EMRFD Message Archive 4782

Message Date From Subject
4782 2010-06-23 08:51:11 Glen Leinweber Regen musings
Was thinking that regenerative receivers often substitute
FETs for valves in "tried-and-true" circuits, where one
device does all the work - amplification, detection,
oscillation. An interesting variation in EMRFD combines
a Colpitts transistor that provides negative resistance,
with a separate FET buffer/detector.
Seems to me that a single-stage regen requires conflicting
attributes that hobbles ultimate performance:
1) high linearity - for smooth entry into oscillation without
hysteresis, and to avoid cross modulation by strong adjacent
2) non-linearity - to convert the RF signal to audio the amplifier/
oscillator must be non-linear to act as a mixer.

The negative-resistance generator (oscillator) requires very
little gain, if the RF resonator components are low-loss - it
doesn't take much positive feedback to boost Q from a
hundred to thousands. It seems that some circuits keep gain
low by operating at low current and/or low voltage. This
can't be good for linearity.
The multi-stage approach makes sense to me - make the
negative resistance generator as linear as possible, and add
a linear buffer that feeds RF to a high-level detector. While
more complex than a single stage regen, performance
ought to be enhanced, with smooth regen control, and
better crowded-band signal handling.
Or is this a wrong-headed approach? Is the regen concept
attractive because it is a simple low-parts circuit with
adequate but not outstanding performance?
4783 2010-06-23 21:51:54 KK7B Re: Regen musings
Hi Glen,

I would suggest that the only "wrong-headed approach" to a regen is attempting to understand the subtleties without having one working
4784 2010-06-24 05:52:36 davidpnewkirk Re: Regen musings
4786 2010-06-26 04:15:19 Tim Re: Regen musings
I have both built regens and tried to model them in LTSpice.

In real life, any regen (tube or FET or bipolar) is vastly more linear than you would think from my Spice simulations.

In fact, what surprised me most was this: the tendency to use active devices in regions of low current/low voltage for regeneration, where the device is nonlinear in the "smooshy" sense, is actually good for the regeneration linearity.

Devices that limit "hard" do not do well in a regeneration section. When we think of wide dynamic range stages we often think about devices that we never ever let get to the stage of clipping or flat-topping, in fact we have an urge to never want to run them where they'll be exhibiting gain compression so we make them linear right up to the point of clipping or flat-topping. As a result we often think that wide dynamic range stages have to limit hard. I'm not so sure this is true or even desirable. But looking at SPICE simulations or scope traces we might believe this.

But devices that limit hard are poor choices for oscillators. Regens, which are oscillators where we want a smooth transition right at the cusp of loop-gain-equals-one, are a place where you actually want all the smooshiness that you can get.

I think that this is representitive of a common and good approach to oscillator design, where non-linearity in the active device is actually important. And devices that exhibit "hard limit" in their non-linearity are poor choices. You want a device that goes into gain compression, rather than hard limiting. And tubes (esp. triodes but also pentodes operated with plate loads that are low enough impedance to give the smooshiness) and FET's are very representative of devices that go into gain compression smoothly. Bipolar transistors can also achieve smooth gain compression instead of hard limiting, but many of our instincts in designing bipolar amps drive us into making stages that clip and flat-top instead of smoothly smooshing.

There was a QEX article a month or two ago about the importance of gain compression in oscillators.

Now, as for the detection action, that's something that is related but not exactly the same. I find that devices that smoothly go into gain compression have a square-law type characteristic which is actually quite a good detector. Devices that limit hard, do not have the smoothness of a square-law type detector. They don't detect anything, and next thing you know there's a 90-degree knee and anything above that threshold is detected with no smoothness in take-off.

An article I turned up in the 1960 HP Journal talks about the occasional production line HP Wien bridge oscillators that would have a very low level of distortion - but which had serious amplitude instabilities. The author traces through a very nice analysis that shows that a minimal amount gain compression/nonlinearity/distortion is necessary for even the most cleverly simple AGC-stabilized oscillators.
4787 2010-06-26 19:11:28 w5ida Re: Regen musings
Tim is essentially correct about phase noise being something that was not a common term of use in 1960. For example, see Proceedings of the 1996 IEEE International Frequency Control Symposium, pp. 47-57, available
4788 2010-06-30 06:31:31 davidpnewkirk A not-so-simple regenerative receiver challenge (Was: Re: Regen musi
4789 2010-06-30 08:33:12 Paul Daulton Re: A not-so-simple regenerative receiver challenge (Was: Re: Regen
Spotting wasnt an issue in the days of the regen. Licensed in 1959 I operated in the xtal controled days. We often had qso's 40 khz apart. One would call CQ and tune the entire novice band. Old timers advised us to use a short wire for rec antenna to keep noise down.

Regens are far more useful today than when I started, the hetrodynes from foriegn BC and amateur AM stations made simple receivers even the low cost superhets useless. Now our bands are much cleaner and a regen shines. One of my favorite projects in the last ten years was a 6sn7 regen made to look like a National SW-3. A friend looked at it and said "thats old technology". To which I repllied " Technology hasnt changed its just old hardware".

fun topic to follow.

73 Paul Dault
4790 2010-06-30 10:37:31 ajparent1 A not-so-simple regenerative receiver challenge (Was: Re: Regen musi

Unlike many who started as hams I started as a radio builder.
I say that as I was not really an SWL, I didn't listen for
BBC expicitaly but used SW broadcast and AM broadcast and
ham bands to test ideas. Initial effort was made to have
a radio that was not store bought and could do better than
the early 60s 6 (or more) transistor radio. The challenge
was I lived about a mile from WGSM 740khz 3KW daytime and
wanted to hear WABC 770 about 50 miles away. That forced
evolution from simple crystal sets to regens then to
converter/regen to full superhets with Q-mult(form of regen)
to get the needed selectivity. along the way in the latter
60s I found CB was a place to play by way of 3transistor HT
(anyone remember Lafeyette HA70?) and endevored to build
radios with better range and even better receivers that
only heard the channel I had a crystal for. During that
time I had SX40s and RBO-2 and other radios but limited test
gear so the radios were the "can we be better than.." test.
Back then I was an early adopter of transistors making it
harder due to lack of high frequency capability.

A result of that has been that Q-5er or Regenodyne (to use
another authors name) or down converting mixer/osc with
regenerative IF/detector at fixed frequency has been a
recurring scheme for me. It gets me sensitivity, selectivity
and stability with few parts. Its a foundation superhet.

A few years back I did a ground up tube version as more than
close to minimal 5 tube RX for 80/70m. Classic lineup was
6U8 LO and mixer, 6BA6 IF, 6AU6 regenerative detector, 6CW4
BFO and 6GW8 audio amp/power amp. using stagger tuned HB
IF cans at 455khz with triple tuned link couped sections I
got good skirts at high loss and the 6BA6 was to make up
for that and provide IF gain (manual) control point. The
regen detector was to add additional variable selectivity as AM
detector and with BFO product detector for CW/SSB. Due
to simplicity it all fit into a 8x7x1.75 inch chassis with
power supply (transformer). It's stable as the LO is very
loosely coupled to the mixer for minimum pulling and selective
without being too narrow for AM but a bit wide for SSB/CW.
Being there is no RF amp it's hard to overload. Makes a
fine listen to the band radio and a good sidekick to my
HT37 TX. It is more than adequately sensitive and the usual
antenna is a 14ft wire tacked to the ceiling with a tunable
coupler. Spotting is easy, a toggle switch disconnects
the antenna.

During construction I played with the regenerative detector
and its gain plus the selectivity clearly adds greatly as it
would need more gain for a diode detector and an AGC as well
to keep the IF overload in check. The regen stage provides
more output, selectivity help and tends to limit volume

It could be done with transistors (or fets) just as easily
for those shy of higher voltages and heat from tubes. I've
done it recently as a Fet VFO driving a DBM into a fet at
262khz as regen followed by two stages of audio amp for
headphones. Using 4 transistors the result was very satisfying
as an 80/75M reciever that was operationally simple and could
hear AM and SSB/CW. I later added a transistor for BFO
rather than an oscillating detector as it was more stable
and could exploit the regen selectvity and gain. Since the
regen was not needed to be oscillating or adjustable the
variable resistor was replaced with BFO switch making
the front pannel controls tune, audio gain, bfo on/off and
bfo tune. The trigger for this was Rick's minimum gain
simple superhet and used the same mixer, vfo and audio
but the IF amp and crystal filter was reduced to a regen
very loosely coupled to the DBM output.

I like minimal radios when they are not minimal to the extreme.
When you reach some lower level the simplcity adds operational
overhead and can be both interesting or annoying. Adding
to a simple design to move toward something more sophisticated
often helps correct the operational limits and can be fun
exploration. The best example is MW and VLF regens can
exhibit all the desired traits and advantages and a frequency
converter can isolate it from the antenna and move it to the
desired band.

One can never build to many receivers.