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 9 10 ... 18 Next   Go Down
  Print  
Author Topic: Designing & Building a High-Peformance Subminiature-Tube Regenerative Receiver  (Read 177100 times)
KB1WSY
Member

Posts: 1309




Ignore
« Reply #60 on: May 11, 2015, 06:13:29 PM »

NA4 (not asked) lower voltages are possible and some tubes work better than others down there.
Obviously you need more feedback and many users use a large inductance audio choke for the plate resistor in your version as it has less DC resistance and high impedance at audio. A typical choke might be several dozen Henries.  You may try a tube audio output from a AA5 (primary side is usually 5Kohms at audio), others have used ignition coils, pulls from old radios and small (physical) 220 to 6.3v transformers. I have a 50 henry choke from a seriously messed up BC221 and that works well.    In the old tube days this was called impedance coupling and it relied on the high AC impedance.

The detector choke-load experiment

If you've been following this thread (not necessarily recommended!) you'll know that I've been puzzled by the difficulty of getting these little 1AD4 tubes to oscillate at anything lower than the "nominal" B+ of 45V. The "old-timers" from nearly a century ago apparently recommended running a regenerative detector at low plate voltages, among other things because it is supposed to provide smoother regeneration control.

Tonight I replaced the 10KΩ detector plate resistor with the 10KΩ primary of a very small tube audio output transformer (Hammond model 144Q). This was at the instigation of KB1GMX and also because Charles Kitchin, N1TEV, uses just such a trick in his "junkbox special" regenerative receive -- his design is solid-state (http://www.arrl.org/files/file/Technology/tis/info/pdf/9811qex026.pdf).

It works! Until now I needed a B+ of at least 42V to get any oscillation going [this statement is wrong; see next post]. After replacing the resistive plate load with the aforementioned Hammond inductor, I started lowering the B+ voltage (by removing 12V batteries from the series chain):

--Original B+ voltage of 46V on the DMM: good reception using that little inductive load, in fact it seemed a bit better than with the resistive load, but not dramatically so.

--Removing one battery, the B+ falls to 34 volts on the DMM. Wow, I can get oscillation, in fact it's almost on the same spot on the screen-regen dial as it was before! Now, is the reception "better" than before? I'm not sure; perhaps. I think it is "calmer" but just as sensitive. A fair amount more listening is needed, to be sure.

--Removing two batteries, the B+ falls to 23 volts on the DMM. If I now advance the screen-regen dial almost all the way, I can get oscillation going, but it's "hair-trigger" and only fractionally below the "loud squeal" point. Reception: I can hear some stations, but overall, not great.

This is very interesting! Further experimentation needed!

Note: This was a crude experiment because I lowered the B+ not just on the detector, but on the audio stage too, simply by removing batteries. Next thing I'll try is putting a potentiometer in the B+ supply to the detector tube, but keeping the full 46-to-48V B+ on the audio stage.

It is also tempting to try an inductor with more "oomph." In other words, not this baby output transformer, but something bigger. Kitchin used the primary of a 6.3V heater transformer.

73 de Martin, KB1WSY
« Last Edit: May 11, 2015, 10:48:57 PM by KB1WSY » Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #61 on: May 11, 2015, 09:14:30 PM »

The detector choke-load experiment (continued)

Here are the various loads tested today. From left to right:
(1) A 10K resistor (the original load in this circuit).
(2) The primary of a Hammond 144Q tube output transformer, published impedance 10,000Ω, the DC resistance on my DMM is 620Ω.
(3) The primary of an old tube output transformer from the junkbox, DC resistance 390Ω.
(4) The 117VAC primary of an old power transformer from a tube radio, DC resistance 26Ω.



Contrary to what I said in my previous post, even with the 10K resistor load the tube *does* oscillate at the lowered B+ voltages (34V and 23V). What seems to have happened is that the addition of the screen "throttle capacitor" a few days ago improved not only control over oscillation, but also increased the tube's ability to oscillate in the first place. I had not noticed this at the time, because I didn't test it with reduced B+ at the time.

Anyway, to summarize: I believe that any of the three tested inductive loads are better than the 10K resistor in my tests. Regeneration control is a bit smoother and there is noticeably more gain. SSB signals in particular have very good audio quality and are easy to tune and very stable, which is no mean feat for such a simple set (I'm not particularly interested in SSB as a mode, but it's one of the useful receiver tests).

I wouldn't call the improvement a "slam dunk" -- time will tell. I think (2) the little Hammond transformer does about as well as (3) the much bulkier older output transformer. As for (4) the old power output transformer, it's not noticeably better or worse than the other inductors although there is perhaps a loss of gain.

The experiment is a success: an inductor is better than the resistor. I am leaving the little Hammond transformer in the circuit, replacing the 10K resistor.

So is the set's performance better if you have a lower B+ (for example 34V instead of 46V)? For sure, regeneration control is very smooth -- but the difference compared to normal B+ is fairly subtle. I think what I'll do is put a potentiometer in the B+ feed to the detector tube, allowing experimental variation of the B+ on that tube without reducing B+ to the other stages (the audio stage, and the future RF stage). I will mess around with various settings of that pot in the future and see whether this mantra about "lower B+ on the detector provides better performance" is borne out in practice.

73 de Martin, KB1WSY
« Last Edit: May 11, 2015, 10:50:49 PM by KB1WSY » Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #62 on: May 11, 2015, 10:44:50 PM »

Lowered B+ voltages, controlled with potentiometer

I added a 50K pot in the B+ supply to the detector tube. This allows the detector plate voltage to be set anywhere between 15V and 45V, while keeping the full 45V for the other stages in the set. To my surprise, the set still oscillates vigorously with only 15 volts on the detector plate, although achieving this does require setting the screen regeneration control near the top of its travel, and adjusting the throttle capacitor to get oscillation. Reception of signals is possible at that voltage. At a somewhat higher voltage (between 20V and 25V) the set performs pretty well; and if I'm not mistaken, the S/N ratio on certain strong signals is actually better at this low B+ compared to the 45V "default."

A couple of things:
(1) To humo(u)r the late, great Regeneration Old-Timers I think I may leave this detector B+ voltage pot in place. It will be fun over the next few months to compare "full B+" reception with their suggested "lower B+" setting. The number of knobs on this set is rapidly increasing....
(2) I am now optimistic that I could (later) build another version of this set, this time using the 2E32 subminiature pentodes rather than the 1AD4s used in the current set. I had originally started on a 2E32 project because that tube has modest power requirements: a 50mA filament current (half as much as the 1AD4) and a "normal" B+ of only 22.5V, but I could not get the tube to oscillate, come what may. I am now fairly sure that was because I had not included a capacitor in the "throttle capacitor" position. The 2E32 has more modest performance (transconductance of 500µMo , compared to 2000µMo for the 1AD4) but it would seem ideal for a portable set.



Time to go to bed.

73 de Martin, KB1WSY
« Last Edit: May 11, 2015, 11:08:37 PM by KB1WSY » Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #63 on: May 12, 2015, 08:16:50 AM »

The usual requirement for marine radios was that anything over 50 volts had to be accessible only by the use of a tool such as screwdriver.

Today I deliberately put the thumb and index finger of one hand across the B+ supply (nominal 48V). I couldn't feel anything. Then I pressed my fingers harder into the terminals and felt a very slight tingling. The 23A batteries I'm using now are very low-capacity (55mA/H). I'll soon be switching to a bank of 7.4V Lithium-Ion rechargeable batteries, with 7 in series giving about 52 volts. (Actually they are marketed as 9V batteries, but the nominal voltage is 7.4 for LiOn). Compared to the 23A's, these have 10 times more capacity: 600mA/H. I will repeat the experiment, with the new batteries, carefully and touching the terminals with the back of my fingers so that they bend away from the terminals if I get zapped!

I was wearing rubber flip-flops as it happens (=irrelevant since one of my fingers was grounded anyway!) and my hands were dry. I suspect that under a rather unlikely set of circumstances, 48V could be dangerous. In a marine setting however, with all that salt water around, I would definitely be taking precautions. (I come from a family of sailors.)

Edited to add: Actually what I'm more worried about is fire! With the really messy construction technique I'm using, a short across one of the three power supplies can't be ruled out. It already happened a few weeks ago, and melted a 1.5V AA plastic battery holder. I can imagine that 50VDC shorted to ground, with high-capacity batteries, could generate a lot of heat. The radio is built on a wooden breadboard and sitting on a wooden workbench. My only safeguard against fire is to make sure that I never leave the radio on when I leave the shack, so I have been getting a little paranoid about making sure that those little knife switches are in the "off" position before leaving the room.

73 de Martin, KB1WSY
« Last Edit: May 12, 2015, 08:40:09 AM by KB1WSY » Logged
WB6BYU
Member

Posts: 18516




Ignore
« Reply #64 on: May 12, 2015, 03:13:28 PM »

A fuse or circuit breaker on the battery packs would, of course, reduce the danger of a short circuit.

I like using the PTC Thermistors that are sold as current limiters / self-resetting fuses.  Something like this:

http://www.digikey.com/product-detail/en/0ZRC0025FF1E/507-1332-ND/1560187

for example will pass 250mA but if the current approaches 500mA will switch to a high resistance state
that limits the current to a safe value.

Yes, it looks funny to see what appears to be a disk ceramic capacitor in series with your battery supply...
Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #65 on: May 13, 2015, 04:58:56 AM »

Something is screwy here

Before proceeding with construction of the RF stage, I need to solve an issue that's been swept under the carpet until now.

The main pill-bottle tank coil has a diameter of 1.28" (including the diameter of the 20AWG wire), 14 turns, with a length of 0.7". If you use Wheeler's formula:

L = (Radius2 * Turns2) / ((Radius * 9) + Length * 10)

= 6.8µH

The trouble is that I calculated the LC circuit to use a 10µH inductor. I could understand if there were a small discrepancy, but this is a very large error. With the coil described above, I am bringing in a range of about 6980KHz to about 7150KHz. This just isn't right!

Could one of you good souls do me the favor of checking my calculations? The first one to check is the inductance calculation (above). The second one is to check whether I've made an error in setting up the capacitance network.



As you can see from the schematic excerpt, there are four capacitors involved, in a parallel and series combination. The fixed capacitors are silver mica. The variable is a Hammarlund MC50S, straight out of its original box, with the part number stamped on the porcelain. It has a meshed capacitance of 50pF and I am assuming an unmeshed capacitance of 5pF.

Based on those assumptions, I get the following values for the capacitor network:

--Maximum 51.7pF.
--Minimum 48.5pF.

For parallel capacitors I'm using C = C1 + C2. For two series capacitors I'm using (C1 * C2) / (C1 + C2).

The final calculation is LC equation for frequency. With C = 51.7pF and L = 10µH, I get frequency = 7000KHz. If I stick in the actual putative value for L in my current build (14 turns or 6.8µH) I get f = 8488KHz (!!!!!).

Formula used:

f = 106 / (2π * √LC) where units used are KHz, µH, and pF.

(I have done all these calculations on paper, and also checked them with online calculators.)

If my calculations are indeed correct, my proposed troubleshooting checklist is in this order:

(1) Double-check the wiring (again).

(2) Check the solder joints for continuity (again).

(3) Replace the pill bottle with an Amphenol 1.25" form, just in case there's something screwy about the plastic material.

(4a) Edited to add: remove all of the fixed capacitors. Connect the variable capacitor across the coil and mesh it fully to 50pF. Check the reception frequency with my signal generator (sig-gen frequency double-checked on another receiver). Tuning will be non-linear but it's just a test.

(4b) Replace the mica capacitors; I suppose it's possible one of them is way off-value (I don't own a capacitance meter, and my grid-dipper only provides approximate values when used as a capacitance reckoner).

(5) Edited to add: physically rebuild the radio so that the tuned circuit is separate from other components. For instance, in the current build, some of the capacitors in the LC circuit are very close to the plate choke and the plate bypass capacitor (and the LC circuit shares a grounding lug with the bypass cap).

(6) Any other suggestions, if none of the above works?

BTW the antenna is currently connected to the "hot" end of the tuned circuit, via a gimmick capacitor (two wires twisted together, two turns).

Many thanks!

73 de Martin, KB1WSY
« Last Edit: May 13, 2015, 08:16:17 AM by KB1WSY » Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #66 on: May 13, 2015, 03:12:37 PM »

So I've now done all of these things:

--Double-checked the wiring.

--Checked the solder joints for continuity.

--Replaced the pill bottle with an Amphenol 1.25" form, just in case there's something screwy about the plastic material in the pill bottle. Made a new coil on the Amphenol form, with 18 turns and 0.75" length (which should yield 10µH).

--Removed all of the fixed capacitors. Connected the variable capacitor across the coil and meshed it fully to 50pF. So now we have a bog-standard LC circuit with a single inductor and a single capacitor.

Checked the reception frequency with my signal generator (sig-gen frequency double-checked on a digital receiver). Result: the set is now tuning a range that begins at 5.9MHz (when it should be around 7MHz). So there is nothing wrong with my capacitance calculations apparently, because substituting a known 50pF capacitor (replacing the capacitance network) led to a similarly screwy result.

I also removed the "gimmick" capacitor from the antenna line, and it made no appreciable difference to the frequency.

I am really, really puzzled.

I am convinced that my coil must be roughly 10µH (and I have ordered a fancy B&K LCR meter so that I can check this for sure tomorrow!). I am shunting it with 50pF. So why is the set tuning at about 5.9MHz? What is the dynamic inside that regenerative set that can shift the frequency down by more than 1MHz?

The upside is that I am enjoying all sorts of exotic shortwave broadcast stations!

To be continued....

73 de Martin, KB1WSY
« Last Edit: May 13, 2015, 03:15:42 PM by KB1WSY » Logged
WB6BYU
Member

Posts: 18516




Ignore
« Reply #67 on: May 13, 2015, 03:17:55 PM »

You might try the Hamwaves inductance calculator, which is supposed to be much more accurate than
Wheeler's formula:

http://hamwaves.com/antennas/inductance.html
Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #68 on: May 13, 2015, 03:36:24 PM »

You might try the Hamwaves inductance calculator, which is supposed to be much more accurate than
Wheeler's formula:

http://hamwaves.com/antennas/inductance.html

I tried it, but "an error occurred when solving for beta" (whatever that means).

I was already aware of the hamwaves calculator. Fact is though, my current experimental results are way off from what any formula yields, even if Wheeler's is only an approximation.

I have wound quite a lot of coils in the past three years and never had this kind of error. So my thinking is that the inductance calculation is (approximately) correct, but there is some stray capacitance somewhere in the circuit that is lowering the resonant frequency. The hunt for that capacitance is ongoing.... To get 5.9MHz at 10µH would imply a capacitance of 73 pF. The coil is currently shunted with 50pF, so somewhere in my radio's LC tank circuit, there is an extra 23pF (approximately). At least that's what I think.

Thank you for your suggestion though!

73 de Martin, KB1WSY
Logged
WB6BYU
Member

Posts: 18516




Ignore
« Reply #69 on: May 13, 2015, 04:54:12 PM »

All coils have some self-capacitance, and this effect is worse when the turns are close-wound
because the capacitance between adjacent turns increases (especially as the air dielectric between
the turns is replaced with shellac or some other form of insulation that has a different dielectric
constant.)  Typical recommendation is to maintain about 1 wire diameter spacing between turns
(typically by winding the wire with a second one, or a piece of string, as a spacer and then
unwinding it when the wire is secured.)

You also have to include the input capacitance of the tube - should be given in the data sheets
as the capacitance from grid to cathode, but also to the other grids.  This can be increased due
to the "Miller effect", though I'll admit that I don't understand it or where it applies.
Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #70 on: May 13, 2015, 05:12:08 PM »

On a totally different subject, I think I am beginning to understand better the old timers' advice to run the regenerative detector tube with a low B+. Although I have seen various versions of this advice, it often seems to pertain particularly to *AM* reception.

This evening I was listening to "Radio Alexandria" on 7490 kHz as a technical test. They finished their broadcast with an extended section of Beethoven's "Eroica" symphony. (I had never heard this station before, but for what it's worth, it seems to be characterized by a combination of "prepper," "environmentalist" and somewhat religious themes.)

I set the "regeneration throttle" capacitor to somewhat below the oscillation level -- as is recommended for AM -- but it still sounded distorted (in real life I am in the music business). Then I lowered the detector-tube B+ down to a very low 15V (as compared to a nominal 46V) and boosted the regeneration a bit (using the throttle capacitor) and the music fidelity improved dramatically.

So, if only because I enjoy listening to short-wave broadcasts, I think it's important to keep some kind of control over the B+ on the detector tube.

Whether this has anything to do with improving overall "communications receiver" performance (as opposed to lessening harmonic distortion on AM, which is not of much importance to most hams) I am not sure.

Edited to add: later this evening I heard a Rachmaninov piano concerto on 7320kHz. Extremely distorted at the full B+. Once again, when I reduced the detector tube's B+ all the way down to 15V and re-adjusted the regeneration throttle capacitor, the music fidelity dramatically improved. Alright, it's not "hi-fi" but the difference was substantial.

73 de Martin, KB1WSY
« Last Edit: May 13, 2015, 05:55:28 PM by KB1WSY » Logged
KB1GMX
Member

Posts: 1824




Ignore
« Reply #71 on: May 13, 2015, 07:16:21 PM »

Miller effect is (in terminally rough form) grid to plate capacitance multiplied by the gain of the tube.
For multi-grid tube that is a small number as the purpose of the screen grid is to isolate the control
grid from the plate.  In most tube, transistor and especially FET (mos or not) it represents a variable
capacitance with the problem of undesired negative feedback.

For those tubes the input C is roughly 4pf mostly structural.  Miller C is very small for the most part.

The Wheeler equation is a bit fussy as it does not seem to work well with other coils in proximity
(mutual inductance).   Also the error tends to be greater if there is stray capacitance (leads and
other).  Add to that your capacitors are likely 5 or 10% tolerance units.  So error will happen.

if you get "beta" error then he coil length for the diameter of the coils is impossible example
a coil of .1" diameter wire and 10 turns cannot be a single layer .5" long.   Also possible if the
outside diameter is not larger than the form plus twice the wire diameter.  Basically two solid
objects can't logically be in the same space.  That and division by zero.

The AADE LCII is a good LL-C meter for measuring caps and coils.  When I've had problems
like that measuring the parts usually found something was not as expected.  Typical is marked
caps were not even close. 

Grid dipper can be accurate if the reference coil is known value and the actual frequency is
checked with a receiver. 

If the tickler winding is wound over the grid winding then you have more capacitance.
If the RF stage has a coil to couple to the grid winding that adds physical capacitance.

You can figure out that missing capacitance (assume for the moment the coil is as calulated)
by working backward from where you are to where you should be to get the excess added
capacitance.  Hint if the total capacitance is about 51pf than an additional 10 pf (tube input
plus other coils) would move you down about the right amount.

Do not forget the random wire running all over the breadboard.

RE: regenration and old timers.  You run regeneration ONLY for AM.  For CW and SSB
you require oscilation making the radio an autodyne (direct conversion) receiver.  Two
different critters with different operating criteria.

Now the distortion could be plain overload, enough gain and too much signal.  The other is
enough regeneration and you get sidebands being deleted from oddly, too much selectivity.
Add to that a regen is NOT a linear detector.

Increasing the feedback winding (turns) and spacing it further away can make for running
with less screen grid at low voltages.  FYI the feedback winding (tickler) is also a possible
load on the grid coil lowering selectivity and gain.  Counter intuitive but its coupled to the grid
and has the tube plate as a load.

Like I said before, its a great engineering exercise to try and optimize a regen as you have
many controls and just as many things that do whatever they want.  Result, banging head
against wall.  Lots can be learned.


Allison
Logged
VE3LYX
Member

Posts: 814




Ignore
« Reply #72 on: May 13, 2015, 07:52:20 PM »

A simple radio project seems to be getting awfully complicated.
Build an oscillator. Under power it so it just below oscillation point and install a set of earphones in series with the B+ or B-. You will have a working regen. Raise the voltage and remove the phones, sub in a key and you have a nice CW transmitter. It really is that simple. After that you can add extras you want like an audio AMP, a filter etc. A RF amp to a regen is totally unnecesary. Nice to talk about but of no real purpose. A regen when done right will out hear any radio you currently own. It is very easy to overload them antenna wise and an Rf amp just adds to that problem. I am very partial to the twinplex twin triode design but have currently in shack 8 regens in service. Most fun was the POW regen built from garbage dump stuff. Also have some solid state ones. Clough's patent (1927)using an absorption loop to control regen is another very interesting design. Biggest problem with most is regen pot failure. Throttle cap method works but is not as nice as a twinlex with pot control. 
Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #73 on: May 14, 2015, 05:11:37 AM »

if you get "beta" error then he coil length for the diameter of the coils is impossible example a coil of .1" diameter wire and 10 turns cannot be a single layer .5" long.   Also possible if the outside diameter is not larger than the form plus twice the wire diameter.  Basically two solid objects can't logically be in the same space.  That and division by zero.

Here's what I entered into hamwaves (http://hamwaves.com/antennas/inductance.html):
Diameter 32.5mm
Length 19mm
18 turns
Wire diameter 0.81mm

I'm always getting the "beta error." But I don't see anything wrong with the values entered.

When I've had problems like that measuring the parts usually found something was not as expected. Typical is marked caps were not even close.  

When my new LCR meter arrives today I will check everything. Coil on its own, coil in circuit, with tickler mounted, without tickler mounted. Capacitance in various connected/disconnected scenarios.

Thankyou for listing the numerous "culprits" for extra inductance and capacitance! It will be interesting to track them down.

Ultimately I'm just interested in knowing what's going on; the set is working and that's what counts.

A simple radio project seems to be getting awfully complicated.  

That's why I started this new thread about building a "high performance" regenerative radio. The intention is to use all possible techniques toward that goal, even if the resulting set is more complex than a typical regen. It is getting far away from the other thread's theme of a "portable hiking radio" and that is done on purpose. I have learned an enormous amount of radio and electronics in the process.

A RF amp to a regen is totally unnecesary. Nice to talk about but of no real purpose.

I have heard many opinions on that. I don't expect it to improve performance in any obvious way apart from lowering susceptibility to antenna movement in the wind and providing a steady load to the detector. Plus, elimination of stray radiation from the oscillator.

Edited to add: I am also interested in finding out whether an "active" RF gain control (controlling the grid bias) will outperform "passive" gain limitation in handling strong signals.

The only way I will find out for sure, though, is to build it! There's nothing like hands-on experience.

73 de Martin, KB1WSY
« Last Edit: May 14, 2015, 12:23:59 PM by KB1WSY » Logged
KB1GMX
Member

Posts: 1824




Ignore
« Reply #74 on: May 14, 2015, 05:57:27 PM »

Here's what I entered into hamwaves (http://hamwaves.com/antennas/inductance.html):
Diameter 32.5mm
Length 19mm
18 turns
Wire diameter 0.81mm

I'm always getting the "beta error." But I don't see anything wrong with the values entered.<<


I had to enter a frequency to make it happy and I used 7mhz.  Result 9.6uh give or take.

If you out in the boonies the RF amp is optional.  If you ahve people that you want to annoy with
wideband regen his noise its optional, If your using it for CW (oscillating) and want to annoy
people its optional.  Regens radiate RF.  The RF amp is not there for gain (never hurts) but
to isolate the detector from the antenna.

That and the RFamp can be used as frequency converter later!

Allison

Logged
Pages: Prev 1 2 3 4 [5] 6 7 8 9 10 ... 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!