Call Search
     

New to Ham Radio?
My Profile

Community
Articles
Forums
News
Reviews
Friends Remembered
Strays
Survey Question

Operating
Contesting
DX Cluster Spots
Propagation

Resources
Calendar
Classifieds
Ham Exams
Ham Links
List Archives
News Articles
Product Reviews
QSL Managers

Site Info
eHam Help (FAQ)
Support the site
The eHam Team
Advertising Info
Vision Statement
About eHam.net

donate to eham
   Home   Help Search  
Pages: Prev 1 2 [3] 4 5 6 7 8 ... 18 Next   Go Down
  Print  
Author Topic: Designing & Building a High-Peformance Subminiature-Tube Regenerative Receiver  (Read 177107 times)
KB1WSY
Member

Posts: 1309




Ignore
« Reply #30 on: May 07, 2015, 01:53:21 PM »

I think you need a capacitor to ground between the junction of the tickler coil and the RF choke. Something like 100pF.

Come to think of it, I think that's where the variable "regeneration throttle" capacitor would go, in the older designs that use it. I was already planning to try this method of regeneration control.... I could wire the set with both the capacitor control and the potentiometer control and see which works better....

73 de Martin, KB1WSY
Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #31 on: May 07, 2015, 06:19:06 PM »

This evening I've spent a little while fiddling around with the current receiver, prior to attempting some of the coil and circuit changes mooted above. So, this is the version with (nominally) a 50pF/10µH tuned circuit, with the coil wound on a 1.25" diameter pill bottle.

In these evening hours plenty of signals are coming in on 40m, many of them heavily overloading the receiver -- this is not just the powerful SSB stations but lots of CW. The sensitivity is impressive so I have great hopes for this receiver once it's been improved!

Here's one new nugget. Until now I've been using a fixed 20pF series resistor on the antenna input (the schematic shows a variable capacitor, but actually I've been using fixed ones and playing around with the values). The capacitor is necessary because if you connect the antenna directly to the "hot" side of the tuned circuit, the damping is heavy enough to kill oscillation. It also helps control overload.

What I hadn't appreciated until now is the substantial "detuning" effect of this series capacitor. I had initially adjusted the value of the coil (adding or removing windings) with the 20pF capacitor on the antenna. This evening, when I changed that capacitor to 10pF, it seemed to be helpful in reducing overload (especially on SSB signals) but it also shifted the resonant frequency of the receiver upwards by at least 200kHz (out of the CW part of the band and into the top of the SSB area). Reducing the capacitor again, to 4.7pF, took the receiver out of the ham bands altogether and into SW broadcast.

So I suppose it's not surprising that, among other things, I've been having trouble using a "true" 10µH coil in that tuned circuit: the series capacitor is dramatically shifting the resonant frequency, so I've been having to adjust the inductance of the coil to compensate and bring the resonance back into the ham band.

The antenna is currently connected (through that capacitor) directly to the top of the tank coil. I have also tried winding an antenna coil above the tank coil, with the bottom connected to ground and the top connected to the antenna, but the set is almost dead when I try that.

If the antenna is connected directly to the detector, stability is very hard to achieve. Not only are there substantial "hand capacity" effects but almost any variable factor at the front of the detector messes things up in one way or another. So it will be very interesting to see what happens when that is replaced by the (relatively) stable load from the plate coil of a preceding RF stage.

Judging from the fact that when I wound an "antenna coil" above the detector coil, the signals were very weak, the coupling from the RF stage will have to be fairly tight and/or the gain from the RF stage will have to compensate. We'll see.

73 de Martin, KB1WSY
« Last Edit: May 07, 2015, 06:25:59 PM by KB1WSY » Logged
W1BR
Member

Posts: 4195




Ignore
« Reply #32 on: May 07, 2015, 08:51:15 PM »

The antenna coupling capacitor can also end up limiting your tuning range, if it is set too tightly, or it can kill regeneration as you noted.

You might try playing around with a link coupled antenna on the coil...

Pete
Logged
G3RZP
Member

Posts: 1315




Ignore
« Reply #33 on: May 08, 2015, 02:01:46 AM »

The 1927 ARRL design used a fixed 0.45pF. For a variable coupling capacitor, you want something no bigger than 2pF, with a minimum  of around 0.2pF. That usually means making one....
Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #34 on: May 08, 2015, 02:24:54 AM »

You might try playing around with a link coupled antenna on the coil...

As I mentioned earlier, I have already tried an antenna coil wound above the tank coil, with its bottom grounded -- I didn't know the term "link coupling" until today but that's what it was. It worked, but sensitivity was extremely low.

The 1927 ARRL design used a fixed 0.45pF. For a variable coupling capacitor, you want something no bigger than 2pF, with a minimum  of around 0.2pF. That usually means making one....

A few minutes ago I tried a gimmick capacitor (I simply twisted the incoming long-wire antenna onto another piece of wire connected to the antenna input; only a couple of turns). This worked, with no overload. I don't yet know how sensitive the set is, in that configuration, because it was so "detuned" that all I could hear was a couple of broadcast shortwave stations. Now, let's rewind the coil and bring the ham band back into tuning range ... and see what happens. With the gimmick capacitor, regeneration occurs rather earlier in the dial rotation.

Edited to add: When I got into this hobby three years ago, and started building radios, I remember thinking that I would be spending much of the rest of my life winding coils....

An exciting election night in the UK. I'm from a multigenerational British Labour Party family. My grandparents helped run a constituency party office in deep-blue Hampshire and I was an adolescent Labour canvasser there (a hopeless task!). I see that in their constituency, the UKIP candidate came second to the winning Tory, this time around. On the one hand, given my quasi-tribal political proclivities, I find the national result depressing. On the other hand, I am always inspired by the spectacle of democracy in its full glory -- despite the unsavory aspects and low opinion of politicians that RZP sometimes evinces, and with which I sometimes agree! I was particularly impressed by Ed Balls' not only congratulating his opponent, but telling her that she will make a good MP. There really aren't that many countries where power changes hands (or, in this case, doesn't change hands nationally) so gracefully.

(For hams outside the UK: yes, there is a famous British Labour Party politician whose name actually is Ed Balls.)

73 de Martin, KB1WSY
« Last Edit: May 08, 2015, 02:38:19 AM by KB1WSY » Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #35 on: May 08, 2015, 02:53:11 AM »

Success! I added another turn to the coil, bringing 40m into range. With the "gimmick" antenna capacitor, sensitivity seems decent (a bit hard to tell for sure but at about 6 a.m. local, I am hearing plenty of CW stations).

Now here's the really interesting part: hand capacity effects are sharply reduced. I can now actually tune in a signal, then take my hand away from the tuning dial, and the change in CW beat note is relatively small. With the old antenna capacitors (20pF or 10 pF) a signal could drift completely out of range when I moved my hand.

I am not totally sure why [reduced antenna capacitor = reduction in hand-capacity effect] and would appreciate enlightenment on the theory side but this is extremely encouraging.

If this change is as beneficial as it initially seems, that's great news. Such a simple modication, eh?

Edited to add: Upon further testing, I notice that regeneration is extremely smooth. At one end of the 50K pot, nothing. Rotate about one quarter of dial, good for AM/SW-BC reception. Rotate another quarter, threshhold of "noisy" oscillation, good CW reception. Another quarter, reduced sensitivity but good for strong SSB signals. Fully rotated, a loud squeal!

Edited again to add: I am now able to ground the set and the set continues to work (in the past, with the "old" 20pF antenna capacitor installed, grounding the set killed the regeneration). I also grounded the small alumin(i)um front panel (confined to the area around the tuning capacitor). When the set is grounded, the regeneration control is somewhat rougher and the set is noisier -- there is somewhat better reception without grounding. We'll work on that; it is mainly an issue of S/N ratio rather than sensitivity per se.

Another edit: the audio volume is considerably lower than it was before (mind you, it was quite deafening before). This is probably partly because of reception conditions in the early daylight hours here, but I suspect it's also because of the change in value of the antenna capacitor. May need to add another stage of audio eventually, depending on what happens when the RF stage is added.

And another: I just checked the tuning range (for this, I use a Sony SW-1 miniature short wave receiver with digital readout, and spot the regen receiver's oscillator). The width of the tuning range is now 165 kHz. According to my calculations, with the current values of the capacitor network in the detector stage (see schematic http://tinyurl.com/lva95cm), the width should be 230 kHz. Presumably the restricted tuning range is due in part to the presence of an antenna "capacitor" even if it's only a low-value "gimmick." A reminder: if you are looking at the schematic, the current build doesn't yet include the RF stage. The antenna is currently connected to the top of the detector tank coil, via the gimmick.

The 1927 ARRL design used a fixed 0.45pF. For a variable coupling capacitor, you want something no bigger than 2pF, with a minimum  of around 0.2pF. That usually means making one....

Couldn't you get the very low-C value by using a small-value variable in series with a fixed capacitor?

73 de Martin, KB1WSY
« Last Edit: May 08, 2015, 05:53:52 AM by KB1WSY » Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #36 on: May 08, 2015, 06:07:43 AM »

The 1927 ARRL design used a fixed 0.45pF. For a variable coupling capacitor, you want something no bigger than 2pF, with a minimum  of around 0.2pF. That usually means making one....

Well, if you confine yourself to fixed values, Mouser has plenty of choice:



You could mount a selection on a rotary switch but I presume the capacitance of the switch would overwhelm the capacitors themselves!!

A strong argument for adding that RF stage is that varying the value of the antenna coupling cap at the front of the RF stage will then, presumably, have no major effect on the range of the detector tuning circuit. I have this extraordinary ambition to have a calibrated main tuning dial one day!

73 de Martin, KB1WSY
« Last Edit: May 08, 2015, 06:13:10 AM by KB1WSY » Logged
KB1GMX
Member

Posts: 1824




Ignore
« Reply #37 on: May 08, 2015, 09:26:36 AM »

Need fractional to small value caps.... Coax, nominally 20pf/foot (you can find
the exact amount for what you have).  So a 1" piece is 20/12 or 1.6pf. 

The usual twisted wire "gimick" also works and if you have a capacitance
meter you can set it.  Generally for regens small coupling is better as it
reduces the antenna loading on the tuning and increases the Q of the tuned circuit.

If the circuit you published is where you are, more or less, the distance
from the plate coil of the first tube  (RF amp) to the second (tuned coil)
can be increased with the same improved Q and more stable tuning at
the expense of gain which likely with a RF amp ahead will not be a problem.

Hint if you can make the coil such that the RF coupling, main tuning,
and tickler are adjustable for position you may find better optimization.


Allison

Logged
KB1GMX
Member

Posts: 1824




Ignore
« Reply #38 on: May 08, 2015, 10:25:02 AM »

Part 2...

Screen bias regeneration control and throttle capacitor.   The answer is both.

The detector is grid leak.

http://www.angelfire.com/electronic/funwithtubes/Grid_Leak-1.html

Short form when the grid  has a positive signal there will be current flow (maybe very small)
and the capacitor in the regen circuit will acquire charge and in the negative part of the cycle
act as a bias.  Its the early approach to self biased.  Since a regen is either oscillating or not
that self bias helps to keep things sorta tamed down.  There is a magick point for the R and C
that is generally best determined by experiment as for a given tube and circuit there is wide
variation possible and a wide acceptable range of values.  Hint is the currents are small so
the resistor used as the discharge path should be large if that is the case then the C must be
small but there is a balance there.  If to large you get problems with controlling the electron
stream in the tube (they have finite leakage due to electrons that collide with the grid).  If the
C is too small you can't couple RF to the tube well enough.  the Why on that is the grid
at low frequencies is a capactitor (two plates separated by an insulator and vacuum is a
good insulator) and that means the grid input has a finite capacitance.  Its best if the coupling
cap C is about 10x more than the grid internal capacitance (from the data sheet 1AD4 about 4pf)
so a 50pf to 100pf cap is about right.  Since the resistor discharges that cap there is a time constant
and acceptable range for that.  Too high and you get audio tone, too low and there is signal distortion.
Start at 1M and 100pF as starting point.  Use a good low loss Ceramic (C0G) or dipped silver mica
for lowest loss.

Same site and different page:  http://www.funwithtubes.net/

Now, controlling a regen.  Three popular ways, control the gain of the device (screen bias),
the feed back path (tickler with throttle capacitor) and mechanical adjustment of feedback
(rotating or moving the tickler) relative to the grid coil.  First, they all work.  One of the great
engineering head banging exercises is optimizing a regen, to many variables, some you
can control.

Using pentodes, there is a relationship to screen voltage to plate(anode) voltage. 
That being, lower screen voltage lowers available gain and also lowers the strong
signal handling ability.  Increasing plate voltage increases signal handling (overload).
So for regen work a low screen voltage make the tube lower in gain and a bit tamer. 
Higher plate voltages allow bigger signal without distortion.

Throttle cap or mechanical feedback adjustment.  An amplifier becomes an oscillator 
when the ratio of gain to feedback exceeds 1.  The more gain in the amplifier the less
feedback needs to oscillate. A regen is an oscillator with a total gain (at signal frequency)
of just less than 1 (like .9999998).  At that point the signal is amplified greatly but there is
not enough total feedback to sustain oscillation.  So for CW the gain is more than 1,
for SSB more than 1, and for AM just barely 1.  We can control the feedback (assume
fixed gain devices) by reducing the means needed to achieve some measure of control
of the signal fed back.  A throttle cap is one way.  Moving the coil or turning it 90 degress
will reduce what is coupled back.

Now we put the two last paragraphs together.  With more screen voltage you get more
gain and returned signal but, only to a point then gain gets too high and regeneration
does not transition softly to oscillation but instead pops sharply.   Even the lowly battery
tubes have more than enough gain to oscillate vigorously so controlling feedback and
gain you can achieve a measure of control and stability.  So the answer is, use both,
and experiment.  Vacuum tubes are adjustable gain devices (either by bias in any
form or even lowering the heater current) and you have a coil for feedback plus
random wires that also contribute feedback.  So there are a few "knobs" you can
tune for best results.
 
So when you mix all of that plus mechanical layout optimization seems hard but experimentally
its very achievable.   The very best regens are mechanically strong with good RF layout
as well as operating point selection by bias and other means.


Allison
Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #39 on: May 08, 2015, 10:26:12 AM »

Need fractional to small value caps.... Coax, nominally 20pf/foot (you can find the exact amount for what you have).  So a 1" piece is 20/12 or 1.6pf. 

The usual twisted wire "gimick" also works and if you have a capacitance meter you can set it.  Generally for regens small coupling is better as it reduces the antenna loading on the tuning and increases the Q of the tuned circuit.

If the circuit you published is where you are, more or less, the distance from the plate coil of the first tube  (RF amp) to the second (tuned coil)
can be increased with the same improved Q and more stable tuning at the expense of gain which likely with a RF amp ahead will not be a problem.

Hint if you can make the coil such that the RF coupling, main tuning, and tickler are adjustable for position you may find better optimization.

Allison

Thanks! The RF stage is not built yet. First, I will be experimenting with a new form factor for the coil: reducing the diameter from 1.25" to 1" to try to get improved Q (because it will make the diameter/length ratio of the tank coil closer RZP's suggested 1:1).

The 1.25" diameter pill bottle has been easy to experiment with because I have been winding the coils and securing the ends directly onto the outside of the form with Scotch tape, then running the leads horizontally to the nearby terminal strips (I have not been drilling holes in the form or running the leads vertically down the inside). This makes it extremely easy to experiment with (1) number of turns and (2) spacing between them. It is sub-optimal in the sense that these horizontal leads (parallel to the turns in the coil itself) must be adding phantom inductance to the coil, but at the moment I'm just trying to get orders of magnitude for number of turns and spacing between them, before making better coils on better forms.

I ordered some 1" polycarbonate (Lexan) pipe, hoping to use the same "trick" for experimentation -- RZP point out that it's very lossy, but I was hoping to use 1" Millen 45005 forms for the final coils and he says they themselves are lossy (because mica-filled) in about the same order of magnitude as the Lexan. So perhaps the Lexan is appropriate for experimentation after all. But now I'm not so sure. The main problem with the Millen forms is that they only have about 1.5" useful height, which is a problem considering I need to wind three coils on them (tickler, tank, coupler from the RF amp). I'll need to use thinner magnet wire (so far, it's been 20AWG) to get all the coils to fit, and won't have much leeway to move the coils apart from each other.

I wonder how important the 1:1 D/L ratio is ... right now it's more like 1.5:1, on the 1.25" pill-bottle form. I have lots of Amphenol 1.25" polystyrene forms; once the basic coil parameters are "figured out" experimentally on the pill bottle, the final coil versions could be on an Amphenol form. Not only are these excellent forms, but using a plug-in coil would enable me to add more bands later (specifically, 20m). One way of getting the 1:1 ratio would be to space-wind the form, eh? Using a length of string or hookup wire for the spacing, and then removing it afterwards?

Yet another possibility would be to use lengths of Air Dux coil. I have a fairly large variety of these in all sorts of wire thicknesses, pitches, diameters and lengths.

Still pondering all this ... will probably experiment with various options. As I said earlier, it seems that building radios often boils down to: "winding coils."

73 de Martin, KB1WSY

Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #40 on: May 08, 2015, 10:38:53 AM »

Screen bias regeneration control and throttle capacitor.   The answer is both.

That's what I had pretty much already decided -- I will install both the screen voltage control, and the throttle capacitor. I'll pass on the mechanical/variometer approach for the time being.

Thank you for your detailed explanation of the workings of the grid-leak detector.

First, they all work.  One of the great engineering head banging exercises is optimizing a regen, too many variables, some you can control.

It is indeed mind-boggling. Ironically, it was once a rite of passage for most hams to build a regen set as a "beginner's radio" when in fact the workings of the circuit are very complex. You can build a one-tube radio that combines RF amplifier, oscillator, detector and AF amplifier in a single tube. The component count is low but I wonder how many novice builders really understood what was going on -- and how many of them managed to get the decent results that are possible with a carefully built and optimized regen.

Thank you also for your detailed explanation of the wonders of regeneration.

One day I'll find a book on the great legal battle between DeForest and Armstrong ... the technical arguments must have been quite interesting!

73 de Martin, KB1WSY
Logged
WB6BYU
Member

Posts: 18518




Ignore
« Reply #41 on: May 08, 2015, 12:18:28 PM »

Switching small capacitor values gets difficult because the switch wiring itself adds capacity.  A better
approach is to mount a small variable capacitor on insulated stand-offs with an insulated shaft to a knob
on the front panel.  Getting a cap where the stray capacitance across it may be the biggest problem -
a neutralizing capacitor (two parallel flat plates, one attached to a screw thread to adjust the spacing)
might do the job.
Logged
G3RZP
Member

Posts: 1315




Ignore
« Reply #42 on: May 08, 2015, 01:32:27 PM »

The L/D ratio is very non critical from about 0.75:1 to about 2.5 :1 and you won't notice the difference.

Consider the antenna as a low resistance in series with a capacitor if it's under 0.25 wavelengths long. Now at one frequency, a series RC circuit has an equivalent RC parallel circuit, so you can see that the antenna with the series capacitor looks like a shunt capacitor and resistor across the detector tuned circuit. The smaller the series capacitor the smaller the effective shunt capacitor and the less the detuning.

Coil former losses can generally be ignored in receivers unless Q is very important, but a lot of the pre-WW2 coil forms had ribs along their length so the coil winding was effectively spaced away from the form to improve Q......The 'kid's crystal set' of the 1920's with a cats whisker and lead pyrites crystal used the cardboard oatmeal drum to wind the coil on.
Logged
G3RZP
Member

Posts: 1315




Ignore
« Reply #43 on: May 08, 2015, 01:36:32 PM »

I omitted to mention that assuming Qs of 75 in the RF stage grid and plate (probably optimistic) any plate load over about 70kohm for the pentode should be stable. It's plate impedance is about 500kohm, so the damping effect of the plate impedance on the detector should be pretty negligible.

Alison, do you know who first established the stability criteria for tuned tube amplifiers? (H<2)
Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #44 on: May 08, 2015, 01:41:18 PM »

The great grid-leak experiment:

Keeping the 100pF grid-leak capacitor and varying the grid-leak resistance:

--470kΩ = narrow band of useful regeneration on the regen pot, very faint signals audible.

--1MΩ = similar to above, narrow regeneration range, slightly more gain on signals.

--2.2MΩ = much more vigorous oscillation, broader range of oscillation, more signals audible, somewhat "rougher" oscillation threshold (this is the resistor I've been using all along, up 'til now).

--4.7MΩ = even more vigorous oscillation, and yet more signals, additional roughening in the threshold.

--6.8MΩ = the most signals, the strongest oscillation, roughest threshold; I listened for audio distortion and didn't hear any, but the caveat is that there wasn't a lot of activity on the band here and no really strong signals that would invite overload/distortion.

Unfortunately reception conditions weren't great. There was a loud broadband cyclical whooshing sound that sounded man-made and located nearby; and it was the early afternoon, not a strong time for 40m propagation.

I've decided to leave the 6.8MΩ resistor in place. I rather suspect I'll end up going back to something closer to the original 2.2MΩ value, but it's worth experimenting.

It's tempting to put a potentiometer in that place ... presumably it ought to be close to the tube with short leads, and with a shaft extender to reach the front panel.

73 de Martin, KB1WSY
« Last Edit: May 08, 2015, 01:49:15 PM by KB1WSY » Logged
Pages: Prev 1 2 [3] 4 5 6 7 8 ... 18 Next   Go Up
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.11 | SMF © 2006-2009, Simple Machines LLC Valid XHTML 1.0! Valid CSS!