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: [1] 2 3 4 5 6 ... 18 Next   Go Down
  Print  
Author Topic: Designing & Building a High-Peformance Subminiature-Tube Regenerative Receiver  (Read 177210 times)
KB1WSY
Member

Posts: 1309




Ignore
« on: May 05, 2015, 05:06:10 AM »

This thread is split out from my separate "Tube Portable Transceiver" thread. My current challenge: designing and building a subminiature-tube regenerative receiver with good sensitivity and stability, plus the best selectivity that can be achieved with a regenerative design.

In his article "High Performance Regenerative Receiver Design," Charles Kitchin, N1TEV, argues that a properly designed regenerative receiver is "quite capable of direct-conversion or superheterodyne-level performance, although it does require greater operator skill." The article was published in the Nov/Dec 1998 issue of QEX
and is available on the ARRL website: http://www.arrl.org/files/file/Technology/tis/info/pdf/9811qex026.pdf. (I thank eham member JAHAM2BE for bringing it to my attention a couple of years ago.)

Kitchin reminds us that the regen was the standard design among hams of the 1920s and early 1930s. He argues that today's viewpoint of regens as "only suitable for beginner experimentation" is erroneous. This attitude, he says, arose in part because regen circuits designed for novices in the 1940s and '50s had poor performance. Much of the regen-design knowledge possessed by the "old-timers" of 100 years ago had been forgotten.

Here are some design elements of a high-performance regen (these points are a mixture of Kitchin's, and mine). Note that Kitchin's published designs are solid-state; so part of the challenge is to "translate" them to a tube context.

--A broadband (untuned) RF stage preceding the detector; this prevents radiation by isolating the oscillator from the antenna; reduces loading on the detector; provides a constant load for the detector; and reduces instability resulting from an antenna that swings in the wind. The gain of such a stage does not have to be substantial and can even be negative, because of the high sensitivity of the following detector stage. Kitchin argues against using a tuned, selective RF stage although some other authors (Lindsay Publications, "The Impoverished Radio Experimenter") do advocate a conventional tuned stage (with its own separate tuned circuit, either ganged or un-ganged with the detector tank circuit).

--Some kind of variable input attenuation, to control overload, which is a common problem with regens. This could be a resistive or capacitive control in the antenna circuit; or some kind of control on the gain of the RF stage.

--A "throttle" variable capacitor to control regeneration, eliminating the "hysteresis effect" ("overshooting") and the beat-note drift of the nowadays more-common potentiometer or the detuning effect of a variometer.

--Voltage regulation to hold the detector voltage constant; Kitchin's solid-state designs use a Zener diode for this.

--Something that is confusing me quite a lot. Both Kitchin and some other sources argue that the gain of the detector should be kept low (with a JFET, by keeping high negative bias) because this permits "very smooth regeneration control." Kitchin says that "the old-timers of the 1920s knew this and operated their tube detectors from low supply voltages" to achieve this. However it is not clear to me whether this is necessary only for good reception of AM signals; or whether it is also a good idea if (as is my case) you are only interested in CW. In "The Impoverished Radio Experimenter" it is recommended to run the detector at such low gain that it is operating in the non-linear part of its curve; does this mean that detection is, in effect, a form of harmonic distortion??

--Use of a non-metallic cabinet, apart from an (optional) grounded metal front panel to reduce hand-capacitance effects. "A metal chassis, shield cans etc. all absorb energy from the main tuning coil and add to its losses, which directly affect the overall circuit Q and the selectivity of the receiver" (Kitchin).

My challenge is to replicate the above design elements using small, low-current subminiature tubes. My "baseline" is the 1AD4, a sharp-cutoff RF pentode with a directly heated 1.25V cathode/filament and a nominal B+ of 45 volts. I've already successfully built (with a lot of help from eham Elmers!) a conventional "beginner's regen" using two such tubes. It has none of the above-mentioned "high performance" design elements and furthermore I took little care in construction, wanting instead to do a rapid "proof of concept." The receiver does work. It is very sensitive; however it is also very unstable, easily overloaded and extremely sensitive to hand capacity effects or indeed the presence of any objects in its vicinity. It is hard to get optimal loading; and grounding the set shuts off regeneration altogether. These are the typical "issues" that cause today's builders to shun regenerative sets.

My first question is: With the tube constraints that I have (directly heated filament), what sort of circuit would function well as a broadband RF amplifier in front of the detector? Kitchin uses a grounded-base transistor. What would be the hollow-state equivalent: Can I use a conventional RF amplifier circuit and if so what does the input circuit in front of the grid look like? Or do I need to figure out a grounded-grid or a (degenerative) cathode-follower design in order to get wide bandwidth? If so, how are those designs possible with a directly heated tube? Alternatively, should I use a tuned RF amplifier, as advocated by "The Impoverished Radio Experimenter" even though it's a PITA to introduce tracking between the RF stage and the detector?

73 de Martin, KB1WSY (with apologies for the typo in the word "Performance" in the topic title; once posted, it cannot be fixed....)
« Last Edit: June 02, 2015, 03:10:58 PM by KB1WSY » Logged
G3RZP
Member

Posts: 1321




Ignore
« Reply #1 on: May 05, 2015, 07:35:09 AM »

The best regenerative detector I had used a 6AU6. The control of regeneration was by varying the screen voltage and it was SMOOTH! Something like an RF choke in the plate, then a 1000pf to ground and a 47k plate load to an 8 mfD electrolytic and then via 1k to about 250 volts of B+. From the plate, a 100pF fed the tickler coil. I used a 6BA6 RF amplifier with a tuned grid and a pot - probably about 5 k - in the cathode for an RF gain.

Now if we look at the US Navy RAL design from 1936, it has two RF stages using 6D6s, a 6D6 detector, a 41 output and a 41 AGC detector: the regeneration was by controlling the screen and suppressor of the detector and the tickler was in the cathode of the detector. It covered 300kHz to 23 MHz in 9 bands, using 3 coils per stage, each one of which had a tuned winding with two taps giving three possible inductances.

The 2 RF stages was probably to give good isolation between detector and antenna: the receiver is not so much built like a battleship as needing a battleship to carry it! It is 13-5/16 inches high, 18 inches wide and 16-3/32 deep and weighs 69 pounds. The power supply is 12-1/4 by 14 by 8 and weighs 41lbs...But they have a reputation as probably the best TRF made. Its sister RAK covered 15kHz to 300kHz in 6 bands

Now as you have been talking of only covering 7 MHz, there's no need to worry about tracking an RF amp - a single tuned circuit in the RF amp grid will suffice. I would still go for a metal box from the viewpoint of hand capacity, but space the coil at least two diameters away from the metal. By varying the bias on the tube (you will need a C battery), you can get variable RF gain to avoid overload. Bear in mind that with the very low voltage supply, overload will happen much more easily than with high voltage tubes.

Try looking in the pre WW2 QSTs for TRF designs.

Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #2 on: May 05, 2015, 08:27:21 AM »

The 2 RF stages was probably to give good isolation between detector and antenna: the receiver is not so much built like a battleship as needing a battleship to carry it!

I see what you mean! The RAL manual is here: http://bama.edebris.com/manuals/rca/ral/.

I will trawl through the pre-war QSTs.

Now as you have been talking of only covering 7 MHz, there's no need to worry about tracking an RF amp - a single tuned circuit in the RF amp grid will suffice.

So presumably, at the front of the RF amp, an antenna coil with its secondary forming the L part of a fixed LC tuned circuit, peaked for whatever favorite part of the band? Maybe I'll try building one experimentally with a 1AD4 and putting it in front of my existing, sub-optimal "breadboard" set as yet another "proof of concept." Then I'll come up with a full schematic for the "high performance" set and we can look over it -- this time, "design before building" (until now, it's been the other way round!!).

So now we're working on the RF amp, I have two questions on other stuff:

--Concerning the AF stage. Am I right that if I put in a two-tube AF stage (triode, then pentode) with relatively low gain on the triode, it will load the detector output less and help with stability? (Perhaps just a resistive load on the triode rather than a choke.)

--Does it make sense to add voltage regulation, and if so how? With the relatively low B+ none of the usual VR tubes is appropriate.

73 de Martin, KB1WSY
Logged
G3RZP
Member

Posts: 1321




Ignore
« Reply #3 on: May 05, 2015, 11:11:46 AM »

Quote
--Concerning the AF stage. Am I right that if I put in a two-tube AF stage (triode, then pentode) with relatively low gain on the triode, it will load the detector output less and help with stability? (Perhaps just a resistive load on the triode rather than a choke.)

The choke presents a high impedance load for the detector with a low resistance. Going to resistance coupling means losing plate volts and load resistance, Getting the right value of resistor is then a compromise. With two AF stages, you get around the loss of signal by going to a resistor: if you used a pentode stage, you could have a lower plate load on the detector and thus more volts. But you might get the noise of the AF stage becoming a problem. Probably not, but worth noting the possibility.


Quote
--Does it make sense to add voltage regulation, and if so how? With the relatively low B+ none of the usual VR tubes is appropriate.

No, but at  least 2mFd and preferably more across the B+ supply. If you use an electrolytic, make sure it is on the rx side of the B+ ON/OFF switch.

I would suggest you go on both RF and detector for about 10 microhenries and around 50pF to tune it. The higher L circuit gives a higher impedance.
Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #4 on: May 05, 2015, 11:59:54 AM »

No, but at  least 2mFd and preferably more across the B+ supply. If you use an electrolytic, make sure it is on the rx side of the B+ ON/OFF switch.

OK. I have just added a 10µF/100V electrolytic across the 48V B+ of my ramshackle breadboard regen test-bed. There is no audible difference in stability, but the causes of instability in this crude set probably don't have much to do with the purity of the B+ supply. However, the new capacitor does seem to have slightly improved the smoothness of the regeneration control -- but it was already rather smooth anyway. With the new capacitor, the onset of regeneration appears slightly farther round the dial and with a slightly more gradual transition.

I would suggest you go on both RF and detector for about 10 microhenries and around 50pF to tune it. The higher L circuit gives a higher impedance.

I will start by building a "breadboard" RF amplifier and connect it to the existing breadboard set, with the LC setup as you suggest. When I build the "high-performance" set I will probably play around with LC ratios to see what works best.

I would still go for a metal box from the viewpoint of hand capacity, but space the coil at least two diameters away from the metal.

I know from my experience with building the ARRL "3-Transistor Receiver for the Beginner" that you can build a good regen, with high sensitivity, even if it is completely enclosed in a metal cabinet. In fact my initial build of that project was just an "open" metal chassis/panel combination and I ended up building a metal cabinet to deal (successfully) with hum issues. Adding the cabinet didn't appreciably damage sensitivity. That set has hand-capacity issues, but they are very mild compared to my "open breadboard" tube experiment.

With the current "high-performance" project I will first attempt to do things the "N1TEV way" i.e. with a largely wooden chassis, with metal only on the front panel; but will do it in such a way that metal shielding/groundplane can be added experimentally if there are problems with hand capacity or hum. Kitchin's point is that the knowledge of The Ancient Ones concerning building a good regen is worth paying attention to, including using a breadboard-style design rather than metal; so I will give it a go, and change course if it doesn't work.

Edited to add: one possibility would be to have a metal groundplane for most of the set, except for the detector stage and the area around the detector coil. I do want this set to "play nicely" with the rest of my station, including being connected to the ground system and not drifting all over the place when other equipment is nearby/active.

73 de Martin, KB1WSY
« Last Edit: May 05, 2015, 12:44:19 PM by KB1WSY » Logged
G3RZP
Member

Posts: 1321




Ignore
« Reply #5 on: May 05, 2015, 05:11:29 PM »

If you look back to the TRF sets in the 1938 RSGB Radio Amateur's Handbook, you can see a move to metal cabinets and chassis....Same as the admittedly,  pretty crappy, RAF aircraft TRF sets of the pre WW2 period.
Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #6 on: May 05, 2015, 05:52:51 PM »

Now as you have been talking of only covering 7 MHz, there's no need to worry about tracking an RF amp - a single tuned circuit in the RF amp grid will suffice. I would still go for a metal box from the viewpoint of hand capacity, but space the coil at least two diameters away from the metal. By varying the bias on the tube (you will need a C battery), you can get variable RF gain to avoid overload. Bear in mind that with the very low voltage supply, overload will happen much more easily than with high voltage tubes.

Why the C battery? No way to get self-bias? How come the audio stage works fine without battery bias? Hmmph, all those batteries....

On another tack, how about a cascode RF amp? They are all the rage in the 1968 RSGB manual, even at HF....

Edited to add: This evening I did a head-to-head comparison between my two regenerative receivers, both on 40m, both with the same antenna, swapping betwen the two. The two-tube 1AD4 "messy breadboard" receiver is hands-down better in sensitivity and is bringing in noticeably more signals, however the ARRL "3-Transistor Receiver for the Beginner" (1968 design) is dramatically more stable and usable at this point. A fun comparison.

73 de Martin, KB1WSY
« Last Edit: May 05, 2015, 06:06:10 PM by KB1WSY » Logged
G3RZP
Member

Posts: 1321




Ignore
« Reply #7 on: May 06, 2015, 01:54:24 AM »

The cascode needs more volts if you use the simpler series circuit: other wise, it gets more complex. You don't NEED an RF stage at 7 MHz from a noise factor viewpoint, but having an extra tuned circuit helps reject those nasty strong broadcast signals above 7.2 MHz. I would go for about 10 microhenries for both detector and RF stage: the higher L/C ratio gives a higher impedance and more step up.

No bias means very low signal handling capability: getting gain control by varying the screen volts also gives signal handling problems. So get the bias by connecting the negative side of the B battery through a resistor to ground: choose it for about 2 to 3 volts drop across it, and now you have a negative bias supply. As done in the BC348 and R1155 receivers.

You may well find less AF distortion with some bias....depends on how big the AF signal is.
Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #8 on: May 06, 2015, 03:12:48 AM »

No bias means very low signal handling capability: getting gain control by varying the screen volts also gives signal handling problems. So get the bias by connecting the negative side of the B battery through a resistor to ground: choose it for about 2 to 3 volts drop across it, and now you have a negative bias supply. As done in the BC348 and R1155 receivers.

You may well find less AF distortion with some bias....depends on how big the AF signal is.

Reading my post yesterday, it looks a bit dyspeptic! I can either use your suggested biasing method, or your original suggestion of a C battery. I'm already using a rechargeable 1.2V AA NiMh battery for the filaments, and can easily add a new, separate "C supply" battery holder using two 1.2V batteries in series -- I recharge those NiMh cells four at a time anyway.

"Leafing through" the digital QST archive from the late 1920s, what strikes me is the sophistication of the regenerative designs: complicated coil combinations, plethoras of batteries, and great attention paid to filtering and bypassing in the power supply lines.

I also found a quote from the November 1929 edition of the Radio Amateur's Handbook:

And now, when the receiver has been built, adjusted and placed in satisfactory working condition it will be permissible to sit back and take a long breath. For the receiver is one of the two essential parts of an amateur station. If the receiver has been correctly built and if the location of the station is satisfactory it will receive as far as any transmitter can send. If it has open tuning scales; if it has lots of sensitivity and amplification; and if it smooth and quiet in operation, it will be a very great comfort and a source of splendid pleasure.

73 de Martin, KB1WSY
« Last Edit: May 06, 2015, 03:14:53 AM by KB1WSY » Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #9 on: May 06, 2015, 05:05:25 AM »

Here's my first try at a circuit for the RF amplifier. This time, I'm trying the novel idea of designing first, then building....



I'll wait for comments, then build a rough breadboard version as a first test.

One question: instead of putting the "C" battery in series with the gain pot, would it make more sense to connect the bottom end of the pot directly to ground and put the negative battery connection on the wiper? Edited to add: in which case I'd need to add a fixed resistor in series, to avoid shorting the battery at one end of the pot travel....

Edited again to add: today I played with the B+ voltage, in an effort to test the assertion (made by Kitchin and others) that The Ancient Ones deliberately dialed back the gain on the detector stage to permit smooth regeneration control. It didn't work: lowering the B+ from 48V to 36V only resulted in an almost dead receiver. It was possible to achieve oscillation at the extreme end of the pot range, but sensitivity was way, way down and I did not notice any greater smoothness. This doesn't mean the idea of operating at low gain is necessarily wrong, however it doesn't seem to be working in the case of my current tube/circuit combination. (It should also be noted that regeneration control with my current circuit is already quite smooth anyway.)

73 de Martin, KB1WSY
« Last Edit: May 06, 2015, 05:18:31 AM by KB1WSY » Logged
JS6TMW
Member

Posts: 1257




Ignore
« Reply #10 on: May 06, 2015, 05:23:15 AM »

I don't think the directly-heated cathode is a drawback, maybe a necessity in a small volume. Wow, but those subminiature tubes were really magic back in the day - not very popular among hams but they were a "missing link" between VTs and transistors and widely used in hearing aids.
Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #11 on: May 06, 2015, 05:34:08 AM »

I don't think the directly-heated cathode is a drawback, maybe a necessity in a small volume. Wow, but those subminiature tubes were really magic back in the day - not very popular among hams but they were a "missing link" between VTs and transistors and widely used in hearing aids.

Yes indeed. For those who may be unfamiliar with these cute tubes, here's a picture of the 1AD4s that I'm using in this set (I bought 30 of them for $29):



Here's my Rat's Nest Experimental Regenerative Radio. I used 1AD4s that haven't had their leads cut for sockets, and the other components also have long leads to make them easier to re-use later. This kind of wild construction is a recipe for instability and leakage ... I'm looking forward to settling on a future "high performance" design and building it properly!



I'm not sure why hams didn't seem to use subminiature tubes much -- even the numerous QST designs for portable gear passed them over and used larger standard battery-type tubes. I have two three theories: (1) I suspect that they were quite expensive, on a "bang for your buck" basis. (2) Logically it probably makes more sense to use PCB construction for a finished project, given the tubes' small size, and those construction techniques were uncommon in the late '40s and early '50s. (3) Those were the days when more power was often regarded as an unalloyed "good thing" and the subminiature tubes seem more in harmony with today's QRP trend.

I like them a lot because they are so easy to experiment with -- similar to the ease with which one can throw together a solid-state circuit. There's no need for an AC power supply and the power consumption is really low. The main drawback is modest performance for the directly heated tubes, but presumably that applies to larger-sized directly heated "battery" tubes too.

Other advantages: (1) Ruggedness. (2) Minimal heat emission (even after they've been on for hours, the 1AD4s are barely warm to the touch) which should make for lower drift in receivers and VFOs. (3) Built-in shielding, in the form of a "sprayed-on" shield on top of the glass envelope (some full-size tubes used this technique too).

Disadvantage: you don't get that warm vacuum-tube glow!

General advantage of directly heated tubes (not just subminis): "instant-on" because you don't have to wait for the cathodes to warm up.

Later on, I'll be experimenting with the cathode-type versions of these tubes (6.3V filaments). These have good performance; some of them are similar in size to the one shown above; there are also subminiature tubes that are very slightly larger (8-pin, circular-body) that would be good in transmitter applications.

You can get an idea of the great variety of subminiature tubes in this Raytheon catalogue: http://www.tubebooks.org/tubedata/RaytheonTubes.pdf.

The tubes are still quite widely available. The sockets are much harder to find; but if you use the leaded types you don't need sockets at all (although that makes them harder to replace if they blow).

(Of course we also had Nuvistors later.)

73 de Martin, KB1WSY
« Last Edit: May 06, 2015, 06:13:15 AM by KB1WSY » Logged
G3RZP
Member

Posts: 1321




Ignore
« Reply #12 on: May 06, 2015, 06:09:03 AM »

The suggested circuit is a no-no. It won't put bias on the tube because of the DC path to ground through the coil. It also shunts the signal path.

Connect the grounded end of the grid coil to ground through a disc ceramic capacitor of  2 to 5000pF, with short leads, and connect the tuning capacitor grounded end to the same point. If the tuning cap is variable, you want the disc ceramic grounded end to be physically on the tuning capacitor. If it's a variable iron cored inductor, connect the tuning cap across the inductor. Connect the variable RF gain pot (I'd suggest something like 50 or 100kohm) across the C battery, and take the wiper through something like 10k to the 'groundy' end of the coil, the disc ceramic capacitor. You want about 4.5 volts in the C battery, and a 0.1 mFd from pot wiper to ground to minimise noise as you turn the pot.

The bypass from the end of the plate coupling winding is a bit on big side: with 1/2 inch leads, it's series resonant about 5 MHz. Use between 2 and 5000pF for that and the screen bypass.

It appears that most pre-war designs of successful TRFs used quite large coils (inch to inch and half diameter), well spaced from metal work, and with around a 1:1 length to diameter ratio for maximum Q.
Logged
KB1WSY
Member

Posts: 1309




Ignore
« Reply #13 on: May 06, 2015, 06:56:54 AM »

The suggested circuit is a no-no. It won't put bias on the tube because of the DC path to ground through the coil. It also shunts the signal path.

I was wondering about that already -- in particular the DC short through the coil to ground!!!

I know almost nothing about how to adapt cathode-type circuits for a directly heated tube. With your help, I am learning fast, it seems. Which is just as well because almost all the circuits in textbooks are for cathode-type tubes, after the first few pages of elementary tube theory with filament tubes.

Meanwhile, you can be grateful that I'm only playing with "low" voltages, huh? In this case the main result would have been a melted battery holder I suspect (yes, I've done that too, a few weeks ago when I shorted a battery).

Given the large number of changes you suggested, here's the revised schematic: is this what you meant?



73 de Martin, KB1WSY
Logged
G3RZP
Member

Posts: 1321




Ignore
« Reply #14 on: May 06, 2015, 09:35:12 AM »

I'd suggest a capacitor from the pot wiper to ground. At least 0.1 microfarad: 1 mFd wouldn't hurt.

To avoid all doubt, "2-5000pF" means "2000 - 5000pF".

Deciding on how many turns for the plate coupling coil is another matter. One can go for a large coupling coil approximately self resonant  but loosely coupled to the detector coil, or for a smaller coil  more tightly coupled. Now you are currently running with a 5 microhenry coil: I would go for 10 microhenries and less capacity. This looks like around 27k with a Q of 60. Now have about a half as many turns on the coupling coil and the plate load for the RR stage is 27/4 kohm and the gain is about 6 to 9. The transformed plate resistance is thrown across the tuned circuit: that will have little effect. So there can be reasonably tight coupling between the coupling winding and the detector.  I wouldn't wind the coupling winding over the tuned winding but put it 1/8 inch away and close wind it.
Logged
Pages: [1] 2 3 4 5 6 ... 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!