TheRadioBoard

[Dan McGillis]

Hi all: I’ve started a new thread to describe another simple JFET regen. The work leading up to this is posted at:

http://theradioboard.com/rb/viewtopic.php?t=1431&start=0
---------------------------------------------------------------------------------------

This layout, JFET Regen-4B, follows the old “Weagant Circuit”, see:

http://theradioboard.com/rb/viewtopic.php?t=1520

It features a Bogen T725 auto transformer in the drain circuit to support a variety of headphones. And it has MUCH more audio output than the previous JFET Regen-2B.

I liked the idea of following a historical circuit. Having a name and a face associated with a circuit is kind of nice. Researching the inventor - Roy Weagant - and his accomplishments has been quite interesting.



The auto transformer coupling scheme is not very pretty but seems to work OK. The effect of changing the Bogen input tap, and the output phone tap are shown below.



There was a big audio output improvement from the previous 2B layout. With Rs = 100k, the rms audio voltage (from a weak test signal) across the piezo headset improved from about 1.5 mV, to about 10 mV.

This circuit’s audio output also behaves in a more conventional manner than the 2B. As you increase the device gain (by decreasing the source resistance Rs), the volume increases - dramatically. A typical response is shown below.



The audio output versus Rs, and the (calculated) JFET transconductance versus Rs track nicely. The audio-output of about 35 mV rms at Rs = 10k is more than a 20x increase from the maximum output levels seen with the 2B regen.

Another change was to increase the throttle capacitor from 365 pf to 540 pf. It’s now a dual gang 270 pf air variable, with a built-in 2:1 vernier, from Leeds Radio. This change had VERY important benefits that are discussed below.
---------------------------------------------------------------------------------------

I’ve been searching for a way to simply characterize the behavior of the JFET regens that I’ve been fiddling with. I wanted a single graphical “picture” - of measured data - that told you at a glance how the thing operated. It’s immodest to say, but I think I found it.

First some background.

1.) So far, the simple balance equation previously described, has worked for every regen and has been a powerful aid in understanding the interaction between regen variables. The balance equation is:

1 = w^2 * Cth * M * K * Qi.

This equation does NOT have to be calculated. It’s a “balance” equation, like a kids teeter-totter with the = sign as the pivot point. To keep it balanced means to keep the regen JUST before it’s oscillation point. If one variable increases, you know another variable (or combination) must decrease to keep the equation - and the regen - balanced. All you need to know is the DIRECTION of a variable’s change -- increase or decrease.

-- w = (2 * pi * freq), so this variable goes up as frequency goes up.

-- Cth = the throttle capacitance (in pf).

-- M = the mutual inductance between the tank coil and the tickler coil. This variable increases as the coupling between the 2 coils is increased (the coils are closer together), or, as the number of tickler coil turns is increased.

-- K = the JFET gain. This variable increases when the source resistance Rs decreases.

-- Qi = the initial Q of the tank circuit - before regeneration. This variable will change as the tank construction is changed AND as the input signal level changes.

a. --- Qi goes up if the input signal level goes down (less loading),
b. --- Qi goes down if the input signal level goes up (increased loading),
c. --- Qi changes with frequency. Probably low at low BCB frequencies, increasing at mid-band frequencies, and decreasing again at higher BCB frequencies.

With all this in mind, the balance equation is quite powerful. You can quickly mentally visualize what will happen if you change something in the regen. And you can easily interpret plotted measurements.

Here’s an example. Suppose the frequency doesn’t change (w^2 is a constant), the source resistance is left alone (K is constant), and the input signal level is steady (Qi is constant). Now, if you decrease the coupling between the tank and tickler coils (decrease M), then the throttle capacitance Cth MUST increase in order to maintain the regen balance. M goes down, and Cth goes up to balance things out. Simple as that.

Mentally visualize the balance equation when looking at data - it really helps.

(Sorry. This reads like a lecture, from a rank novice no less. I don’t mean it that way -- I just get excited about geeky stuff like this.)

2.) The variables w,Cth,M,K, and Qi, are in principle measureable, but there is an important property of the regen that I could HEAR, but had not been able to quantify - so I couldn’t measure it. Until now.

That property is “CRITICALNESS”. Lyle Williams in “The New Radio Receiver Building Handbook”,2006,page 57, describes criticalness as “ -- a tiny movement of the regeneration control causes the radio to jump into oscillation.” It’s not a good thing.

This is a biggie. Some configurations of the JFET regens were just a PITA to operate. They were loud, and they pulled-in the DX, but they were just too touchy to enjoy operating.

Other configurations were a joy. You could slowly advance the throttle capacitor and hear the signal increase in volume and decrease in bandwidth, until finally it sounded like you were listening in a seashell. You could easily adjust the regeneration to give you whatever bandwidth you wanted: wide for music, narrow for DX. Then, as the throttle capacitor was increased some more, the set would go into a gentle, low oscillation - with quiet, easy-on-the-ears heterodynes. Sweet.

Why was one configuration a PITA, and another a joy? “Criticalness”.

Here’s how I’m measuring criticalness. Adjust the throttle capacitor and increase the regeneration until you can JUST hear a 400 hz test signal. Let’s say it measures 3 mV rms across the headphones. Record that throttle capacitance value as Cth’. Continue increasing the throttle capacitance - the test signal gets louder & louder - until the set just goes into oscillation. Record that throttle capacitance as Cth.

The “criticalness” is the amount of throttle capacitance you have to play with between the point where you can just hear a signal, a the point where the set goes into oscillation.

Criticalness = ?Cth = Cth - Cth’.

The bigger this number, the more throttle capacitance range you have between just beginning to hear a signal and having the set go into oscillation.

For this particular JFET Regen-4B configuration, a ?Cth greater than, or about equal to 25 pf made for a pleasant set to operate. Much less than that, and the set was touchy - useable, but touchy.

You immediately see that criticalness is strongly affected by how much throttle capacitor shaft rotation is required to go from Cth’ - to - Cth. Verniers, “band-spread” caps, big knobs, etc. all help and all affect the minimum comfortable ?Cth value for a particular set.

Anyway, ?Cth is the important quantity I was looking for as a means of quantifying and characterizing the “criticalness” of a particular regen configuration. Quantifying it’s friendliness.
---------------------------------------------------------------------------------------

OK, here’s my one-graph characterization of the JFET Regen-4B.



It’s busy - but it’s one graph. Here’s how I read it:

1.) The relationship between the throttle capacitance Cth and the tank-tickler spacing S (coupling) is shown by the curve. There’s a different characteristic curve for different device gains, ie., for different source resistance Rs. The characteristic curve shows that as the tank-tickler spacing increases (ie.,the mutual inductance, M, between them goes down), the throttle capacitance goes up. Just like the balance equation says.

An important observation is that the audio voltage across the headphones - the volume for piezo headphones - is the SAME anywhere on a Cth vs S characteristic curve (for a given input frequency). So, as far as audio goes, one operating point (S,Cth) is as good as another. And more device gain gives you more audio volume. No surprises.


2.) The graph shows that for this setup, in order to use very little coupling between the tank-tickler (large spacing S), the throttle capacitance must be quite large.

This is where the new 540 pf throttle capacitor is important. Previously I had used a 365 pf throttle cap. The graph shows that I was restricted to tank-tickler spacing of about 1/2 - 1” using the 365 pf capacitor. However, a larger throttle capacitor allows me to use very large spacing between the tank-tickler coils. As much as 4” or more, ie., VERY light coupling.

And - VERY light coupling has a significant positive effect.

3.) It turns out, as seen on the graph, the (S,Cth) operating point - although not affecting the audio output - does affect an important parameter; the “CRITICALNESS”, ?Cth.

As you can see, at very light coupling between the tank-tickler (large S, small M), the values of ?Cth increase. The set becomes less critical, less touchy to operate.

You can also see that increasing the gain (by changing Rs) affects the ?Cth values. More gain = more volume, BUT also = more “criticalness” (smaller ?Cth values).
---------------------------------------------------------------------------------------

NOW, there is a rational basis for choosing one set of operating conditions (S,Cth) over another. How much volume do you need (choose Rs), and how much “criticalness” (?Cth) can you tolerate. AND, the less the tank-tickler coupling, the more options you have. But you’ll need a big throttle capacitor to be able to get to those very loose coupling options.
---------------------------------------------------------------------------------------

Bottom line(s) -- as most here know, but I made into a long story --

##### use as big a throttle capacitor as you can; then

##### use as little tank-tickler coupling as you can; then

##### use as much device gain as you want - until the “criticalness”, ?Cth, annoys you.

And,

##### keep the balance equation in mind

##### a single characterization plot tells a good story.
---------------------------------------------------------------------------------------

I’m anxious to see how this little regen does in the upcoming Active Device contest. The number of weak daytime stations heard will tell the tale.

73, Dan

[KE4ID]

Hi, Dan,
Nice job on keeping up with your experimental data from your latest set!

Are you and Mr. Ben Tongue related?

Keep up the good work,
Jack

[Dan McGillis]

Hi all:

Here is an indication of how well the JFET Regen-4B works.

Like most folks, I use the local daytime (groundwave) stations as an indicator of a radio’s capability. If the radio snags the really weak ones it’s a good sign.

The contest put on by Jack and the ‘Bama Boys, and the contest put on by Dave, are a good excuse to spend some hours just tuning and listening. The total # of stations ID’d is an indication of a radios’ capability but it’s also an indication of the operator's skill, and perserverance - both physical and mental - and of course location and skip conditions.

I’ve been using the contest results for the first few DAYTIME hours of listening as a an indication of radio capability. Most all the stations are groundwave, and there’s not much skill or perserverence involved on my part.

Anyway, the table below has a summary of my contest results for the first couple of DAYTIME listening hours. The more weak groundwave stations heard - I presume the better the radio is.



There are no big conclusions beyond the obvious ones. It just shows how effective this simple regen is.

Fun stuff.

73, Dan

[philwashere]

hi Dan, GREAT WORK! i tested a JFET version (will write up eventually) of the Regenerative Plate Detector (see below). 6GK5 Gm is controlled via grid voltage. the valve version worked well so i dropped in a J310 (wanted 2SK125). it differs from your regen above in that i ran the tickler off the SOURCE (unity voltage gain). i went for simplicity (lazy) and do not have the superior antenna input you and Dave use. i drew up a dual-gate MOSFET version using the second gate as regen control (RFC isolating RF from knob). i felt compelled to write as your previous three circuits described each combination except the one i used: DRAIN = HEADPHONE, SOURCE = TICKLER. i mostly use Sparkplug 16-ohm (noise control) phones and a T725.

http://home.comcast.net/~phils_radio_designs/PlateDetectorRegenerator.pdf

regards,
phil [lurking]

[Dan McGillis]

Thanks for the nice words guys - but YOU are the ones that really understand radio. All I'm doing is measuring stuff - and having fun.

Phil, thanks for sharing your web page. That's going to require some time to digest. Sure appreciate it.

73, Dan

[philwashere]

howdy Dan,

i'm still very much a novice at this radio game... but it's a fun hobby and i've met some nice people here on Dave's board. keep up the great work.

regards,
phil

[Dan McGillis]

Hi all:

For fun, I’m looking at the signals at various points in the JFET Regen-4B circuit. I don’t have a way of taking pictures of the scope screen, so I just drew the waveforms the best I could. Obviously the shapes and voltages picked off the screen are a bit approximate.

I’m using an Hitachi V-212 scope with a X10 probe. An Eico 315 signal generator supplies a 985 khz carrier modulated by a 400 hz audio signal. The signal is loosely coupled to the tank coil of the regen.

I’m a novice at this, so if something looks fishy or is just plain wrong, please say so!

----------------------------------------
In the first diagram, shown below, the approximate waveforms and measured voltages are drawn on the JFET Regen-4B circuit diagram.



1.) Starting at the lower left-hand-corner, the waveform on the gate looks like a 400 hz audio signal riding on a 985 khz carrier. I guess this looks like what you would expect.(?)

2.) The waveform at the source terminal of the JFET looks approximately like the gate signal cut in half. With 3.42 vdc at the source, the JFET is dc biased very close to the pinchoff voltage (Vp). So I guess some of the bottom half of the gate signal shuts the JFET “off”.(?)

3.) The waveform at the drain terminal surprised me. There certainly are RF and AF signal components at the drain, but the RF is kind of sandwiched between 2 audio waveforms. (?)

The throttle capacitor passes the RF component of the drain signal and blocks the AF. The 27 mH choke passes the AF component and blocks the RF.

4.) The waveform at the Bogen input looks like nice 400 hz audio. I guess the Bogen acts as an AF resistor (choke). (?)

----------------------------------------
I suspected that capacitor C1 would affect the regens’ output and behavior, but I didn’t know how. C1 bypasses the source resistor Rs. It’s quite effective at RF but much less so at AF. I’m reading about the negative feedback effects caused by bypassing Rs, but I don’t really understand it yet. Any help would be appreciated.

The waveform at the drain is the one being split apart: RF through the tickler for (positive feedback) regeneration, and AF on to the Bogen. This waveform is quite sensitive to C1. The drawings shown below represent what drain waveforms look like as C1 is changed.



As C1 increases, there is less RF component at the drain.

This has a major affect on the throttle capacitance and tank-tickler mutual inductance need to bring the set just to the oscillation point.

In the graph shown below, the values of throttle capacitance (Cth) and spacing (S) between the tank-tickler coils, is plotted for various values of bypass capacitor C1.



As C1 increases from 0.22 nF to 100 nF, more throttle capacitance is needed (less resistance to RF), or tighter tank-tickler coupling (smaller S) is needed to bring the set to the oscillation point. There is less RF at the drain pick-off point to work with as C1 increases, so the throttle capacitance and/ or the coupling must change.

In addition to the major affect on Cth and “S”, there seem to be other effects associated with changing C1.

-- The set becomes unstable if C1 is reduced too far. This may be what Krystallo was experiencing.

-- The audio increases as C1 is made quite large, but it begins to sound muffled or distorted.

Whether C1 is “small” or “large” I think depends on its relationship to the source resistance Rs -- ie., to what DEGREE C1 is bypassing the resistor. If Rs is changed, the acceptable range of C1 values will probably change. And changing C1 - as just shown - will require a change in Cth and “S”.
----------------------------------------

This darn set is EXQUISITELY interactive. Everything seems to depend on everything else. Push here -- and it pops out over there, and there.

Fun stuff.

Again, if this is nonsense, please say so -- it’s a learning experience.

73, Dan

[gusnaz]

Dan,

[Ham-er]

Dan,

Again you are doing some excellent work. By using a source resistor to bias the jfet, you get some degeneration. Bypassing the source resistor for RF or for AF reduces or eliminates that de-generation FOR THOSE FREQUENCIES ONLY.
It all comes down to the C1 reactance compared to the resistance of the source resistor, AT audio or, AT RF, or possibly both.

Some degeneration can be used to stabilize an amplifier and to broaden its response. If you did not bypass the source resistor at all the de-generation produced can have the effect of "cancelling" the modulation.

Some things for you to try:
Leave part of the source resistance un-bypassed.
Use a quartz crystal, or ceramic resonator, or series tuned circuit as the bypass for the resistor instead of just a capacitor.

Have fun, and keep on learning.
73's
kb0lxy

[Dan McGillis]

Thanks for the encouragement fellas.

Ham-er, thanks for taking the time to explain things and for suggesting other paths to explore! Comparing C1's reactance to Rs - at both 985 khz & 400 hz - does indeed shed some light on what's going on.

I'm struggling to get my head around the idea that there is both negative feedback (RF & AF from Rs) AND positive feedback (RF from tickler) going on at the same time.

Lots more to read about and try to measure

73, Dan

[macrohenry]

Hi, Dan, have you considered putting your findings in an additional internet space, like a webpage? For now your postings are easy to find, but as time goes by, they will surely get moved down the list.

Your work has had at least two positive effects on me, one being that I enjoy using your findings as reference material. The other is that it makes it clear to me how I want my regens to respond to my touch. Viewing the continuum graphically is very helpful in developing such perspective.

Is there an article archive similar to Ben Tongue's you could use?

Macrohenry

[Dan McGillis]

Mac - it’s great when someone as knowledgeable as you finds something useful in this. I just worry that because I have so little regen experience, I may be overlooking big issues, or making serious errors. Hope not.

As far as an archive of the data goes - Dave has kindly offered to put it on his website in his “visitors tech articles” section. I’ll post an address when the article is finished. (Thanks Dave !)

73, Dan

[OErjan]

thankyou Dan, you inspire me, I am going to try a FET regen this weekend (yet again)

[Dan McGillis]

OErjan - great. I’ll be VERY interested in what you find and in what you have to say!

Mike Tuggle certainly blew us all away with his FET regen in the 1AD contest.
---------------------------------

In the spirit of a “work log”, here’s another entry - on source resistance bypassing.

Ham-ers comments about bypassing and comparing C1’s reactance to Rs for BOTH RF and AF got me looking for a simple way to graphically visualize bypassing.

I’ve seen a “rule-of-thumb” in the ARRL Handbook that says C1 would be considered an “effective” bypass capacitor if XC1 ~ Rs/ 10. This is obviously frequency dependent since XC1 = 1/ (2*pi*freq*C1).

There’s a computer simulation of the bypass effectiveness of C1 in a JFET amplifier at:

http://www2.eng.cam.ac.uk/~dmh/ptialcd/spice/simulation.htm

which shows that a signal is “almost” bypassed (the gain is reduced by only -1.5 dB) if XC ~ Rs/ 3, and that there was no gain degradation if XC ~ Rs/ 33

So, I’m going to pick a number between 3 and 10, say 6.3, that will describe when “most” of an ac signal is bypassed by C1. Why pick 6.3 ? Because that’s ~ 2*pi. Then I can say “most” of an ac signal - but not all - will bypass Rs if XC ~ Rs/ (2*pi) , in other words if

(Rs*C1) is greater than (1/ f).

This is easy to remember and easy to plot. It’s helpful (at least to me) when trying to picture what combination of C1 and Rs will bypass “most” of an RF or AF signal. This simple time-constant inequality is plotted in the graph below.



The graph shows that, for example, if you use C1 = 1 nF ( the black dots) and Rs = 10,000 ohms, you are in a place where:

-- the RF is “mostly” bypassed around Rs for any BCB frequency, and
-- the AF is “mostly” NOT bypassed around Rs (& therefore there is AF gain degradation due to DEgeneration).

Also included on the graph - for information - are the calculated bias point drain current and transconductance (gain) for this particular J310 JFET as a function of Rs. The graph gives me a quick visual picture of SOME of the consequences of choosing particular Rs and C1 values.

Of concern is another discussion about bypassing in the ARRL Handbook (I have a 1967 edition - which shows my age). The gist of the discussion is that because of distributed inductance, there is a practical limit to the size of capacitor that is effective for bypassing. ARRL suggests that at low and medium RF frequencies, the bypass capacitance should be no bigger than a few nF’s lest the capacitor start behaving as an inductor. So maybe I should restrict the maximum practical C1 to the “blue curve”, C1 < 10 nF for RF bypassing. And if AF bypassing is also wanted, use a large ADDITIONAL capacitor - parallel to C1. I know Exray & Curt Reed have discussed this for power supply bypassing. We’ll see how it goes for source resistance bypassing for a regen.

73, Dan

[Ham-er]

Dan,

Nice work once again. I am goin to "steal"(copy) that chart!

Your comments about lead inductance of the bypass cap remind me of an old war story.

When I was a repair tech for radio shack on one of thier computer assemby lines a fellow technician once commented "You cant convince me that you need all those small caps connected right close to the power pin of each chip. Since they are all connected across power and ground then one big power supply filter cap should do. They are all in parallel on the power lead anyways!"

What he was not considering was that the long 5 volt circuit board run has considerable inductance at some of the higher frequencies in noise. Those caps are there to bypass that noise to ground. Because of that inductance, the caps are NOT all in parallel at those noise frequencies.

BTW I have a year 2000 ARRL handbook, a 1942 ARRL handbook, and a "west coast" handbook by William Orr, 1975.