Mayhem on AmRad 40 m. band

[Mike Watterson]

The few hours life was being over driven for power, proximity fuzes are related to radar, and that was only some types.
They can easily be longer life than RCA B7G 1.4V battery valves. In radio use the filament was run at 1.2V from NiCd and an oscillator may still run down to 0.7V or 0.8V unlike the 0.95V to 1V of 1R5/DK91 and later replacements. The only non-military two-way radio was a USSR model for forestry workers. Some models manufactured up to 1991.
There are some articles here and on the Radio Museum. Look for articles by Joe Sousa, an expert engineer in Linear Technology now part of ADI.

The 1j29b is essentially two valves in parallel sharing anodes (all rod tubes have a pair of anodes like flat rods. The g1 is a pair of flat rods very close to cathode filament. The g2 and g3 are focus rods either side of the beams. There are two beams. The 1j29b will do 1W. With one filament only it can replace most 1.4V audio Pentodes, but almost all the rod tubes work to over 100 MHz. The Russian sets used a 1j37b for AGC as the g1 connections are separate, thus a differential voltage (either polarity) steers the beams off the two anodes and reduces gain. They can also work as unusual mixers or frequency doublers or detectors.
I've used a pair of 1j29b to replace a broken KLL32, with suitable g2 pull down resistors. The two 1.2V filaments in series (common at one end anyway) work well enough off the 1.8V to 2.3V Lead acid cell I have. The g2 current sort of mirrors anode, so the g2 has to connect to an HT line or resistive divider. Strange instabilities will result using a resistor to set g2 current, because that doesn't work! It's part of an electrostatic lens so needs a voltage.

[Jez]

Your post last night led of down a rabbit hole searching for info on these and yes I read the articles by yourself and Joe Sousa at radiomusuem and the entire 10 page thread here, which I just replied to. I need to get some of these!

Something that has me most confused though is that I'm seeing thing's like "max anode current 8mA" and that HT should be keep relatively low like 45V if using this max figure... and then I'm seeing such as "a pair of these in push pull can give 4W output" and above "The 1j29b will do 1W".

It would be interesting to see how they perform as hi fi audio amplifiers, pre amps etc that is.

The directly heated cathodes could prove "fun" here! I'm thinking of constant current sources as a solution to this.

[Mike Watterson]

I've done graphs of filament IV. And Joe of performance vs Vf. You do not need a constant current or even constant volts, just noise free. More sensitive to Vg2 than Vf or Va, due to electrostatic lens.
Yes, the j29b will do 1W DC to 100 MHz, or 4W in class B push pull (I use a small dual 110V/120V primary mains transformer. People have built entire FM stereo radios, HiFi Amps d two way QRP. There is only any point to the really high power ones if using 1.2V or 2.4V NiMH, as it's 220mA 1.2V and 110mA at 2.4V.

[Jez]

I'd not be using batteries! Hence the CCS power for filaments. I can't see how the published specs allow anywhere near those outputs.

[BusyBee]

I'd not be using batteries! Hence the CCS power for filaments. I can't see how the published specs allow anywhere near those outputs.

I have used constant current but found it far more fiddly to setup than my usual alteernative - voltage from a DC regulator. I tend to use one of the IC regulators, like LM317, just using it's internal regulator which is around 1.25V, so extremely simple, then put in a series choke to the filament to drop the excess voltage with it's series resistance (and also work as the choke which these are often used with for RF circuits.

Tracy

[Murphyv310]

I'd not be using batteries! Hence the CCS power for filaments. I can't see how the published specs allow anywhere near those outputs.

Interesting you say that the specs differ from what's been achieved.
Look at the PL509, PL519 and EL519. Many amateurs built linear amps and got powers way above the specified figures, the reasons are the HT level was much higher and the duty cycle enabled short bursts of much higher power.
The humble ECL80 Pentode section will handle 1.3kv on its anode!
Again a true class B stage will deliver more than double the same valve would running class A, sorry to preach to the converted!

[Mike Watterson]

I'd not be using batteries! Hence the CCS power for filaments. I can't see how the published specs allow anywhere near those outputs.

Constant voltage is simpler and fine. CCS might be the fashion for valves, but I've seen no evidence that it superior to a good design otherwise. The filaments have a significant temperature co-efficient, especially the 1j24b.

Class A has 50% efficiency.
True Class B push pull is much better and you double the power because two devices. It's almost 79% efficiency for linear RF. Audio might want a little more bias (some variation of class AB). Class C is only good for CW, digital OOK, FM and RTTY or similar.

Also regular audio has a lower average power (so called RMS power) than a continuous sinewave. The power is actually limited by the filament on direct cathode battery valve, hence take care to run at 1.1V to 1.2V on power amps and decouple the centre tap on series 1j29b. Due to poorer filament tolerances than the Western 1.4V 50mA nickel or 1.4V 25mA tungsten any series operation of multiple valves needs select on test and ballast resistors as different models are all different currents like the low profile 1940s German metal can valves (used till 1950). When designing for parallel only the filaments can be optimised to reduce power. The 1.4V 50mA Octal Sylvania, 1.4V 50mA B7G RCA and 1.4V 25mA B7G Philips designs deliberately use the same filaments (not optimised) to allow series or parallel operation. The series only mains valves (U, C, H, P) are optimised which is why the higher power ones progressively use higher volts.

The rated figures are for particular life and performance. Philips on the LX527AB and similar Push-Pull DL94 pair (A Philips made part) uses 153V HT (after subtracting LT) on mains. The Philips datasheet has 120V!

The limit on battery valves, especially the rod pentodes, is ultimately average filament temperature.

So 1W Pa gives 1W max audio and uses 2W in Class A.
But 1W PA is 2W Pa on Class B, because there are two. That then is about 7W audio (At 78% that's 0.78 x total 9W). So a 400mW Pa for two valves in Class B would be nearly 3W. If filament is at low end of recommended volts, then 4W at reasonable life.

If it was parallel Class A like an FT101 using 6146, 40W Pa is total 80 W Pa (because there are two) and 80W RF. But if a wideband transformer was used (that was big enough!) then a pair of 6146 would give 280W instead using Class B. They didn't do that because such an RF transformer would be very expensive and the mains transformer HT would be increased from 160W approx to 360W! Cheap to get 80W by two 40W tubes in parallel. Though people think it's a 100W rig.

FM or CW only rigs are simpler. Class C. Can beat Class D on a good day.

Digital at sau 16 QAM or higher needs a progressively more linear design. A Class B RF TX can to an extent cheat with filtering.

[Jez]

Although I use valves quite a bit I've never had anything to do in practical terms with directly heated ones so I may be making a silly beginners mistake here.... but the reason I'm talking about floating CCS 's, one for each valve, (and a couple of silicon diodes across each filament to prevent over voltage), is that audio, probably multistage high gain amplifiers, is my main interest here (I'll have to do a superregen though! ) and surely the heater IS the cathode hence all valves would "see each other" through the heaters/be cathode coupled to each other, at least partially, if not isolated from each other AC wise? I may well need to take NFB back to the cathode of the first stage for example.

Obviously the usual filament chokes as used at RF would be useless at audio frequencies. In fact some probable uses I may have would incorporate DC feedback so even a truly huge choke would be no good.
The only thing that springs to mind otherwise is that all valves would have to be grounded cathode at AC, no degenerative cathode feedback allowed, and each valve in fixed bias... Hmm come to think of it that's probably why there was a separate bias battery in early 2V accumulator powered heater radios  cathode followers and GG could be "fun"!

If that's the case then being limited to fixed bias grounded cathode would limit their usefulness to me somewhat.  I guess RF powered heaters from multi secondaries could be another solution.

[Mike Watterson]

Bias batteries in early 2V sets was nothing to do with cathode circuitry, it was partly to do with with cost of a resistor and that the o/p valve (often the only one with negative grid bias) was colour coded to set the bias. Originally 9V or even 12V packs of B cells tapped every cell (nominal 1.5V). Also being a triode, the anode current would vary as HT ran down. This is a problem for moving iron 2K to 4k cone speakers as the armature would move closer to one pole. There was also a mechanical adjustment, but that was to give more sensitivity for quiet stations. So the grid tap could be reduced as HT aged.

Mallory came up with the idea of a sealed mercury button cell for the O/P pentode (or triode) on later battery or mains sets. A battery of them would be made. However that idea wasn't a success. The cells were used in WII for long shelf life in proximity fuzes in shells and bombs then in early transistor hearing aids. That made them a success. So later he bought a licence from either Eveready (NCC/UC /Ucars) or British Ever Ready (which was founded by US company) for alkaline cells which Ever Ready or Eveready (not sure which) wasn't interested in selling. Mallory marketed those as Duracell, so successfully they renamed the company. Both Ever Ready (Gold Cell) and Eveready (Energiser) belatedly sold them.
Mallory's package technique was later used for Silver Oxide, Alkaline and Lithium cells.

The battery sets got rid of the bias pack when they were Pentode based on o/p and HT voltage not so critical. A resistor was placed between HT- and LT- to develop the negative bias for the o/p Pentode. The other valves use grid leak space charge to bias negative. Interestingly the DK9x or 1L6 L O grid can be DC filtered with 2 x 1M in series and give a negative enough bias for class AB on the DL96 in Push Pull, so some sets don't sacrifice the HT. A 90V layer pack has an endpoint of 54V. But the B7G 1.4 valves will work lower than that, so some sets used 45V (earphones only and 22.5V end point) or 67V in PP3 style packs but longer. The 22.5V in an Avo was used in hearing aids. Older ones used two packs. Later sub-miniature hearing aid valves are 0.7V filament to allow two (the entire amp) off one 1.5V cell and one 22.5V cell.

The only time (and very very rare) that isolated filaments used via a bifilar RF balun (the DC cancels) was the cathode of a grounded grid RF pre-amp. In such case the DC is the regular LT and 0V. I don't know of any cascade amp which would need a floating LT.

A stereo decoder might want a balun or transformer on the cathode, but it would be 19kHz and up, so not big and the filament connections could be bifilar only needing to be a choke at 19kHz to 54kHz (or maybe just 19kHz) and a capacitor to couple. Grundig did a single pentode decoder using also diodes as mixer and audio transformers. It doubled the 19kHz pilot at the cathode to get the 38kHz to mix into the L-R DSBSC and then two transformers.

Better valve amps in radios use both feed back to O/P g2 from a winding on transformer primary and feedback from speaker or extra winding to input. All the cathode resistors are ONLY for grid bias, all decoupled at audio. So that architecture will work with direct filaments. You still need some negative bias from somewhere for the o/p and that is usually (as in all battery valve sets approx 1937 to 1962) a resistor between HT- and 0V carrying entire HT current.

[Jez]

Some interesting historical perspective there....

If/as/when I get some rod pentodes to play with I think I'll be using my method. It's comparatively complicated but I reckon it should satisfy the needs I have for circuitry with cathode biasing, grounded grid, cascode amps and cathode followers etc.

What I'm thinking of is a master filament supply of something like + & - 12V (another separate isolated set of supplies floated up at say 40V or whatever for high cathode potential uses such as cathode follower or cascode as necessary) very well smoothed and regulated, which power current sources and sinks. The valve filament connects between them and has a pair of 1N4148 or a 1.25V bandgap reference between filament + & - to maintain 1.25V actual filament voltage (hence actual filament see's voltage drive) whilst a DC servo op amp adjusts one of the current sources to centre the cathode voltage to where it is required. Op amp and current source forming a sort of DC transconductance servo loop. This last bit would also allow bias to be accurately adjusted and maintained as desired to any reasonable point between around + & - 8V ish, hence simultaneously providing adjustable fixed bias whilst the cathode floats at many Meg Ohm away from ground and rails from DC to at least a few 100 KHz (filament chokes or common mode choke could be used to extend it to MHz etc if needed. Obviously the scheme is mainly intended for audio work). Et voila, instant "indirectly heated" valves   There could be a gotcha I haven't thought of here of course!

Probably talking 6 transistors and an op amp per rod pentode to do this. Worth it? For potential high end hi fi use's then yes.