EL84, Balanced Input, Push-Pull Amplifier

[Cressy Snr]

They seem to like GU50s on DIYAudio.

OTOH have you seen a KT120 in the flesh?
Built like brick outhouses, they make KT88s look like toy valves

[Paul Barker]

Yes indeed tough construction normally sound good. No I haven't seen a KT120. With luck you may have it built in time for OWsten. When is that anyway? Even if I don't get time to change this amp it is bringable as it is, since it will show what it is likely to sound like when the Rf isn't there to anoy.

[Ali Tait]

15-16th June.

[Paul Barker]

ta

[Cressy Snr]

Here we have the same amp in an ultralinear configuration:


This is the better sounding amp of the two that use the ECC82 as the VA/driver.

Having read up extensively over the past week the original Hafler/Keroes paper concerning their development of the UL circuit, after the Blumlein patent had expired in 1953, I realised what were the faults with my original UL EL84 circuit.

I had wrongly omitted the cathode bypass capacitor to the output pair.
This was wrong for two reasons:

a) the amp is class AB and omission of the cap is not good for intermodulation distortion when the amp enters class B at high output levels.

b) the cap is needed when operating the output pair as ultralinear, because if it is left out, you get the degenerative feedback around the cathode/grid circuit raising the valve's output resistance, which in turn destroys the benefit of the lowering effect on the output impedance caused by the negative feedback around the screen/anode circuit. This lowering of Zout is a feature of UL you want to preserve, otherwise, why bother?

What you actually end up with, by leaving out the cap is basically an ultralinear output stage that behaves like a pentode in terms of its output resistance, which tips up the balance and destroys the bass unless feedback is used, which you wouldn't have had to use if you'd fitted a bypass cap in the first place
And there I was whining, earlier in the amp's development, about how UL was nowhere near as good as triode or pentode operation; making assumptions on the basis of a flawed information.

So what I have now is a proper ultralinear output stage that has most of the lowered output impedance a triode produces, but with the added benefit of the power of a pentode, which is, after all what the point of UL is.

No additional feedback is required as the inherent feedback between the screen and the anode does the job of flattening frequency response and damping the speaker perfectly well without it.

The sound is clean, low in distortion and fatigue free, with oodles of tonal loveliness and a sweet, detailed presentation; a presentation that works beautifully with classical, jazz, soul, rock and R&B.

[Paul Barker]

You could use seperate unby-passed cathode resistors to allow incursion into class B but you would have the higher impedance to deal with, or you could use fixed bias (or what the dynaco boys like, a little of each).

[Cressy Snr]

Aye could do that, but I'll be incorporating all these ideas into the KT120 amp rather than doing anything further to the EL84. The poor thing could do with cutting some slack

I've just had a horrible thought:
on a previous iteration of the amplifier, when it had the input transformers, I had a distributed load topology on the output stage, plate-to-grid feedback and no bypass cap; that's three local feedback loops, all at the same time.
Its a wonder the output stage didn't disappear up its own backside. Thing is, it sounded fine to me.
Just shows what a finely tuned ear I have (not). I'd best get back to my Max Bygraves box set

[Cressy Snr]

By eck I cut drilled filed and bled this morning and replaced the ECC82s on the front end with a lovely pair of Brimar black glass 6SN7.
They dwarf the EL84s which is fun.

I'm just playing around with the amp now because I can.

I'd like to fit my prized 13D1s, which are the same valve as the sought after Brimar brown base 6SN7 apart from the 25V heaters. I have a 25V toroid left over from the 13E1 project that I'm sure will shoehorn into the chassis.

[Cressy Snr]

Update on the amp:
I couldn't get the 25V toroid into the chassis, it was simply too big. However all was not lost as I found a 30VA 2x12V toroid in my oddments box. Putting both windings in series gave 25V at each base with the valves plugged in. The transformer fitted perfectly into the bit of space left in the chassis. It sits neatly next to the main power toroid and heats the drivers separately from the rest of the valves.

When I had the 13E1 SE amp, I had had a tantalising taste of the beauty of these driver valves, but unfortunately, the 13E1s' biasing became uncontrollable soon after, making the whole thing unusable, so I never really got to appreciate just how good these drivers are.

They have languished in their boxes until now. Their sound is easy going and natural.

[Cressy Snr]

These will make great drivers for the KT120 project. Really great drivers.
Shall I shut down development on this amplifier?... probably.

[Paul Barker]

Probably.

[Paul Barker]

Probably.

[Cressy Snr]

OK development has halted on this particular project.
Here's the format it will be coming to Owston as:



Those big micanol base 13D1 (25SN7) valves don't half glow lovely

Whether push-pull or SE, that warm orange glow is one of the things that make this valve business such a rewarding pastime.

[Cressy Snr]

Seeing as I absolutely adore the EL84 and the sound it makes, I am about to start tomorrow on a new balanced push pull chassis for 12W worth of the buggers.

So tomorrow, it's a 500 x 500mm sheet of 1mm steel, a rubber mallet, a vice, some G clamps, punches and drills, primer and paint.

A bit of bending, panel beating and painting and hopefully, we will end up with a chassis, something like this:



OK this is for KT120 but you get the general idea

[Dave the bass]

Shoddy!