TheRadioBoard

[Dan McGillis]

A JFET Short-wave Regen

Hi all.

This is my first attempt at building a short-wave regenerative receiver. And this is another one of those long winded work logs. Writing this stuff forces me to clarify my thoughts and to document results in case someone else can use them. If there are glaring errors, I hope somebody will (politely) point them out. This is for fun and for learning.

If you want to skip all the verbiage - here’s an executive summary:

1.) The receiver is an all JFET regen with an untuned RF amp and an audio pre-amp. It covers about 5.5 - 12 MHz. Total current drawn at 9v is about 1 mA.

2.) The Hartley regen detector is from Wes Hayward. It uses 2 separate toroid coils that are NOT magnetically coupled. NO TAPS.

3.) The band spread scheme - also from Hayward - uses “common” large value air-variable capacitors and is very effective.

4.) A throttle capacitor off the detector’s drain controls regeneration. There is a slight change in tuned frequency - noticeable when copying CW - as the throttle cap is adjusted. Suggestions as to the cause of this would be greatly appreciated.

5.) Two antennas are available - a 200’ long wire, or, a 7 MHz folded dipole. Figuring out how to couple the radio to these antennas is a work in progress. Suggestions appreciated.
---------------------------------------------------------------

Like my previous broadcast band regen attempts:

http://theradioboard.com/rb/viewtopic.php?t=1431&postdays=0&postorder=asc&start=45

http://theradioboard.com/rb/viewtopic.php?t=1678&start=15

this one uses JFETs - because I don’t know anything about tubes. I don’t know much about JFETs either - but I’m trying to learn.

Eventually I’d like to wind up with a regen receiver that will be used on the 40m ham band (7 MHz) to copy relatively weak Morse code (CW) signals.

Since my previous broadcast band regens used an Armstrong tickler coil arrangement, I wanted to try something different for the SW regen. A Hartley circuit seemed like a good starting candidate since I’d read somewhere that it potentially had a tuning range in excess of 2:1, whereas the Colpitts had less. I’ll try a Colpitts variant later for a dedicated 40m regen receiver.

I also wanted to learn how to wind and use toroid coils. The very thought of trying to measure and adjust low inductance toroid coils using my old grid-dip methods caused me to “invest” in an AADE L/C meter. Boy, what a great instrument. Should have gotten one years ago.
---------------------------------------------------------------

After searching the web, the 2 sources of JFET Hartley regen circuit info I “borrowed” from the most are:

“Experimental Methods in RF Design”, by Wes Hayward (W7ZOI) et al, 2003;

and Charles Kitchin’s ‘98 QEX article, “High Performance Regenerative Receiver Design”,

http://www.arrl.org/tis/info/pdf/9811qex026.pdf

Wes Hayward used a variant of the Harley circuit that does NOT use a tapped coil - there is no transformer action. Since my arthritic fat fingers don’t like fiddling with coil taps - especially on tiny toroids - his Hartley regenerative detector design really caught my eye.

He also talked about a way to use “common” large value variable capacitors in a bandset - band spread configuration that looked interesting. I “borrowed” that idea too.
---------------------------------------------------------------

Here’s the schematic of the SW regen as currently breadboarded:



It’s a busy schematic because I try to cram as much info into it as possible for reference.

1.) The battery bypassing and decoupling scheme is what worked for the BCB regens as the battery aged - so I used it again here. The 2.5 mH chokes were the only small chokes I had. The total current drain is just under 1 mA - which should be good for battery life.

2.) The audio preamplifier circuit is also from the BCB regens. It’s tailored to my bad ears. You can use a sensitive piezo headset connected directly - or - for a lot more audio oomph, connect a Radio Shack amplified speaker (essentially a LM386) to the output and connect the piezo headset to that. The piezo headset has a LOT more volume than a “Walkman” type headset so a diode limiter is also used on the RS amplifier output.

3.) The Hartley regen detector is from Hayward. It uses 2 separate toroid coils that are NOT magnetically coupled. NO TAPS. The inductive reactance at 7 MHz is about 247 ohms -- in the ballpark suggested by Jim Kearman on this RadioBoard.

Hayward says that “some experimentation” is required to set up controllable regeneration. Indeed! The number of turns on L2, the value of R1, and the gain (gm) of Q2 all affect the amount of energy fed back. Increasing the # of turns on L2, decreasing the value of R1, or increasing the Q2 gain will all increase regeneration. The trick is to adjust them so that the feedback can be controlled by the throttle capacitor Cth.

The plots below give some idea of the interactions observed. I don’t think I’ve found the “best” set of conditions yet - but the thing works pretty well as-is.



Lowering Q2’s gain (gm) - ie. using a larger source resistance Rs - makes the transition into oscillation much less abrupt. And, the affect of Cth on the tuned frequency is reduced. With L2 fixed, varying R1 changes the required Cth vs frequency curve.

4.) The band spread scheme - shown above the schematic - works very well. But I had to use a spreadsheet to manage the calculations and explore the options as all the capacitor variables were changed.

The way I elected to wire-up my dual gang Leeds Radio capacitors leaves the frame ungrounded. Even though a plastic shaft extension and a grounded planetary are used, there is enough hand capacitance to be annoying when tuning CW. I’ll have to fix that on the next go-around.

5.) The throttle capacitor is off of Q2’s drain - shunting RF to ground. The less throttle capacitance, the less regeneration. Kitchin - in his QEX article - shows a similar throttle cap arrangement for a Hartley regen. He suggested that the RF circulates in a loop from the drain of Q2 - through the throttle cap - to ground - through L2 - back to the source terminal of Q2. I don't know - but it works pretty well. Tuning an AM station, you get that nice seashell sound effect as you adjust the throttle cap and creep-up on the oscillation point.

6.) I’m using a second winding on L2 to couple-in the RF signal from Q1 into the tank circuit of Q2. When the RF output from Q1 was capacitively coupled directly into the tank circuit of Q2, changing the antenna coupling capacitor affected the tuned frequency. I don’t know why. This method of magnetically coupling the two stages stopped that.

7.) When copying CW with the receiver - the regen will be oscillating. I’ve read that an untuned RF amp in front of the regenerative detector will help keep the regen’s signal from being radiated and causing interference. (I know some say that with a JFET regen the radiated power is too small to be concerned about, and the real reason is to present a constant impedance to the following stage. Again, I don’t know.)

Anyway, I wanted to try designing a simple grounded gate JFET RF amp, Q1, for the purpose of presenting a known load to the antenna. If there are other benefits - great. So I followed Professor Kenneth Kuhn’s lecture notes (a gold mine of information!).

http://www.kennethkuhn.com/students/ee351/text/

8.) The way the rf signal is taken from the antenna and input to Q1 is not shown in the schematic. I’m still experimenting with, and learning about, different antenna coupling methods - which I’ll discuss in a separate section below.
---------------------------------------------------------------

Here’s some pictures of the breadboarded radio.





1.) I wanted to see if you can make a decent “low frequency” SW prototype regen on one of those plastic plug-in breadboards. You can! Even with my sloppily wound toroids hanging over the board, the set works at least up to 12+ MHz.

2.) As-is, the receiver is stable enough to easily copy a CW or SSB signal over a relatively long (~ 15 min.) time period without tweaking anything. I was surprised at how well it worked for CW. It is temperature sensitive, but changes are slow and it isn’t a problem (for a receiver).

3.) There are two major annoyances when copying CW:

-- the slight amount of hand capacitance previously mentioned,
-- and the biggest annoyance, a slight change in frequency as the throttle cap is adjusted.

You don’t notice it tuning AM stations but it’s apparent when listening to CW. I’d like to figure out what causes this and fix it. ANY SUGGESTIONS WOULD BE APPRECIATED.
---------------------------------------------------------------

Now for the antenna coupling which is a work in progress. Suggestions are appreciated.

I’m trying to use 2 different antennas: a 7 MHz folded dipole, and a 200’ long wire used for crystal radios .

The folded dipole - at it’s 7 MHz resonant frequency - presents about a 300 Ω resistive source to the receiver. But the long wire’s resistance ( impedance) may range from about 50 Ω to well over a 1000 Ω as the frequency is changed. The maximum power is transferred from the antenna to the receiver when the antennas’ impedance (at the end of the transmission line) approximately equals the radio’s input impedance (resistance).

A.) Using the 7 MHz folded dipole antenna.

If I’ve read Prof. Kuhn’s lecture notes right ( a big IF) - and did the calculations right - the input resistance to the common gate RF amp Q1 depends on the source resistance Rs used, as shown in the following plot.



The plot (if right) shows how the input resistance of a common gate JFET amp varies with Rs for 3 types of JFET.

By choosing a J310 JFET for Q1, and a source resistance of 4k7, I think the input resistance to the receiver is about 300 Ω. That’s a good match to the folded dipole antenna so that the relatively weak CW signals at 7 MHz should be received well, but other frequencies less well.

Since some powerful short-wave stations are transmitting on the “other frequencies”, they should be attenuated but the receiver should readily pick them up. Hopefully, my 2 strong local BCB signals at 790 & 1470 kHz will be attenuated enough to prevent them from overloading the receiver.

The next figure shows the simple coupling scheme used for the folded dipole. It’s just - I think - a 1:1 transformer.



On the receiver side, the 1.18 µH inductance and the 470 pF coupling cap form a series resonant circuit, having minimum impedance around 7 Mhz. I don’t know if that makes any difference.

Although I’ve only played with this scheme for a few days, it seems to work well. Relatively weak (judging by reception on a ham rig) CW signals on 40m are picked up nicely. There has been no BCB overloading (so far) and SW signals are heard all across the 5.5 - 12 MHz operating range.

B.) Using the 200’ long wire antenna.

I’ve got a lot to learn about coupling this beast into the receiver. I know it’s WAY to big for a regen - a shorter piece of wire picks up lots of SW stations at night. But the shorter piece of wire doesn’t pick up the relatively weak 40m CW signals.

I’ve tried the simple coupling scheme shown below - with mixed results.



It brought in the weak CW signals at 7 (&10) MHz all right - but there was significant BCB pickup also. And a few off-frequency SW stations overloaded the receiver at night.

Any advice on designing antenna coupling circuits for a long wire antenna to a ~ 300 Ω load would be greatly appreciated.
---------------------------------------------------------------

Conclusion? The receiver works good! With a few annoyances to be fixed.
---------------------------------------------------------------

If you got this far without falling asleep - thanks for reading it all.

If there’s stupid stuff in here - please point me in the right direction.

It’s a learning experience and lots of fun for an old retired guy.

73, Dan

[KR1S]

Dan,

Good work! And thanks for all the data, which I am still processing. FETs rule! While you are experimenting, you might pick up some J202 and/or J203 FETs, cheap from Mouser, and try them in Q2. They are audio FETs. A J201 wouldn't oscillate reliably at MW in mine, but the 202 and 203 are a little hotter and should work. I ended up using a 2N5639 switching FET (also cheap from Mouser), which gives me exceptionally smooth regeneration. Q1 will set the noise floor for the receiver, so if you have a low-noise FET like your J310 there, a little more noise in Q2 won't be noticeable. I can't tell the difference, noise-wise, between a J310 and the 2N5639. My writeup is here: http://qrp.kearman.com/html/mwregen01.html

If you only want to tune 40 M, hunt around for a small variable and pad it with fixed caps to reach the total C required. Bandspread and band-set caps are such a nuisance!

As far as interactions between controls causing frequency shifts, that is pretty hard to eliminate. Everything is hanging on an oscillator, and anything that changes the current through the FET is going to affect how the FET interacts resistively and capacitively. A better route is to make it stable enough that you rarely have to adjust the throttle cap, at least over 20 kHz or so. Yes, signal strengths vary, but it's possible to have a regeneration setting that will still let you tune around without fiddling. That, IMO, is the sign of a good regen. If all else fails, try a 150k or so resistor across the tank. This method is frowned on my some purists but FETs are harder to control than tubes, and the goal is to make a usable receiver, not satisfy the elitists.

It shouldn't drift! Find and fix whatever is causing that problem. It may be as simple as building a good enclosure, which will help with the hand-capacitance problem, too.

Mis-matching the antenna for out-of-band signals isn't going to make a big difference. Just match your antenna and never mind about what's going on elsewhere.

BCB break-through is another critter. You may experience it with regeneration backed all the way down. There are a few ways this can happen, including the first audio stage acting like a crystal radio. A rudimentary low-pass filter made from a fairly large rf choke and a .001-uF cap, on the input to the audio amp, may work. A high-pass filter on the receiver input would work, too, if the problem is detection in the rf section of the receiver. But then you have to deal with matching the filter etc. Good shielding may be all you need. And bypass-bypass-bypass dc lines, headphone jacks, etc.

For the 300-ohm input transformer, did you wind it bifilar? Don't know that it will make much difference, but might be worth trying. I know you hate to wind toroids (am I the only person who likes winding toroids?), but give it a go. Instead of a -6 toroid you might try 9 bifilar turns on an FT37-43 ferrite core. For broadband transformers the rule of thumb is to make the winding reactance 4 times the circuit impedance at the lowest frequency. Nine turns on an FT37-43 comes out to 27.3 uH, 1200 ohms at 7 MHz.

A 200-ft wire should let the receiver blow your ears off. With it or the folded dipole, it's okay to be concerned about interference caused by radiation. Yes, the radiated signal will be minuscule, but there's a thing called "good practice."

Keep up the good work! And don't let those toroids get you down.
_________________
Jim, Amateur Radio Station KR1S
Celebrating My 50th Year In Ham Radio


Just build it!
http://kr1s.kearman.com/
http://qrp.kearman.com/

[Ham-er]

[Ham-er]

[Dan McGillis]

Thanks for the thought provoking comments guys. You’ve given me more to read up on -- which is great. I really appreciate it!

Jim:

-- Thanks for the practical advice about the bandspread & band set. I agree!

-- And that’s good advice about adjusting conditions so that the variation of the required Cth vs frequency is at a minimum at 7 MHz. I hadn’t thought about it like that. If you look at the 2nd data plot - by dumb luck I think that’s about where I’m at. (ie. 7 MHz is about at the minimum on the red curve.) And indeed, I really don’t have to adjust Cth when tuning from say 7.025 to 7.050 MHz. Guess it’s a case of “if it hurts when you do that - don’t do that”.

-- re. the slight thermal drift. I’ve got to do some more reading about that. The J310 spec sheet shows me the (Vgs/Vp) ratio at which Id doesn’t change with temp - and I’m not operating there. Hmm.

-- I’ll have to more carefully re-evaluate the BCB “breakthrough” with the 200’ antenna in light of your comments.

Sure wish I knew how to design an “Antenna tuner” for a 300 Ω load.

-- I’m beginning to like winding toroid critters. Just takes practice I guess. My wife says “they’re CUTE”.

-- No, the 300 Ω input transformer was not wound bifiliar. Good suggestion. -- I presume I’d get more signal transfer with the tighter coupling (?) But won’t 43 material give me greater losses at 7 MHz.?

Ham-er:

--Phase shift ! I hadn’t thought about that. More reading to do. Thanks!

-- I’ll take a look at making L2 smaller. Will have to re-adjust R1 & probably Q2’s gm too. Based on Jims comments, I’ll be looking for a flatter Cth vs freq curve at 7 MHz. Will have to make more precise measurements also.

-- And I didn’t know about mixer noise either. Fact is, I don’t know anything about noise issues. More reading to do.

Good stuff! Thank you again guys.

73, Dan

[KR1S]

Dan:

1. Drift: At the current you're using, I wonder if it isn't stray air drafts. There can't be that much heating due to current flow.

2. Antenna tuner: Less necessary with a receiver. It wasn't always a problem with transmitters until we went to touchy transistors with limited breakdown voltages and untuned low-pass filters that only worked properly into a 50-ohm load. On the receive side, reflected energy won't hurt anything. An antenna tuner isn't really a tuner, it's an impedance transformation device. You've got a 50-ohm source and an unknown complex-impedance load, and you want to transform that load to 50 ohms so the transmitter is happy. Sometimes "tuning" an antenna will peak the signal, if you happen onto an L or C value that compensates for whatever the antenna looks like at that frequency. The input of your FET preamp is pretty forgiving (and untuned), so it isn't very important to try to match it.

3. Type 43 material makes excellent transformers at 7 MHz. You don't start to see appreciable loss until around 18 MHz, and even then it isn't bad. I used a type -61 core on my BCB receiver because I had the idea to float the input to "break" the shield and prevent noise picked up by the transmission line from getting onto the receiver chassis. That turned out to be unnecessary and it works just as well coupling directly into the preamp.

73,
_________________
Jim, Amateur Radio Station KR1S
Celebrating My 50th Year In Ham Radio


Just build it!
http://kr1s.kearman.com/
http://qrp.kearman.com/

[Bob Weaver]

Hi Dan,
Good work!

I've always liked Hartleys, and this is a very interesting version.

I've got a bunch of JFETS sitting around waiting for to be used, and a shortwave receiver is starting to look better and better all the time.

If you're unhappy with the frequency shift caused by the throttle capacitor, you might want to explore other regen control options. For example, you could remove the throttle cap, and then replace the 47k source resistor Rs with a 47k pot. You could try it with the wiper connected directly to ground, and then connected through a relatively large capacitor (>.01µF) to ground. It might work better; it might be worse, but it never hurts to experiment.

[Dan McGillis]

Good stuff again - thanks fellas.

Jim:

-- Stray air drafts are exactly where the slight termal drift is coming from. If I gently blow across the circuit, there is a slight (~ 100 Hz) frequency shift. You’re right - need a box! Thanks.

-- Yep, I know it’s not an antenna “tuner”, but I just can’t break the habit of calling it that. Your comments & suggestions give me hope of finding a simple way to tame the 200’ beast. I’ll have another go at it today if I can dispose of - or get out of - my “honey do” list.

-- ok on the 43 material. I’ll try a bifiliar 1:1 transformer by twisting 2 #24 wires together - then wind that on a toroid. I presume that should work.

Bob:

-- YES, a pot to change the gain. I’d forgotten about that option. Think I’ll try keeping the throttle cap AND add a small pot in the Rs path - then use the pot to tweek the gm a bit as a fine tuning regen knob. Worth a try. Thank you!

-- Boy, if YOU make up a JFET SW regen, I know it will be a real beauty. Please keep us informed if you make one. I’d sure be interested in what you come up with and HOW you came up with it.

73, Dan

[KR1S]

[Rune]

Nice stuff you have going Dan.
For your drifting problem you could try to shield the detector JFET with a little metal cup over it with some cotton stuffed inside. That really helped on my regen. It will also shield the JFET from RF pickup from the surroundings.
My regen worked much better with the components placed on a proper PCB, so that should be worth a try also.

[Dan McGillis]

Thanks Rune.

That's a good idea about the little metal box shield + cotton. I'll try it.

Your regen looks very professional. Eventually I'll do a proper job and solder everything down, but for now - it's still "under construction".

73, Dan

[Ham-er]

A 300 ohm ATU?

It can be very difficult to design an ATU to have a CONSTANT 300 Ohm Z at all its various range of frequencies.

It would require the use of Differential reactances. (large range differential capacitors?)

And indeed it may not be necessary at all in a receiver.

I can think of one kind of compromise solution, if you dont mind some loss. In a regen a small amount of loss might not even be an issue.

You can use a 3dB or 1dB resistive "pad" attenuator on the input or output of the ATU so that circuitry outside of the ATU itself "sees" 300 Ohms.

[Dan McGillis]

Thanks for thinking about the ATU issue Ham-er.

For now I'll follow your suggestion and presume it's not necessary.

My cup "runneth over" with to-do items for this regen. I'll try to clear them away, then revisit the ATU issue.

73, Dan

[Dan McGillis]

Lets talk a bit about the slight but annoying frequency shift as the throttle capacitor is adjusted.

1.) It’s only an annoyance when copying CW signals -- and when you then try to adjust the receiver to be just over the oscillation point. (I assume this is the point of maximum selectivity in the oscillation region. ??)

2.) Bob Weaver tweaked my memory about controlling the regeneration with a pot. On some of my BCB regen breadboards, I had used a pot as part of the source resistance because I wasn’t sure what value of Rs I needed. It made a nice fine tuning knob for regeneration control. (Changing Rs changes the JFET’s gain.)

3.) I tried the same thing for this regen. Instead of a 47k resistor for Q2’s Rs, I substituted a 22k resistor and a 27k pot with the wiper to ground . The pot is an old WeCo (Western Electric) “Type J” - whatever that means.

-- I set the pot at 25k,
-- set the throttle capacitor so the receiver was oscillating,
-- and tuned in a CW signal at 7.030 MHz.

Then I varied the pot to take the receiver slightly more into oscillation, or just slightly out of oscillation. It only took about +/ - 500Ω for this particular setup.

4.) The frequency DID still vary a little - but no where near as much as was the case using the throttle capacitor alone.

There was no hint of hand capacitance when tuning the pot, and no scratchiness.

The adjustment in & out of oscillation was very smooth with no hysteresis and it had a very nice “feel” to it.

I think it’s a winner for CW. Thanks Bob!

73, Dan

[KR1S]

Quite a few regen builders here also hang out at the Yahoo regenrx group. http://groups.yahoo.com/group/regenrx/ A recent post, which I think you can read without joining the group, discusses the characteristics of various methods: http://groups.yahoo.com/group/regenrx/message/8723 The author's conclusions, if I'm understanding him, is that only by adjusting gain and frequency in separate stages can you mitigate the problem you describe.

I think we can summarize by saying it's very difficult to adjust the level of oscillation without also affecting the frequency. You vary drain current in a FET by varying the gate-channel width, which has to affect input capacitance IMO. You'll notice the shift right away on CW or SSB. In fact, the regen control can be another way to fine tune a signal. Regens are two-handed radios.

In addition to smooth regeneration control, what's more important for a CW or SSB regen IMO is whether strong signals pull the frequency. Can you tune ever closer to zero beat with a strong signal without the detector suddenly "pulling in" due to phase locking? Avoiding pulling-in and the ability to move smoothly across the threshold into and out of oscillation are the hallmarks of a good regen. If it meets those criteria, it's not hard to fine tune a signal. Once you find the right level of regeneration you can tune across the signal to set the beat note or zero beat an adjacent interfering signal.

73,
_________________
Jim, Amateur Radio Station KR1S
Celebrating My 50th Year In Ham Radio


Just build it!
http://kr1s.kearman.com/
http://qrp.kearman.com/