EL84, Balanced Input, Push-Pull Amplifier

[Paul Barker]

Yes the harmonic equaliser of the WE ony differs from what you have in terms of ratio of upper and lower resistor selected. Your ratio is way different from W.E.. That is why I have gone with very low resistance to work with for balancing the currents on the output transformer, which I have yet to establish whether it would be sufficient, allowing me the adjustment of the W.E.H.E. (Western Electric Harmonic Equaliser) with a variable tap along the cathoe resistor. In an ideal world one would seek the help o a spectrum analyser to tune it in. But I imagine the W.E. voltage devision of about 10:1 would be in the right ball park and adjustment to sound choice would be sufficient information to "get it right". W.E. used an equivalent value cap of 16uF because they series connected to achieve the voltage capability, in our time we have voltage tolerant caps a plenty. I don't haver a clue how they arrived at what they arrived at either in cap size of voltage devision of the resistor, but I am certain they didn't just guess it.

You may have noticed that my choices have compromised my operating position as total cathode resistance is now 312 ohm, but the text book class A cathode resistor value should be 250 ohm. Basically I lifted the operating point of 285v -19.5 bias with current of 78mA via a 250 ohm resistor. But as my power supply is in units of 48v my operating point will end up with lower anode voltage than spec, which will effect the current pulled across the resistor so the bias may not alter much and the current will be less than stock operating position. But the only reason I am altering the operating position from stock (which was chosen many years ago by the manufactureres for best power output in class A PP) is to facilitate W.E.H.E.

The other original plan I have forgone is fixed bias. With fixed bias and only balancing pot on the cathodes I could tune in the exact operating point. But that would remove the option of employing W.E.H.E.

I am also interested in trying the Quad II cathode feedback. But I would have to stumble over a pair of Quad II output transformers cheap first.

[Cressy Snr]

Cheers Paul,
I think as per usual, mine is complicated by the fact that my bypass cap goes to B+ instead of ground So I have a WE harmonizer, bypassed by a WE connection. I'm injecting common mode something or other into the still point at the centre of the OPT Primary.

Whatever is happening, I can't deny, the result is very nice.

[Paul Barker]

Well I don't have a clue what is going on, so I am just copying. I hope it works.

I imagine it is possible to get a satisfactory blend of methods as you have found.

[Paul Barker]

So question Steve? As you have yours built, what is the ratio along the balancing pot which provides balance on each channel? Clearly it depends on the matching of the valves, but it would be useful to know anyway how a physical example has turned out.

[Cressy Snr]

Hi Paul I have 47 to 53R on one channel and 48 to 52R on the other, with everything nulled out; so there is an average 2R5 difference to each of the ends of my 100R balance pots.
So the pots are not far off their centres.

I bought my JJ EL84s at different times as two matched pairs, so providing your valves aren't too far different, I would think you'd get away with the 25R pot.

Here's a shot of the audio circuitry with the two aluminium cased ultrapath caps bolted through the rear panel.

There's more holes in this box than a swiss cheese, so the time is fast approaching for a request for a nice box from our Ant.

Very simple circuit, for a PP amp

[Paul Barker]

It's a tidy build non the less.

[Cressy Snr]

Was experimenting yesterday and today with 5687 subsitutes with interesting results. 7044s were disappointing in this particular application as they were horribly microphonic and far too bassy for the system as a whole. This was odd as they used to be my absolute fave driver valves for DHTs, but driving EL84s was not successful. I've no doubt that they were easily capable of driving the output stage but their tonal characteristics were incompatible with it.

I decided to blow the dust off a pair of E182CC, that Will had given me; years ago now it was. I had never got on with these valves as they had sounded a bit thin and strident with 2A3s and the TJ mesh plate PX25s I used to have.

With PP EL84s however, they seem to be a match made in heaven. Lovely tone, relaxed and liquid sounding, with nary a hint of stridency. Just shows that in the right circumstances, a valve you had given up on, can reveal itself to be not the dog you thought it was.

My four new ML6 as two diff pairs, driving push pull KT61/6P25s would be a larf I reckon, but that's for another time.

Our Ant is on with the permanent housing for the amp.
It will look something like this example from Trafomatic Audio:

http://www.trafomaticaudio.com/premise.php

[Cressy Snr]

The design continues to be refined, more blue labels signifying changed component values.
It's the input/driver differential stage that is getting all the attention.
The output stage is fine as it is for the moment.

The anode resistors have been dropped to 10K, to increase the current through the driver valves.
I would like to lower the value of the cathode resistors to bump the current to around 18mA per side(current is now 13.4mA per triode)
but if I did it now, the anode resistors would be on their dissipation limits, not good for reliability.

When the amp is rebuilt into our Ant's box, the driver stage cathodes will be treated to a lovely Takman resistor each.
Their anodes will get Mills jobbies. This will allow the driver standing currents to be increased.
Output stage cathodes will also get a set of Mills 12W wirewounds to replace the cluster of parallelled up, 5W metal oxides
The amp, really is sounding beautiful now. The tone is smashing and listener fatigue is non-existent.

[Cressy Snr]

So in the end I decided to stop farting about and make the amp as simple as possible.

There were for me, too many variables at work, with the ultrapath cap and the balance pot and the harmonic equaliser all moshing around together with the audio signal.

So everything around the cathodes of the output stage has been stripped away, leaving just the essence of the design in place.

The ultrapath cap has been left in place and whaddya know it still sounds as well as it did with all the other clutter around the cathodes.
The whole thing is now very hard wired, with solid copper bus bars connecting the EL84 cathodes together.
More copper bus bars run from there to the ultrapath caps and a great big pair of 270R aluminium clad cathode resistors are connected by more copper bus bars to the ground bus.



I stopped before I disappeared up my own backside. That feels good.

That I have been to get this quality of sound out of a quad of cheap EL84s is reward enough for me.
I think in the case of push-pull, with ultrapath caps, IMHO attempting WE harmonic eq is unnecessary.

I have two good amps now, this EL84 and of course the much beloved Mazda amp. Here's to the next one

[Neal]

Good one Steve look forward to hearing it at Owston!

[Paul Barker]

I can't blaim you at all Steve.

I was only thinking myself about the whole balancing push pull output stage conundrum.

the idea of balancing the valves by vertue of the cathode wiper is flawed. All you can hope to do is balance a static state. Once the music plays you have no idea from one level to the next how well balanced the pair may be.

All you can really do is buy a curve tracer, match your valves by printing the trace on tracing paper and overlaying pairs of valves until you have best match. In the amp make sure the output transformer is wound on a split bobbin exactly equal, side for side. In the cathode a shared resistor. That is the best you can ever hope to do.

Then you know at one point in the valves life you have balance.

As for the WE connection or the ultrapath, or the issue of cathode bypass cap?

More problems of the same making. At different levels and for different distortions different methods will benefit.

Maybe plumb for one and live with it, have a switch to eliminate it and make sure it brings net benefit to the music type and mood you are in at that time.

Probably because my transformers are imbalanced I shall statically balance my valves through the cathode trimmer. But it may not make a blind bit of difference when the music plays.

Even in power supply transformers a perfectly balanced split bobbin makes a difference. What I mean is a centre tapped secodnary needs to be balanced a split bobbin with equal turns and dc resistance on each half. I wound one for Darren back in the day and he said it made a huge difference to the sound of his amp.

though with your super power supplies you would probably even out the vageries of an imbalanced transformer, Darren's old fashioned power supply approach (which I taught him) would yield much fruit from every subtle change. His amps always sound organic.

Onwards and upwards waving the simplicity flag!

Regarding my 6v6's. No iodea if they are a match, when I am wealthy (a day which is getting further away from a life which is getting closer to it's end) I shall have a tube curve tracer and match my valves.

Maybe I succeed anyway with push pull, or maybe I dust off the 211's. Or maybe I build the OTL again.

[Cressy Snr]

Cheers Neal/Paul
For me it is no coincidence that two of my best sounding amps are, in terms of their audio circuits, the simplest. There's not much to either this one or the Mazda amp.
Mind you, Steve S has been going on about the virtues of simplicity for more years than I care to remember

[Nick]

All you can really do is buy a curve tracer, match your valves by printing the trace on tracing paper and overlaying pairs of valves until you have best match. In the amp make sure the output transformer is wound on a split bobbin exactly equal, side for side. In the cathode a shared resistor. That is the best you can ever hope to do.

Play sine wave through amp, into spectrum analyser, adjust for least 2nd harmonic. That way you are balancing the valves and the transformers.

Not saying you need to do it, its just simpler than a valve tracer. The valve tracer is better, but a lot of pain.

[Paul Barker]

Thank you that sounds like a plan.

Presumably Andrew's distortion meter would do the same job. We could all tune our push pull amps once a year.

[IslandPink]

Yes , then the next step would be to put in two slightly unmatched valves, get a bit of 2nd harmonic into the mix, and enjoy the improved tone !

( - which is after all, what the harmonic equaliser does to a small extent )
Oops, maybe I'm not welcome at this party now ...