PY 81 as power rectifiers?

[Terry]

Or at least I'm not clever enough to do it

That's today - there's always tomorrow!

Terry's suggestion of grid stoppers is a good one ...

Err, no! The grid stopper suggestion was Amie's!

What I suggested was putting an RC network in the earthy end of the grid leak to improve the smoothing of the bias on the triode's bias supply, effectively making it independent of the pentode bias, even though both will still be derived from the same source.

[Kalee20]

Thanks, Terry... Sorry, Amie!  

What I suggested was putting an RC network in the earthy end of the grid leak to improve the smoothing of the bias on the triode's bias supply, effectively making it independent of the pentode bias, even though both will still be derived from the same source.

Not sure that the idea of an RC network in the earthy end of the grid leak to improve the smoothing of the bias on the triode's bias supply would give benefit, I'd suggest the reverse.

As an extreme, consider the triode grid hard-coupled to the cathode. If the pentode current wiggles a bit (noise) then the cathode voltage will wiggle, and the triode grid will be taken with it. So there will be no wiggle in the voltage between triode grid and cathode.

Consider the other extreme, the triode grid hard-coupled to 0V. If the pentode current wiggles a bit (noise) then the cathode voltage will wiggle, and as the triode grid is grounded, the full wiggle will appear between triode grid and cathode. This will be in such a phase to reinforce the original wiggle, hence positive feedback and oscillation is likely

Increasing the capacitor between triode anode and pentode grid will aggravate the condition unless the cathode decoupling capacitor is increased with it. But as the output transformer is easily the limiting factor in the bass response, there probably is not really much point in going above 2.2nF. It's easy to verify, though.

[Terry]

Not sure that the idea of an RC network in the earthy end of the grid leak to improve the smoothing of the bias on the triode's bias supply would give benefit, I'd suggest the reverse.

Considering that no bias current flows in the grid leak, the series resistor can be quite high - I think I suggested 100k or more - and, with a suitably high value of decoupling cap, should completely isolate the triode grid from any feedback from the pentode's cathode.

The size of the capacitor is limited solely by the RC time constant so that there is no undue delay in generating the bias voltage as the valve warms up.

The alternative, of course, assuming that the valve type remains unchanged, is to AC couple the triode grid and fit a 10M grid leak between the grid and earth.

[Amie]

Hum cancelling....So...What if anything, went wrong ?:

https://www.radiomuseum.org/forum/grundi...guage_id=4

Lawrence.

Just to be clear, I meant what went wrong in the radiomuseum article insofar that his calcs didn't stack up to what he thought they might.

Lawrence.

It's beyond me, I've got a handle on why the balance current deal didn't really get off the ground, but hum cancelation seems a little pointless, given the difficulty in realising it and with the large valUe electrolitics available?

Terry/kalee20....
I think I will just poke the components leads through the tags and solder not wrap and solder! 

More experiments needed on the biasing?

Amie

[Terry]

Terry/kalee20....
I think I will just poke the components leads through the tags and solder not wrap and solder! 

Perfectly acceptable way of mounting components - think of the way they are fitted in a printed circuit board!

[pwdrive]

It's beyond me, I've got a handle on why the balance current deal didn't really get off the ground, but hum cancelation seems a little pointless, given the difficulty in realising it and with the large valUe electrolitics available?

Amie

Yes, for cancellation the correct transformer is needed.

Have a look at Schades curves to see the effect of increasing the reservoir has on the ripple voltage in percentage terms etc, the whole article is well worth a read, Shades article starts on book page 174 (then scroll down for the curves):


http://www.nj7p.org/Manuals/PDFs/Books/RCA_1940_VTD.pdf

Lawrence.

[Kalee20]

Terry/kalee20....
I think I will just poke the components leads through the tags and solder not wrap and solder! 

More experiments needed on the biasing?

Amie

Definitely!

Anywhere you read. "Make a good mechanical joint by wrapping the wire around the tag with pliers. The solder is there to complete the electrical connection," just forget it. A quick blob of solder is all that is needed, especially when experimenting. You need to be able to take the joint apart without overheating the other components as you struggle to unwind your beautiful wrap while keeping it molten.

The good-joint technique is for when making a final version that you want to work reliably for 20 years. But not for the lab where you may change your mind about something after 5 minutes use.

Yes, please experiment with the biasing! You won't hurt the triode, even if triode bias is zero. The pentode won't like under-biasing as the current is only limited by the output transformer's DC resistance, so a bit of care is needed (but unlike a transistor, the ECL80 will not see instant death).

[Kalee20]

Not sure that the idea of an RC network in the earthy end of the grid leak to improve the smoothing of the bias on the triode's bias supply would give benefit, I'd suggest the reverse.

Considering that no bias current flows in the grid leak, the series resistor can be quite high - I think I suggested 100k or more - and, with a suitably high value of decoupling cap, should completely isolate the triode grid from any feedback from the pentode's cathode.

The size of the capacitor is limited solely by the RC time constant so that there is no undue delay in generating the bias voltage as the valve warms up.

I'm still not in agreement - any chance of putting up a quick sketch Terry? The unwanted feedback to the triode, is to the triode CATHODE, not the grid, and the cathodes can not be isolated.

[Terry]

As an extreme, consider the triode grid hard-coupled to the cathode. If the pentode current wiggles a bit (noise) then the cathode voltage will wiggle, and the triode grid will be taken with it. So there will be no wiggle in the voltage between triode grid and cathode.

Consider the other extreme, the triode grid hard-coupled to 0V. If the pentode current wiggles a bit (noise) then the cathode voltage will wiggle, and as the triode grid is grounded, the full wiggle will appear between triode grid and cathode. This will be in such a phase to reinforce the original wiggle, hence positive feedback and oscillation is likely

The real problem is that the triode needs a different bias voltage to the pentode, so option 1 above is not viable. However, if the grid voltage is decoupled as I said before but the decoupling cap is connected to the cathode, rather than to ground, the problem will no longer exist.

A better solution, though, would be to connect the 0V line directly to the cathode and use a 10M or greater grid leak on the triode returned to ground for the triode bias - which is the way that Grundig did it in the previous circuit.

I'd prefer a valve with separate cathodes for the triode and pentode sections but that's a personal preference. However, as Grundig have demonstrated, there is no reason why Amie's ECL80 can't be used in a configuration that guarantees success.

[Kalee20]

That would indeed work Terry!