EL84 p-p Resurrected

[Mike H]

I was trying to find out where I was posting about this originally, reason I can't find it, it wasn't my topic it was Steve C's! I was just barging in ...

Anyway, it was about this.....

Chassis-Image-004.jpg (104.42 KiB) Viewed 3376 times

-
And I couldn't spell 'resurrected' ...

You may or may not remember the mains TX burned up (got too hot inside the case), so it had an 'external one' bodged up on the end of an umbilical to make work, and then after its last Owston trip the main switch broke, and it's been waiting in a corner ever since.

[Mike H]

Finally had a grope around the innards today, and discovered one of the o/p transformers was wired up wrong, the o/p was on the 16 Ohms tap not the 8 Ohms tap! This is an error from the beginning. What's significant about this (or vaguely significant) was that I had to reduce the open-loop gain because the right channel, only, oscillated at about 1 Hertz. Can you guess which channel was wired correctly? The right. The left had reduced gain due to the transformer presenting 4 kilohms to the o/p valves instead of 8 kilohms (I'm guessing).

I reduced the gain of the 1st stage, we will see how that goes in due course. There's always the option to convert the 1st stage to triode mode if there's any more trouble.

Photo after I'd finished changing the components around the V1 stages.

EL84 p-p_5697.jpg (136.07 KiB) Viewed 3375 times

[Mike H]

Before and after —
Schematic as it was, R15 was added to reduce gain of U1 stage ... which is a bodge ...

EL84_p-p_amp_005B.jpg (86.29 KiB) Viewed 3372 times

[Mike H]

Revision 006B — maybe a better way is halve the values of the anode and screen resistors (values from Mazda or Philips datasheet) ...
-

EL84 p-p Amp 006B pentode-2.jpg (96.48 KiB) Viewed 3372 times

-
We shall find out in due course!


For reference, the 'K' - 'E' block represents a pair of CCS, each as below — the intention was to provide automatic balancing of the EL84 quiescent currents, without resorting to presets etc.
-

EL84_CCS_2.jpg (41.83 KiB) Viewed 3372 times

-
The Zener prevents the Voltage climbing when the o/p stage is driven into class B, and is currently 13V. The normal level at no signal or low level A-B mode is 11 - 12V. DB2540 is a depletion mode MOSFET. R2 sets the current at 40 mA.

Hope that makes sense!

Laters .....

[ejam]

Mike H

R1 sets the current, not R2. Also why the super high value of R2 at 100Meg, effectively a short? Your grounding at point E makes R2 redundant.

[Nick]

Not sure I understand why you would use a pair, instead of just one set to 80ma. Also, that zener worries me. It just looks to me to be a way the valve could fail and kill the mosfet. If you want it to run in B, then I would think a resistor would be a better choice. C1 also worries me, I can see it making a great oscillator with the CCS. But not tried it, so will be interesting.

Allan Wright was a proponent of the idea, and this is a good set of info (also with a better use of the zener I think).

https://www.tubecad.com/2011/08/blog0210.htm

[Daniel Quinn]

I admire the people who work with such voltages and direct coupling of components . I'm however glad I've only got one arm/hand so I can't try , I'll simply look on with envy .

[Mike H]

Mike H

R1 sets the current, not R2. Also why the super high value of R2 at 100Meg, effectively a short? Your grounding at point E makes R2 redundant.

You're right, R1, not R2. R2 only exists to satisfy LTspice by providing a path to ground. LTspice does not allow floating nodes. Although I could have deleted it from the diagram before posting and avoided confusion.

Thanks Nick. We did have a discussion about it a few Owstons ago.

There are 2 CCS because the tubes are not a matched pair. They do measure differently. I want the quiescent current in each half of the OPT to be equal. What I have at the mo turns out to be very like the 2nd from bottom schematic in your link.

Roger your advice re the Zeners, but I didn't have any high power ones. EDIT - I'll look into this.....

No I don't know why the cap is 220 uF — I'll look into that. Maybe the value came from a published schematic somewhere?

Daniel — yes, which is why safety procedures must be applied!

[Mike H]

In other news, today I made the supply input filter board...

-

Filter board_5699.jpg (139.05 KiB) Viewed 3242 times

-
-

Filter board_5700.jpg (59.62 KiB) Viewed 3242 times

[Mike H]

Right so, I've decided the current source board can be reconfigured as follows...
-

Current Source Board V4.jpg (65 KiB) Viewed 3216 times

-

Based on .....
-

Push-Pull Cathode Bias Two CCS Amplifier.png (38.5 KiB) Viewed 3216 times

-
Biggerer Zeners on order from RS.

[Nick]

That's very similar to what I do with the DAC PP output stage. I would still use 1k or so gate stoppers on the mosfets though. Also my normal advice, those mosfets are (in my experience) very very easy to damage with static. That's why I use the IXY depletion mosfets.

[Mike H]

Thanks for that — well I've changed my mind again, very late last night and following your earlier advice, I made a dynamic executive decision and opted for Keep It Simple Stupid, so today I made up another board comprising of just cathode resistors and capacitors. Zeners could be added, but not got them yet, an order is in with RS, but it doesn't need those to work.

Another perturbing and persistent issue was, LTspice doesn't like the separate CCS's, and keeps pooping out with errors. A single CCS, or resistors & caps, it likes those. So that also influenced my thinking.

I also reduced the caps to 47 uF, don't know where the 220 uF came from. Similarly, the 1.8k resistors feeding both EL84 screens. I must have been looking at another schematic from somewhere?

I changed the 1.8k (R13 in schematic) to 100R, LTspice likes that too and can show a discernible increase in output power. It's just short of 30W peak before clipping. So we're going with that.

Pix: 1. what I'm now calling the bias board, Zeners could be added later but are not vital...

EL84 p-p bias board_5702.jpg (119.59 KiB) Viewed 3124 times

-


Board circuit is...

EL84 p-p Amp Bias board V5.jpg (31.3 KiB) Viewed 3124 times

[Mike H]

Pix 2. What the innards look like at the moment...
-

EL84 p-p Amp 006C-2.jpg (160.27 KiB) Viewed 3124 times

-


Schematic as of today — values as what's actually soldered in...
-

EL84 p-p Amp 006C.jpg (137.58 KiB) Viewed 3124 times

-


Onwards!

[Mike H]

Could this be the last posting of the schematic? Place your bets now ...

Values as what are soldered in at the mo.....
-

EL84 p-p Amp 006C.jpg (140.66 KiB) Viewed 2926 times

[Mike H]

All resistors carbon film except –
R10, 11: carbon composite.
R5, 13, & bias board: Rapid Electronics PRO1 (1W) or PRO2 (2W) metal film.

Electrolytics: Panasonic radial 450V or 100V low ESR (and bias board).
C2, 4, 5: ClarityCap audio grade polypropylene.
C7: polystyrene film.
C8: 1kV ceramic disc.

Valves –
EF86: Teonex and Chelmer Valve (CVC)
EL84: JJ
Valve bases: B9A ceramic.


I'm a bit surprised that there was more work recommissioning this than I envisaged.

In the end, it's just a basic common-or-garden, cooking EL84 push-pull, no exotic bits. I did have more ambitious ideas originally but when I showed them to LTspice, it threw its toys out its pram and wasn't having any of it. So I thought OK I can take a hint...

Cue by-the-book (or datasheet) input stage and a concertina phase-splitter. Simples.

Last thing I now need to change the heater supply for 12V — been putting it off.....