Golborne Vintage Radio

Mike Watterson Wrote:

Amie Wrote:As a suggested method if you wanted to construct a sensitive infinite impedance detector for the radio using those valves, could you use a floating supply for the filament? A separate LT cell for each valve wouldn't be a hardship....but then why fix what isn't broken?

I can't see how that does anything much as the filaments are the cathodes and the input impedance is much higher than regular valves.

My suggestion was more from the point of view of the linearity and ability to handle a wide variation of signal levels without distortion.

Amie

Why the "infinite impedance" detector is pointless for a two valve TRF:
The so-called "infinite impedance" detector is just a cathode follower, i.e. Anode is common. It has no gain at all. The valve needs to be biased just to cut-off. It isn't necessarily lower distortion than a diode, except for larger signals. Because the gain =1, it's a bad application for weak signals or low level signals. An 1N60 or OA81 etc is better.

You do not need a floating cell. A filamentary (direct cathode) valve can be "common grid" (Cathode is input, Anode is output) or "common anode" (grid is input and cathode is output).
Historically a transformer wired as a common mode choke is used for RF (the DC + and - currents cancel, but the RF "sees" the two coils in parallel, which doesn't 1/2 the inductance if they are closely coupled but does double the Q). Two coils in series ON THE SAME core gives x4 inductance. A typical transformer would be a 2V valve interstage transformer or a 2V valve TRF audio choke for capacitive coupling. You need three valves minimum if using the "infinite impedance detector". The input impedance is load resistance x gain in series with space charge impedance, not infinite.
Because it's a low level signal and the gain is needed, I can't see the point. A simple way to limit the gain on strong stations is a pot on G2 on the 1st valve. Manual RF/IF gain on the Hallictrafters S72 is done this way, and on some Russian sets with Rod Tubes.

For a direct filamentary cathode valve audio "infinite detector" you can either use a common mode audio choke or a transformer that won't saturate with the filament current. If you have a "floating cell", you are placing what was the anode load in series with cell to -HT and the gain = 1. Also you need SOME current to bias the 1j42a to "almost off". This is best arranged by a resistor to a stack of button cells, though you can try using just the load current. The grid has to be connected to -HT or 0V by SOME sort of DC path unless it's an electrometer, like 10M Ohm.
Actually the very first "button cells" were mercury type and invented by the Chemist that founded Duracell, then called Mallory. They were to bias grids of out put valves without cathode resistors and decoupling capacitors, a cheaper solution to last 5 to 10 years in 1930s. It was a commercial failure. His success was selling them to Military in WWII for shell and bomb 'proximity fuze' electronics due to the years of shelf life, even in tropics where Zinc cells might self discharge in a few months.

Thanks Mike.

I'm just having a bit of a problem with visualising the transformer fed cathode circuit.

I have seen demonstrated a homebrew twin tuned trf where all valves were EF50's, 1 RF amplifier 1 II detector, and 1 audio amplifier, feeding phones. Pointless or not it worked well. But of course not rod pentodes!

Thanks.
Amie.

See, THREE valves.
Basically you are replacing a diode with an EXTRA valve.
It's only an advantage if a valve is as cheap as a diode (which they were once) AND you have no RF gain control or large signals.

The transformer wired as dual winding common mode choke is in series between the filament f+ and f- and the f+ supply AND 0V.

An RF one allows you to use the filament as a signal in (capacitors to f- & f+) in common grid mode.

An RF or audio one allows a cathode follower, for signal out via capacitors from f- and f+. The transformer or common mode choke allows the filament current but "blocks" RF / IF / Audio depending on value of series inductance. Admittedly at Audio, with 11mA, the winding resistances to have a decent inductance might mean a higher LT voltage.
It could have a 3rd winding to step up the voltage.

Really it's practical down to maybe 19kHz for a multiplex decoder and not for audio. Also you always, always need an extra valve, because the valve is replacing a diode.

Actually in Trevor's original circuit, the capacitance of the diode and the massive high input impedance might mean that the valve is biased nearly off by space charge and the positive AC through the diode capacitance is more amplified than negative AC. You might get nearly the same result as an "Infinite impedance" detector, but with real gain, just by a 10M Ohm resistor to -1.4V alkaline button cell or -2.9V lithium coin (depending which has the 1j42a nearly off). The cell will have shelf life. Coupled with the capacitor and no diode at all.

In 1949 an EB91 dual diode ( EB91 = 6AL5 = EAA91 = 6B32 = 6D2 = CV140 = CV283 = DD6) was cheaper than 1 x 1N34 or 1 x 1N60.
If you had a AF preamp, you could get one with a diode built in virtually the same price as a triode or pentode on its own. So when superhets with AGC were the norm from 1935, there were also triodes with diodes (free) in the same envelope.
Occasionally there were RF/IF pentodes used with a detector diode, for superhets with no preamp, only one audio valve. Or the EBL1, EBL31, EBL21 and EBL71 output Pentodes with dual diodes.
There is no real need for a second diode for AGC, but it was "free" and usually driven with a capacitor and was across the AGC line to 0V, so it or the main diode ( xAC series triodes and xAF series Pentodes, in 2V, 1.4V, 4V, 5V, 150 mA, 200mA, 300 mA series and 6.3V etc) gives a more negative voltage the stronger the signal. AGC isn't so common on TRFs, except late ones like the Philips Superinducance series.
Early TRFs adjusted the filament current on the RF Triode or SG (RF Tetrode) for a volume control, if it wasn't purely by Reactance feedback.
The Infinite Impedance detector is rare, because before combined triode-diodes and pentode-diodes in superhets the gain was an issue, so reaction, leaky grid, etc was used. Very occasionally in early 1930s to save a valve they used a semiconductor rectifier, I think usually copper oxide. Germanium diodes started to be used in professional equipment in 1938 or 1939. The 1N34 and 1N60 easily available by 1947.
The first common use of semiconductor diodes was in battery radios with FM. Germany started FM in 1949 because of the Allies taking MW & LW allocations. Thus unlike UK 1955 to 1971 (approx) the FM actually had extra stations. The first models used DC90 as FM mixer/osc. One early model might have used the EC90 (150mA 6.3V). They almost all used DK96 as AM mixer/osc with DAF96 as the AM detector and AF preamp. All used two semiconductor diodes for the FM detector. A valve solution was too much filament current. Later the DF96 was replaced by DF97 on some for IF and on all a triodised DF97 instead of the DC90. They are interchangeable with little shift in tuning, though the DC90 is 50mA and DF97 is 25mA. On some DF97 sets you need to cut an extra earth wire to use a DC90, which doesn't seem to affect operation. AFAIK all AM/FM battery valve sets are parallel and the US never made any, because FM was regarded as HiFi only. There was one USA "novelty" battery valve model, an FM super regenerative single valve type.

It would be interesting to see what is the maximum frequency the 1j42a does. The other Rod Tubes certainly are roughly as good as an EF80 for VHF oscillators.

Hi.
An update on some further tests.
Adding in a diode between D1 anode and ground virtually damps the audio to nothing as does a good low capacitance 2.5Mh choke. Seems the only way this performs is a Diode with no DC path or with very high resistance values where there is a marginal performance gain, distortion is also less. I also found two AA143 diodes that work well except for more distortion. I'll now wait till I have the 1N60P diodes I ordered.

I made a super regen set with a 1j24b that works very well right up 108 MHz, in fact I rewound the tuning coil and was able to get TV sound off an aurora standards converter on Channel 7 so well into band 3

I'm just wondering if the 1j42a would work in this circuit, may try one some day.

Murphyv310 Wrote:Hi.
An update on some further tests.
Adding in a diode between D1 anode and ground virtually damps the audio to nothing as does a good low capacitance 2.5Mh choke. Seems the only way this performs is a Diode with no DC path or with very high resistance values where there is a marginal performance gain, distortion is also less. I also found two AA143 diodes that work well except for more distortion. I'll now wait till I have the 1N60P diodes I ordered.

That's really surprising! Just checking, the 'new' diode is the right way round?

All I can think, is with things now 'right' that the 2nd valve is cutting off with the DC form the detector being too much. How about a 10M grid leak, with the AF coupled to it by 0.01uF capacitor, with the detector loaded by 1M (ie across your 100pF)?

Hi
Diode definitely in correctly, worth a try coupling the G1 with a cap and grid leak. Tomorrow now though.

Just been toying this over in my head. As the diode is positive going I'd expect a more positive voltage on the G1 with a strong signal and the Anode volts to drop, the anode volts on the AF amp is 6.2v after being dropped by the headphones. With the resistor network this voltage is constant irrespective of strong signal or off tune. This test was done with a 10m grid leak and 18m from diode anode to ground. What I will do is wind a 1:1 RF transformer and wire the diode direct to the secondary with the other end grounded.
I've got the feeling the issue is pure and simply the extremely high input impedance and anything will damp it. Even touching the diode reduces volume massively.