TheRadioBoard

[Ham-er]

Krystallo,

Having a very high-Q tuned circuit connected to a regen detector IS over kill and therefore may have something to do with the "popping past" a station due to the q-multiplying effect making the selectivity so sharp.

Not oscillating at part of the band though is a separate issue.

The issues you are having may or maynot be due to using DX crystal set components with a regen.

As a test disconnect both the antenna and its associated ATU components to see if it will then oscillate at all parts of the band.

Other regen designs have also been known to jump past a station due to high selectivity, but the solution was usually bandspreading. Also on some bands a resistor was placed across the coil for 1 or 2 bands to purposely de-Q the tuned circuit (only on that particular band).

Also you might try adjusting the source resistor value. All mpf-102's (and other jfets) are not the same even within the same batch from the same manufacturer. The IDss changes from one transistor to another, and so the source resistor which sets the bias may need to be changed.

BTW a gate voltage causing 1/2 pinchoff is considered class-A

Good luck, I hope some of this helps.

[krystallo]

Hey Ham,

I just sort of had a "gut feeling" that there may be issues here ,even before I built this set. But of course I did it anyways ("brave" or stupid-you decide!).


I suppose I can just yank out the litz tank coils (and also if I have to, yank the litz bias coil too) and just swap these out to DCC B'weave coils. That should throw a nice bucket of cold water on any possible coil Q fire!

Also would there be any problems dropping the bias below 2K? If so ,how low could I go before I have an "issue".

-OR- What would be the impact of RAISING the bias R value ??? If valid how HIGH could I /should I try?


de N1NQC

[Dan McGillis]

Hi K:

Here's some thoughts FWIW:

1.) The 2k across piezo ringer elements is probably way to low. Their audio impedance in series is ~ 14k. I'd try >140k across them.

2.) Taking the audio off the source resistance Rs, like regen-2B does, gives you only a small amount of audio to play with. The piezo ringers sound pressure output is proportional to this voltage across them. This voltage goes up as Rs goes up -- even though the gain, gm, of the jfet is going down. So, I'd start at about Rs = 10k and try to hear something from a moderate signal. Then increase Rs in steps until the radio quits working.

3.) Use as large a throttle capacitance as you can,(~ 365 pf). Then, starting with the tickler coil as far away from the tank as you can get it (feet), slowly increase the coupling between the tickler and tank coils. Couple them axially, end-to-end. Eventually the coupling should increase enough that the set goes into oscillation. WATCH YOUR EARS.

If it doesn't, try reversing the tickler leads (phase issue) and repeat.

4.) As hamer said, use 9v. You should have about 2-3v or so across Rs to ground. And indeed, disconnect the antenna. I think you should be able to hear something without it. Then slowly increase the antenna coupling and see how the set reacts - you may have to change the throttle cap setting and/or the tickler coupling as you couple-in more signal.

Good luck

73, Dan

[Ham-er]

[Dan McGillis]

Hi all.

Having spent about 4 months playing with a regen where the audio is taken off the drain,

http://theradioboard.com/rb/viewtopic.php?t=1678&postdays=0&postorder=asc&start=0

I’ve returned to this configuration - with the audio taken off the regen’s source resistor.

The audio output voltage is down about 15 dB compared to the the drain takeoff - but the regen behaves SO much better that it’s worth the hit. With a large source resistance - ie., low device gain - the approach to the oscillation point is slow, controlled and pleasant. And, there appear to be no bad habits such as a tendency to howl etc. It’s just a more gentle beast.

Kitchin consistently recommends taking the audio off the source lead and I’m beginning to experience why. And the old timers seemed to consistently recommend low device gains. Again, sound advice.
-----------------------------------
But -- the audio level is down so it needs a little help. A good reason for me to learn how to design a simple audio amplifier - something I’ve always wanted to do.

Another reason to put a little audio amp after the regen is that my hearing has deteriorated badly due to that disease, A.G.E. Something tailored to suit my particular ears would be a BIG help.

I found an excellent source of design information on the web: Prof. Kenneth Kuhn’s lecture notes.

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

This site is an absolute GOLD MINE! Ken says he gets e-mails from all over the world from folks using his notes.
-----------------------------------
So, here’s my latest attempt at a simple regen + a simple audio amp companion.



It’s very pleasant to operate and is quite sensitive. It will pick up 10w travelers information system transmissions from the PA turnpike - that even my Yaesu 757 transceiver has trouble getting.

A Bogen T725 on the amp’s output makes a nice impedance matching scheme - and hooking a little Radio Shack amplified speaker to that makes for a surprisingly decent armchair radio.
-----------------------------------
Again, I realize that all this is old hat to most folks here, but it’s been a great learning experience and a whole lot of fun. Thanks for putting up with this long-winded “work log”.

I think my next project will be to try a simple 40m CW regen receiver -- I’m following exray’s posts with great interest.

73, Dan

[Dan McGillis]

Hi all.

Here’s the broadcast band JFET regen I plan to use in Dave’s upcoming ‘09 home-brew radio contest, JFET Regen-5D. Even though the layout is still modular on a breadboard, the circuit is finally soldered on a prototype board. It’s “done”.





The JFET Regen-5D circuit is similar to previous circuits in this work log but It does have some nice features the others didn’t have. None of these features are new - but they were new to me and I enjoyed experimenting with them. I’ll pass along some observations hoping they might be of use to other regen novices like me.

1.) A single ferrite rod is used for both the tuning and tickler coils.

The mutual inductance between the tank coil and the tickler coil is adjusted by sliding the tickler coil along the rod near the rod’s end. The mutual inductance between the coils is given by:

M = k*sqrt( Ltank*Ltickler ).

Because the coils are on the same ferrite rod, the coupling coefficient between them (k) is assumed to be constant and probably nearly equal to 1. But the inductance of the tickler coil (Ltickler) changes as it’s moved along the rod. It’s inductance decreases as it moves toward the end of the rod. So moving the tickler coil changes Ltickler, which changes M.

The tickler coil was moved along the rod until the set “just” oscillated at 530 kHz when the throttle capacitance was at it’s maximum value (540 pf). Then the tickler coil was fixed to that spot on the rod. The goal was to use as little feedback mutual inductance as possible and as much throttle capacitance as possible.

2.) A switched dual gang capacitor increases band spread.

I like this technique because it’s simple. I’m using a Leeds Radio dual gang 10 - 270 pf air variable cap in the tank circuit. Any dual gang cap can be used as long as the 2 sections have the same capacitance. The Leeds caps are nicely made, have a built-in 2:1 vernier, and are relatively cheap. Dave has talked about them.

The idea is to switch the two capacitor sections between a series connection of about 5 - 135 pf, and a parallel connection of about 20 - 540 pf. This cuts the broadcast band into two tuning ranges - without changing the tank inductance which would upset the mutual inductance between tank and tickler coils. Choose the tank coil inductance so that it resonates with the maximum capacitance (540 pf in this case) at the lowest desired frequency (520 kHz in this case). A simple approximation is:

Ltank ~ (159^2)/ (Cmax*Fmin^2); with Cmax in pf, Fmin in MHz, and Ltank in uH.

For Cmax = 540 pf, and Fmin = 0.520 MHz, you get the required Ltank ~ 173 uH.

I used a miniature, non-rf rated, DPDT switch to swap around the capacitor sections. Since this switch is smack in the middle of the tank circuit, there is probably a big reduction in the tank Q. But I figured a regen could tolerate low tank Q and still give good results. I’ve used this technique in crystal sets - with excellent results - but in that case I use a good ceramic DPDT rotary switch.

Anyway, this technique easily cuts the BCB into two tuning ranges: 530 kHz to about 1100 kHz+, and 1100 kHz to 1700 kHz+, ie more band spread.

On top of that, this dual gang cap has a built-in 2:1 vernier. So when mated with a common 6:1 planetary reduction drive (see Dave’s site), you get a 12:1 reduction in addition to the split band. Tuning of the BCB is spread out and you can easily separate stations even at 1700 kHz.

I try to judge how much total band spread can be achieved - and whether it will be sufficient - by calculating the number of kHz tuned per 5 degrees of tuning knob rotation. With a 1” diameter tuning knob, the minimum angle my shaky hand can reproducibly turn the tuning knob is about 5 degrees. Larger diameter knobs of course improve things.

Doing the calculations for the split band arrangement with this straight-line-capacitance capacitor, and a 12:1 total vernier reduction, gives about 5 kHz per 5 degrees of tuning knob rotation at the maximum frequencies of 1100 kHz or 1700 kHz. Not bad when there’s 10 kHz separation between channels. Even a tuning rate of 10 kHz per 5 degrees of knob rotation is useable - but just barely.

3.) Hand capacitance is “eliminated”.

I wanted to see if mounting the tuning capacitor behind a small 3” x 3” piece of grounded one sided copper circuit board would eliminate hand capacitance effects. It does. I didn’t need a big metal front panel.

But, as you can see, figuring out a decent dial decal to paste on the 3” x 3” panel is still a work -in-progress. I need to reread exray’s posts.

4.) The audio output is diode limited.

For some signals, this regen is just TOO LOUD. Using a piezo element headset, 5 mV of peak audio across the headset is about the hearing threshold for my old ears. But a 50 mV peak audio signal seems loud enough. This set will easily put out a few volts of audio signal if you let it. Ouch. The output needed a simple diode limiter.

So, in my diode stash, I looked for diodes with the highest saturation currents to use as clippers. Some 1N60 diodes had measured values of Is ~ 1.2e-6 amps, and N ~ 1.2. A pair of these clip and limit the audio peak voltage to about 0.2v. That’s still uncomfortably loud - but no longer painful. I wish I had diodes with even larger Is values that would limit the output even more.
------------------------------------------
Bottom line - the set works pretty well and is a pleasure to operate. It has lots of band spread, plenty of volume, and well behaved regeneration.

I do realize that I’ve been babbling on and on about old stuff that’s well known to most folks here - so thank you all for putting up with this work log. But as a novice, it sure has been fun. If I’ve done something wrong or there are better ways of doing things, please let me know. It’s a learning experience.

73, Dan

[Rune]

Nice Dan.
I'm also using a diode limiter in my regen, which is not fully completed yet, but hope to do in soon. I've had some problems so to speak.
Good luck in the contest

[Austin Hellier]

Dan,
I've scrounged enough bits and pieces to build the two FET version of your regenny. However, I'm going to finish the one FET version first.
As for the Rs value, I've been using a 22k across the ceramic earphone, which seems fine. Next, I'm going to replace the air cored coil with a ferrite loop one - I've managed to scrounge a really long ferrite rod - 180mm and will also install the feedback coil, but am not sure that just one turn will do it for me.

With the second FET in place certainly. The author of the article who inspired me along with your rig, suggests that two FETs are better than one for a number of reasons.

Anyway, am going to give it my best shot.

Austin Hellier

[Dan McGillis]

Hi all. This is a “work log” entry -- about the Leeds Radio dual gang “5 - 270” pf capacitor.

In calibrating the tuning dial on the JFET Regen-5D I’ve learned something about the Leeds Radio dual gang 5 - 270 pf capacitor that I thought I’d pass along. Turns out, it’s NOT a straight-line capacity air variable, but rather it’s a straight-line wavelength capacitor.

This means it has better band spread than I had hoped - but - the relationship between capacitance vs dial rotation is more complicated than I had assumed.
----------------------------------------------
Old timers know, but us novices are learning, that the plates on an air variable capacitor can be shaped to give a desired capacitance variation as a function of the angle of shaft rotation. There are evidently 3 common plate designs (Terman, pg. 121):

a.) Straight-line capacity; the plates look to be approximately semicircular and rotate around a central shaft. The capacitance is a linear function of the angle of shaft rotation. For this kind of capacitor, I approximate the relationship of the capacitance (C) to the angle of shaft rotation - or, degree of mesh (M) - by the equation:

C = (( Cmax - Cmin)/180)*M + Cmin.

b.) Straight-line wavelength; the plates still look to be approximately semicircular, but they rotate about an offset shaft. These are designed so that when they tune a coil to resonance, the wavelength at resonance is a linear function of the angle of rotation. The relationship between capacitance and angle of shaft rotation is more complex than the simple straight-line capacity relationship.

c.) Straight-line frequency; the plates look elliptical and rotate about an offset shaft. When they tune a coil to resonance, the frequency at resonance is a linear function of the angle of rotation. Again, the relationship between capacitance and the angle of shaft rotation is complex.

Evidently, straight-line wavelength, and especially straight-line frequency capacitors are more desirable for radio tuning. When tuning an inductor, the higher resonant frequencies are more spread out across the tuning dial. The more common straight-line capacity air variable bunches the high frequencies together.
----------------------------------------------
The dial calibration stage of a project is always interesting to me. I finally get to “see” what the set “hears”.

If you have a signal generator, it’s relatively easy to mark out the correspondence between the tuning capacitors angle of shaft rotation (or a dial #) and the tuned (resonant) frequencies.

But if you have to rely on listening to off-air stations at known frequencies to make out the correspondence between the angle of shaft rotation and frequency - well, things get a little more dicey. The more stations logged the better, but there are inevitable gaps.

For those who like to fiddle with math, an obvious thing to do is to “curve fit” the measured frequencies and shaft rotation angles to a mathematical equation. Then use the equation to calculate the frequency corresponding to ANY angle of rotation or dial setting - or vice versa. With an equation in hand, you can easily make up a table of dial settings for every 10 kHz channel. This is a great way to know EXACTLY where you are on the dial. You can sit on a frequency and wait for that DX station to pop up again.

A “simple” equation that can be made to fit just about any frequency (F) vs angle of shaft rotation (M) relationship is a polynomial equation:

F = a0 + a1*M + a2*M^2 + a3*M^3 + ........

“All” you have to do is figure out the right values of the constants a0, a1, a2, a3, etc.

Fortunately, there’s a web site that will do that:

http://www.shodor.net/chemise/tools/regressionjava/index.html

Enter your measured frequency (F) vs M (or dial setting) data, and the program will calculate the values of the constants ao, a1, a2, a3, etc. .
----------------------------------------------
For the Leeds Radio dual gang capacitor, with both sections in parallel, I found that the following equation fit the measured frequency vs shaft rotation angle data quite well:

F = ao + a1*M + a2*M^2,

where F is in MHz, M is in degrees ( 180* = fully meshed), and

a0 = 5.399722
a1 = 0.279162
a2 = 0.01527

Once you know the frequency (F) for any value of capacitor shaft rotation (M), if you know the inductance (L) in your tank circuit, you can calculate and plot the capacitor’s capacitance as a function of the capacitor shaft’s angle of rotation.

C ~ 159^2/ (L*F^2),

where F is in MHz, L is in uH, C is in pf.

Results for the Leeds Radio capacitor are plotted below.



The red dots are off-air measured data - ignore the one red dot where I obviously made a measurement error. The blue dots are the fitted equation. The black dots are the results you’d expect if the capacitor was a straight-line capacity capacitor.

This plot above shows that the Leeds Radio cap is NOT a straight-line capacity unit, but the plot doesn’t tell us what it is.

However, using the capacitance values C) and converting them back into frequency (F) - or into wavelength (WL) - and plotting F and WL vs the shaft rotation, gives the answer.



The Leeds Radio dual gang capacitor, is clearly, a straight-line wavelength capacitor.

Wavelength is: WL ~ 300/ F, where WL is in meters, and F is in MHz.
----------------------------------------------
Now that I know what the capacitor is, and, I have an equation that describes it’s behavior, I can construct the following useful table for the capacitor:

-- the expected (approximate) capacitance versus the capacitor’s shaft rotation,
-- the expected tuned resonant frequencies (for L ~ 175 uH) versus shaft rotation,
-- the expected band spread versus shaft rotation.



In addition to the VERY accurate calibration chart/ dial that can be made, the equation approach lets you play “what if” games using spreadsheets on your computer. What if I changed the coil inductance; how would the frequency tuning and band spread be affected? What if I used the capacitor gangs in series or in parallel; what frequencies would be tuned and what would the band spread be? What if I added trimmer capacitors in series and/ or in parallel with the variable, etc.,etc. The frequency vs shaft rotation equation gives you a powerful analysis tool.
----------------------------------------------
Geesh, you say, this is making a big deal out of something simple. True, but it’s a hobby after all, and I enjoy the geeky side of it. Plus, it’s kind of fun to first calculate what should happen - then see if it really happens. If it does - or doesn’t - either way, I learn a little bit more about radio.

Thanks again for putting up with the long-winded novice babbling.

73, Dan

[KR1S]

[Dan McGillis]

Hi Jim.

Just so I'm clear about your comments -- when you say " I designed the LC circuits for a mid-band impedance of about 200 ohms." I presume you mean that at the mid-band resonant frequency:

2*pi*F*L = 1/(2*pi*F*C) ~ 200 ohms,

Right?

I gather that the parallel impedance of the tank circuit at resonance which is:

Z = Qloaded*(2*pi*F*L)

is not what your talking about.

I'm just starting work on my 7 MHz regen - and I'm trying to follow your advice. Thanks!

73, Dan

[golfguru]

Hi Dan,

Not trying to criticise or be a "party pooper" but would it not be easier to attach a 1920's 180 degree dial to any 180 degree rotatable cap. and just plot the capacitance with an accurate cap. meter at every 10 or 5 degrees and maybe every degree at the top end, for SLC caps.?

The shape of the plates (mathmatics) would be immaterial to the results and with the capacitance known, it should be an easy matter to match the capacitance curve in any set, by taking a couple of "in circuit" capacitance readings at specific frequencies. (ie. the same curve would just move up or down dependant on circuit capacitances - within a reasonable error margin.)

It would be a "simple" matter then to mark the set dial in frequency or wavelength, as desired, using standard calculations from the adjusted capacitance/frequency data set?

Hope I haven't missed the vital point in your post.

[KR1S]

[Dan McGillis]

Hi golf.

That's not "poopin" on the party - that's a good idea to try.

Someday I hope to have a good L/C meter. But I'll have to convince SWMBO that yet another hobby expenditure is absolutely vital to our very survival.
I'm kinda laying-low for awhile.

73, Dan

[OErjan]

Dan, while trying to build courage to ask permission you could likely get by with bridges.

http://www.allaboutcircuits.com/vol_2/chpt_12/5.html
http://theradioboard.com/rb/viewtopic.php?t=1546
http://theradioboard.com/rb/viewtopic.php?t=1352

quickie ASCII sketch of a usefull bridge below.