Mullard 3 valve 3 watt amplifier circuits

[flyingtech55]

Further research reveals that the ME1400 may be a more suitable version of the EF37A as the first stage operates in 'starvation mode'.

Tim

[barrymagrec]

Back in the sixties when I was still at school I built a 3-3 using an EF36,KT61 and 5Z4 based purely on their free availability. O/P transformer was some HMV / Marconi early fifties unit. Worked a treat.

[bikerhifinut]

I've built this circuit, it is very simple and works a treat.
No doubt it will have shortcomings but it does work.

http://rh-amps.blogspot.co.uk/2013/0...sion-2_26.html

You can substitute the LM317 CCS circuit on the EL84 Cathode with a 270 ohm resistor and the zener diode screen grid feed with a 2.2k resistor, mine runs this way. I don't think the CCS offers any real advantage. I'd also use 47uF on the cathode bypass on the output valve.

Mine is totally silent in the background, no hum and no hiss. I did get a bit OCD about heater wiring and earth routing, and used a 10H choke in the HT supply.
It is also silicon rectified, I do not buy in to the idea that somehow a valve rectifier "improves the sound". Smooth DC is smooth DC............. End of.
Andy

[frankmcvey]

I do not buy in to the idea that somehow a valve rectifier "improves the sound".

You're probably quite right as far as hifi goes. I think guitarists like actually quite like the saggy aspect of a valve rectifier in their amps, in that when you hit a power chord with everything cranked, the HT sags; then as the note is dying away, the HT starts rising towards its quiescent level, increasing the amplification of the dying note through the amplifier chain, thereby giving it a longer sustain.

[Diabolical Artificer]

The EF37A has an OP of around 20v with a THD of 0.07% or something daft (can't find notes) if you use the R values off the datasheet, EG HT 200/250v, Ra 220k, Rg 680k, Rk, 3k9 triode strapped. It is running at about 8uA (starvation mode?)

I The 6V6 is the octal "equivalent" ish of the EL84., but you could use a QQV03-10A. The current thang is to use triode strapped 807's, KT66's etc, poor mans 300B

A..

[Synchrodyne]

To place the Mullard 3-3 in a proper context, perhaps we should avoid conflating latter-day single-ended valve amplifiers, which as RW has said, are essentially a “fashion” item with earlier era attempts to design and build single-ended amplifiers with above-average performance, perhaps edging towards the high-quality zone, and with simplicity in mind. The Mullard designers surely had no doubts that for the best quality and for reasonable power outputs, push-pull outputs were preferable, but also saw a place for a simple single-ended design that within its inherent constraints, was as good as was reasonably possible, and “robust” enough to be built by home constructors with predictable results, hence the 3-3.

To some extent, “fashion” played its part, albeit in a more muted way, even “back in the day”. Post-WWII, conditioned by 1930s thinking, it was the norm that higher quality amplifiers would be of the push-pull type, including those fitted to the upper end mass-produced radio receivers and radiograms. For the highest quality work, triodes or triode-strapped tetrodes/pentodes were de rigueur, although tetrodes/pentodes as such were acceptable for domestic radio receivers/radiograms and PA work. Negative feedback (NFB) was making an entrance and was accepted for its benefits. Thus, for example Williamson and Leak used triode-strapped KT66 push-pull outputs with NFB. Soon, not to have NFB would be seen as a negative. Baxandall proposed a tetrode-based amplifier in 1948, using a 6L6G pair, but this had little acceptance, apparently simply because it used tetrodes. Leak also offered the TL25 with tetrode-connected KT66s, but this was aimed more at industrial than domestic applications.

In this period, and into the 1950s, Lowther offered, at the lower end of its amplifier power output range, units based upon single-ended triode (PX25) circuitry. Above that push-pull PX4 and PX25 outputs were used, with push-pull pentode-connected EL37s for “industrial” applications. One assumes that the single-ended triode design was seen as a simple and economical way of obtaining 5 watts output, enough for some of the Lowther-Voigt corner horn loudspeakers, and also that it was capable of good enough performance that it was not put to shame by its larger push-pull brethren. In an era when push-pull was essentially sine qua non for hi-fi, it would not have been sold because it was a single-ended triode, but in spite of it.

In 1954 the ultra-linear (UL) circuit arrived in UK practice, and within a couple of years became the new norm for the highest quality amplifiers, displacing the triode. Perhaps because UL required the use of tetrodes/pentodes, these became somewhat acceptable even when not used in the UL mode. The original Mullard 5-10 circuit was not UL, that refinement being added later. However, the broadly similar Osram 912 of the same time was UL, and even GEC’s “Junior” design, using an LN309 pair, was UL.

In that push-pull UL era Pye offered the Mozart amplifier, which used a single-ended EL34 in the partial cathode loading UL mode. Not being push-pull probably meant that there would be some sales resistance, but one assumes that Pye, with push-pull UL designs already under its belt, saw it as an economical solution to providing 10 watts at a performance level appropriate to its price point, and probably at lower cost than say an amplifier using an EL84 push-pull UL pair. A very simple design with as few valves as reasonably possible may have been required to facilitate the small case size. For push-pull valve amplifiers, the unit cost per watt curve probably had a steadyish negative slope from somewhere around 10 to 12 watts and upwards, but might well have kicked up quite steeply at lower power outputs, leaving a zone where simpler well-engineered single-ended solutions had a place.


Cheers,

[ms660]

Originally Posted by Diabolical Artificer

The EF37A has an OP of around 20v with a THD of 0.07% or something daft (can't find notes) if you use the R values off the datasheet, EG HT 200/250v, Ra 220k, Rg 680k, Rk, 3k9 triode strapped. It is running at about 8uA (starvation mode?)

I make it 0.8mA (800uA)

Lawrence.

[Diabolical Artificer]

A well written short history of commercial valve amplifiers Mr Sync.

Your right Lawrence, knew it was 8 something, I was typing from a suspect memory. Anyhoo, it's off the mutual characteristics graph.

A.

[kalee20]

Originally Posted by G6Tanuki

Only 3 Watts from a single-ended EL84 though? I'ts significantly underachieving! With the right voltages you can easily get 5W from an EL84.

You can, yes.

The 3-3 wastes quite a lot of power in the EL84 cathode resistor, necessary because it's direct-coupled from the preceding EF86. The DC NFB to the EF86 screen-grid is lovely. So, while not efficient, it is a very elegant circuit. But to get 5W, the HT rail would need to be raised another 50V or more. By contrast, the ECL80 push-pull circuit has quite a low HT voltage.

Originally Posted by Synchrodyne

...The Mullard designers surely had no doubts that for the best quality and for reasonable power outputs, push-pull outputs were preferable, but also saw a place for a simple single-ended design that within its inherent constraints, was as good as was reasonably possible, and “robust” enough to be built by home constructors with predictable results, hence the 3-3...

For push-pull valve amplifiers, the unit cost per watt curve probably had a steadyish negative slope from somewhere around 10 to 12 watts and upwards, but might well have kicked up quite steeply at lower power outputs, leaving a zone where simpler well-engineered single-ended solutions had a place.

Yes, I'm convinced that the 3-3 was designed just to show what could be achieved by a simple single-ended circuit using the latest Mullard valves. They did it because they could. I am sure the push-pull ECL80 amplifier is inherently capable of even better results, if NFB is introduced, and if the cost budget is set to be the same as the 3-3, but input sensitivity would be nothing like as good (something which has not been considered).

Synchrodyne's cost-per-watt is a good figure of merit, and associated with this is the cost of the magnetics! The mains transformer will be very similar for both 3-3 and 3W-ECL80 amplifiers (the 3-3 has a slightly greater power draw). But the 3-3 output transformer will have to be quite a bit bigger as it carries DC and needs air-gap, whereas the ECL80 amplifier doesn't. On the other hand, winding time will be greater as it has two sections to the primary - and that's before pri/sec interleaving! Nevertheless, I'd still predict it will be cheaper. Use the money saved to add a few extra components... which has the greater potential? I'd still say the 3W-ECL80.

[Boater Sam]

Following on from yesterday's new knowledge, how about a transformer-less
single ended push pull totem pole output with EL84 & EL86 (or UL84/PL84)?

[Herald1360]

That would transfer output TX cost to a speaker made of unobtanium.......

Jumping back a bit, the 6V6 isn't a good swap for EL84, EL33 or KT61 are closer characteristics wise. Pricewise it might be better than the other two, but new EL84s are still being made.

I've got some ME1400s, though......

[Radio Wrangler]

The 6V6 does seem to have a very good life expectancy, while the EL84 seems less long-lived. Maybe those nasty brown stains make them all about to peg it?

David

[ms660]

I reckon it's all down to power per cubic cm.

Lawrence.

[kalee20]

Yes. The EL84 is very hard-working for a B9a bottle at full bore. Whereas the 6V6 is bigger so loses its heat better. And it has a 2.8W heater rather than 4.8W.

In the Mullard 3-3, the EL84 should have an easier life though, as it is not run full whack (as far as I know).

[G6Tanuki]

TBH I really never understood why the EL84 became popular when there was the 6BW6 [a 6V6 in a B9A envelope] available.

A pair of 6BW6 in push-pull gets you 30W of audio. And they put much less strain on the heater-supply.

[The EL90/6AQ5 - a 6V6 in a B7G bottle - is pushing things a bit though - they need good ventilation if they're not going to suffer premature death]

[CambridgeWorks]

A pair of 6BW6 in push-pull gets you 30W of audio.

Just saw your quote from r-type site, 30W as in class AB1.
I certainly wouldn't push my luck though! Back in 60s, they were used to modulate a 160M tx or at a push, as a PA stage.
I agree with your other comments regarding EL84 though. Seems to me to be "the latest modern audio output valve of the moment" decades ago, hence used in great numbers. I agree with David, the stains within EL84 (they remind me of tar!) make me think life expectancy is not good. However, ebay can sometimes show such examples that allegedly are "NOS"! I will say no more!
I know, any such EL84 that I have like that are treated with suspicion and usually my contempt is justified.
Rob

[GrimJosef]

Originally Posted by G6Tanuki

A pair of 6BW6 in push-pull gets you 30W of audio.

Yes, the details are here https://frank.pocnet.net/sheets/184/6/6BW6.pdf.

I do so want to build an amp to do that. The 6BW6s actually need to run in class AB2 so the drive would have to be as stiff as a very stiff thing - I'd aim to use the triodes in an E182CC, or possibly a 5687, as cathode followers to feed the grids. The phase splitter might be a Radford/Bailey one based on an ECF82. A single triode (half an ECC83 ?) in front of the whole lot should give enough open loop gain to allow NFB to tame the inevitable distortion (the datasheet says 7% at 30W from the 6BW6s). To keep everything compact I'd use a solid-state rectifier for the output stage HT and an EZ81 to power the rest, including the 6BW6 screen grids. That way the slow warm-up of the EZ81 would protect the 6BW6 cathodes from the 'instant on' anode volts. Ten B9A valves in total would deliver 30WPC in stereo .

Cheers,

GJ

[G6Tanuki]

When I built 'juicy' 6BW6/EL84 push-pull guitar amps, I used a 12BH7 as the driver - it can deliver a bit more 'oomph' than the more-traditional double triodes.

And I used BY127-type diodes for the HT, and a small 24V transformer/rectifier for the negative bias on the output-pair, allowing the cathodes to be connected direct to -ve, which is better than using cathode-resistors when the cathode-current will be varying with drive,

An ECC83 as preamp, a pair of 12BH7 as drivers, four 6BW6: that's only seven valves for 30W-per-channel.

[kalee20]

Originally Posted by G6Tanuki

TBH I really never understood why the EL84 became popular when there was the 6BW6 [a 6V6 in a B9A envelope] available.]

Probably a triumph of good marketing rather than good design!

Originally Posted by G6Tanuki

The EL90/6AQ5 - a 6V6 in a B7G bottle - is pushing things a bit though - they need good ventilation if they're not going to suffer premature death

That one really is a shocker - gets stinking hot. Unlike the EL84, it is a beam tetrode. I did make a push-pull amplifier using a pair, works very well, but the valves are totally exposed so cooling is good.

The EL84 is a bit more sensitive than the 6V6 / 6BW6 / 6AQ5 though, so you're not strapped for gain in a basic circuit.

[Synchrodyne]

30 watts is an order of magnitude above where we started!

Back to 3 watts, the Jason JSA2 was another simple amplifier in this class, with an ECF80 and an EL84 in each channel. It was described in Radio Constructor for 1959 February.

The GEC Junior (Radio Constructor 1956 November) had four valves for 5 watts, but one of those valves was in the control section. The main feedback loop encompassed just the LN309 pair. So, the amplifier proper had three valves including the rectifier.

Re the Mullard ECL80 push-pull design, would it be possible to apply NFB in shunt with the input, via a resistor from the appropriate side of the output winding? Then I think there would need to be a series resistor on the input side. Maybe this would not be of high enough value for say ceramic cartridges, but perhaps enough for modern sources? Power amplifiers that overall look like inverting feedback amplifiers seem to be rare. (Although from the solid-state era, the Quad 303 comes to mind.)

Whilst the ECL8x valves were strongly associated with record players, the ECL82 appears to have had some hi-fi applications, including in the U-L mode, the Radford ST7 being an example. So perhaps it was better than its common use suggested. With the ECL86 I think that Mullard may have had hi-fi applications in mind from the start, additional to domestic TV receiver applications. It seems to be been adopted quite early on by some of the hi-fi equipment makers.

Was the Mullard 3-3 circuit ever used as the basis for a commercial amplifier? The 5-10 and 5-20 certainly were, and I suspect that the Osram 912 might have been the basis for the Dynatron LF10. The book “Audio! Audio!" lists one or two amplifiers with single-ended EL84 outputs, but it is not apparent that any were of the Mullard 3-3- form.


Cheers,