel it's right.  So the circuits now  became this one:

Decent Regen Receiver

Then, everything explained.  Regen Receiver don't need a diode detector: this is not ture.  

Regen does need a detector. The trick is BJT/JFET bias.   Every regen don't use diode detector, all configured to almost cutoff states.  It's not correct to turn off the BJT   by increasing the collector/emitter resistor! The corrector way is increase base resistor, in my scenario, it may need 1 Mohm, even larger. 

Review Circuits Debug Process

There is no loss but gain. Painful debugging process teach me lots of thing instead frustrated me.

parasitic feedback

High frequency parasitic feedback 

Diode detector or make sure the BJT cutoff

Modified Original design

Since  the problem is solved, I consider recovering most of original design. And I come up with this regen receiver finally. 

Design notes:

Alignment notes: ( following this procedure to make it work )

Some Thoughts

What could I expect from a regen receiver?

Coped with a LM386, get gain 30dB, this is possible, but only for strong signal. Most of regen receiver without a high enough gain, say It is sensitive only when listen it with a earphone.

To listen with loudspeaker, a additional gain stage is needed,  a extra audio stage, or a RF stage before detector.

Not really.  For example, I had a 15 meter long wire antenna for HF. This antenna injected too much strong HF stations to the receiver, even the regen theoretically improve the Q,  I,e the selectivity. the whole receiver still been severely overloaded.

It depends.  For simple enough TRF, regen is much better. I could not make a simple TRF beats the regen so far. 

I tuning the regen receiver follow a procedure.  I always tune up, than tuned down:

1)  Use regen control make the receiver just begin oscillating,  you gonna hear a soft pop noise. then tweak it a little back. 

2) Now tune up a little bit, if there is a station, we gonna hear it. 

3) tweak the regen control back more, make the oscillating stop, then tweak the main tune, lock the station.

repeated 1) - 3)

Driven loudspeaker

One Regen Stage following by a 30 dB audio stage does not enough for driven loudspeaker to a comfortable volume. I then attached another RF stage before detector.

Well, It's works.  It could driven the  Loudspeaker now. for long story short, here is a summary feeling about this regen receiver configuration:

Soften Regen Control

Cause the prototype  works drive the loudspeaker. I reckon this one could do better. More stable bais, might good regen control, good impedance match.  

I flatcam it and make a CNC PCB out of it.

What actually happen is disappointed. With a loudspeaker, It's much less sensitive than the earphone. This demand the regen control to be 'soft' --  needs a wider high gain region before bursting into oscillation. In this 'soft high gain' region,  we could located the station, earphone is far more sensitive than loudspeaker, means this soft gain region is very large).  Only if we could hear something, we could know how tune the regen.  

With the loudspeaker, the "soft high gain region" is too narrow,  which gonna missing every stations, except we try hard to wiggle the regen adjust knob, that's experience is too bad.

The loudspeaker prototype  is  coupled to regen stage via a large capacitor, around 4700pf or so. This arrangement might not a bad thing.  Second stages gonna loading the regen receiver, if it's  burst into oscillation, the loading effect might became significant tending to cease the oscillation, the cause list here might be wrong,  except the fact told me so. 

And plus another loading source:  the tap which connected to base,  also contribute to make a soft regen control. Tapped from bottom, ie, 1/5 or so, I get a "harder" control, but connected into 1/3 -- 1/2 tap, the regen is much soft.  With this change, the tickler winding need to increased, and the top band then get harder to start oscillation.

I decide it's the loading might make  the regen control soft: more energy loss in the oscillator, more energy needed  pump into the tickler, loss is everywhere but ergney pump  control  by only regen adjust.

Perfermacy summary for capacitive couple to regen:

Following  Rule learned

Ok, put aside whether the theory is right or not,  that provide me a idea,  I could make the second stage input impedance lesser and get high gain simultaneously: just bypass the emitter degeneration resistor. 130 ohm resistor with a 330k bias,  there is about 1mA quiescent. (last setup get 2 mA or so),  if beta is about 100, input impedance about  2k(low frequency), this is half than before,  and get the extra gain.

 note: the 3dg6 is a low hfe transistor,  for 2N2222, change 330k to  560k or even larger to get 1mA quiescent current.

This configuration really surprise me.

This receiver now is totally OK for sensitivity, regen smooth,  easily tuning,  and these performance stay almost identical while replace  transistor. For second stage, it needs to lower quiescent current for high beta transistor.

The key to make easy tuning and smooth regen control: