(18-09-2021, 02:25 PM)Mike Watterson Wrote: [ -> ]Naturally direct filaments can't use a cathode resistor...
But whyever not? I have an Ever Ready model 5033 which uses a directly heated triode (S30D) as output valve. It has its own filament winding on the mains transformer with the 'cathode' taken as centre-tap of this - with a resistor and bypass capacitor to chassis!
A friend has been experimenting this week using 1j29b's fed from AC using a few turns thrown onto a toroidal mains transformer, with encouraging results. This allows the 'cathode' connection to be elevated easily.
I've wondered about how the 1p24b would cope. With its thicker filament, temperature variation with AC will probably be a comparatively minor matter. There's still the voltage variation along the filament, but if 1.2V operation with the two halves in parallel is used, no part of the filament will be more than 1.2V x root-2 (= 1.68V) away from the mid point, at any part of the cycle. And as the 1p24b does not have a very sensitive grid, I'd expect this to make little difference. There's only the magnetic field due to filament heating current, within the valve, left to consider. And, putting as many filament windings on a toroid, each at just 1.2V, is the work of a few minutes each!
Yes I'd agree Peter.
I've done it with a separate DC supply and it works fine.
A toroid with separate feeds sounds good to me.
(18-09-2021, 05:47 PM)Kalee20 Wrote: [ -> ]...if 1.2V operation with the two halves in parallel is used, no part of the filament will be more than 1.2V x root-2 (= 1.68V) away from the mid point, at any part of the cycle.
Too late to edit... so just drawing attention to my basic cock-up here.
With parallel filaments (1.2V total) the maximum voltage difference from the mid point is of course HALF this figure, or 0.84V.
And of course, the two ends of the filaments will be opposite polarity so will have opposite effects... the total current flow will be approximately what you'd expect from a unipotential cathode. It's only the non-linearities and any construction asymmetry which'll prevent the cancellation being perfect.
The little Ruskies should be ideal for this because the suppressor is taken to a separate pin, it can be taken directly to deck, as opposed to being tied internally as is the case with Dx96 series of B7G directly-heated valves.
(18-09-2021, 05:47 PM)Kalee20 Wrote: [ -> ] (18-09-2021, 02:25 PM)Mike Watterson Wrote: [ -> ]Naturally direct filaments can't use a cathode resistor...
But whyever not? I have an Ever Ready model 5033 which uses a directly heated triode (S30D) as output valve. It has its own filament winding on the mains transformer with the 'cathode' taken as centre-tap of this - with a resistor and bypass capacitor to chassis!
Obviously possible using a separate supply, battery or mains. As I implied on the o/p valve bias system on battery portables. I wasn't explicit enough. Another trick is a choke (1:1 transformer) with the DC currents cancelling feeding the filaments. Easy at RF & IF, larger at AF. Allows cathode follower or grounded grid.
VFDs often use a centre tapped filament supply to reduce brightness gradient across a panel.
(17-09-2021, 06:30 PM)BusyBee Wrote: [ -> ]I have found great differences as previously reported and the results do indicate that individual setting up may well be beneficial. My samples were 1j24b's. I do wonder whether 1j24b-v's would have closer characteristics but I do not have any to compare.
I think there is only one grade of 1j24b. The 1j29 series and 1p24 series both come in different grades, but the differences may be mechanical or power handling rather than electrical uniformity.
(19-09-2021, 01:49 PM)Mike Watterson Wrote: [ -> ] (17-09-2021, 06:30 PM)BusyBee Wrote: [ -> ]I have found great differences as previously reported and the results do indicate that individual setting up may well be beneficial. My samples were 1j24b's. I do wonder whether 1j24b-v's would have closer characteristics but I do not have any to compare.
I think there is only one grade of 1j24b. The 1j29 series and 1p24 series both come in different grades, but the differences may be mechanical or power handling rather than electrical uniformity.
Thanks for this. Now I am reminded I do think you may have mentioned it before but I had forgotten. It does make me wonder all the more about how the Russians handled such variation.
(21-09-2021, 06:38 AM)BusyBee Wrote: [ -> ]Thanks for this. Now I am reminded I do think you may have mentioned it before but I had forgotten. It does make me wonder all the more about how the Russians handled such variation.
Design, multiple HT rails and dividers on g2 (which isn't really a variation issue).
Regular valve variation is huge. On indirect heaters the cathode resistor hides it. Some designs and parts of circuits are more tolerant. A good radio may still work well with only 30 % emission on some valves. The UHF tuner valves had the shortest life, though an EF80 elsewhere in a TV might be more worn and it wouldn't matter.
When o/p valves used a grid battery rather than a cathode resistor they had a suffix for grid bias setting or a coloured dot.
Transistors, especially for the first 30 years, had immense spreads of gain and VCEo and VCBo ratings. It wouldn't surprise me if the early 1950s coloured dot ones were select on test as RF, IF or preamp and if later a BC107, BC108 and BC109 were all simply select on test for max operation voltage (BC107) or max gain (BC109). They even had a, b & c gain groupings on some.
Modern VMOS FETS, when miss-applied as CB RF PA and driver need the bias voltage adjusted. There is a big variation in voltage at threshold.
You'll find the 2N3819 j-FET has quite a variation of Id at VGS=0 and cut off voltage.
But the issue is magnified by the rod pentodes because when Ia= max, the Ig2 is nearly zero, uA anyway, but at Ia nearly zero the Ig2 is maybe a 1mA. Thus ONLY a series resistor to 80V HT varies Vg2 too much.
(22-09-2021, 02:18 PM)Mike Watterson Wrote: [ -> ]But the issue is magnified by the rod pentodes because when Ia= max, the Ig2 is nearly zero, uA anyway, but at Ia nearly zero the Ig2 is maybe a 1mA. Thus ONLY a series resistor to 80V HT varies Vg2 too much.
That's interesting, Mike.
I've certainly found that Ig2 is low with rod pentodes. What I don't have data for, is how constant, between samples, is Ig2 as a fraction of Ia.
Ia being nearly zero brings Ig2 up to a maximum surprises me - again, for a given valve, I've found the ratio to be constant. The only circumstances where Ia falls and Ig2 rises, is if anode voltage drops to a low level (below the pentode 'knee') - or if the suppressor voltage is made very negative so that electrons are repelled back to g2 and can't make it to the anode.
The Rod valves I ordered from Ebay have arrived today, I also have 4 extra pieces of glass included 3 x 6C7b Triodes and 1 diode 6D6A, these both being 6.3 volt heaters. It will be some time before I get chance to delve into any of these bits of glass.
Adrian
I presume you mean 6s7b. I had plans for using one of these for an oscillator for a superhet but, after building a test circuit using a fet, I am holding off at the moment as there are more impotant things, and I do realise that the fet will likely be more stable. A valve circuit would give more drive though. The 6d6a I have used but only fleetingly. It is a signal diode giving response similar to half a EB91 when I tested it.
Tracy