Parallel single ended valve amps, any experience?

[Gabe001]

I have a few ideas in my head for a future amplifier build, and I was wondering if I could pick your brains regarding viability of the lead contender.

As some of you may know, I have a push-pull el84 amp I'm happy with, so I'm looking for something different. In my head, this means a valve amplifier that 'editorializes' the sound rather than just reproducing it. Namely, a single ended triode design with a "warm" second harmonic. Now, I'm not going to spend silly money on a DHT, so I'm happy to use el34s or 6l6gcs in triode mode for a "near enough" experience.

The problem with triode amps, as you know, is the power output (barely 4-5w), so I am reading about parallel tubes. 2 triode el34s in parallel, per channel, with decent anode voltage should yeald close to 10w, and with a smidge of global neg feedback, a decent damping factor and hopefully inaudible hum and noise.

In terms of the OPT, I know that it needs half the impedance of a single valve, twice the current rating and the ability to handle twice the watts.

As such, is any OPT suitable as long as the above criteria are met, or does it need to meet other special considerations?

Also, has anyone built something similar? My brain keeps telling me to go for push-pull triode, but my heart says otherwise.

Thanks for your comments
Gabriel

[GrimJosef]

A single-ended OPT has a constant (DC, if you like) magnetic field in its core. In a perfect world it might be designed to cope with that, but single-ended amps are sufficiently imperfect in other ways (some of which you might actually like) that it's not unusual to find that the transformer also just has to 'live with' the DC field. One solution is to make the transformer physically large, so at least the field strength isn't high enough to be troublesome.

On the plus side the low anode impedance means the OPT will have a relatively low turns ratio, which makes design easier.

In answer to 'has anyone built something similar ?', Audio-Note UK make a stereo amp with four EL84s in called the OTO https://www.audionote.co.uk/oto. Some models have the L and R pairs of valves in push-pull, others have them in parallel SE. I've had a few in for service of one sort or another. The push-pulls measure better than the SEs. But I have to say, 'measurist' that I am, that each time I've heard one of the SEs I've been surprised at how good it sounds. The other surprise that you get when come across an OTO, though, is just how heavy they are. A lot of care goes into the design of the output transformers. And a lot of iron and copper does too.

My advice would be to give parallel SE a try, if only to work it out of your system. If you buy decent transformers and take care of them then you can probably sell them again for a reasonable fraction of their original price if you decide you want to convert to PP.

You also mention hum. Controlling 50Hz hum is the same in SE amps as in PP ones. Controlling 100Hz hum is about removing ripple from the HT rail. Brute-force smoothing with large Cs, and a good-sized L if you want to splash out on one, works perfectly well.

Cheers,

GJ

[Gabe001]

Thanks GJ. I suppose if the output transformer cannot handle 2 valves I can convert to a standard se amp and use an 8ohm speaker on the 4ohm secondary.

My first single ended amp had good hum control with a CRCLC so I'll apply broadly the same principles. Iron from China is quite cheap. Chokes and mains transformers I'm not too fussed, it's the OPT I'm worried about. I'll look for an output transformer weighing at least 2kg and perhaps give it a go

[GrimJosef]

In the end your room for manoeuvre depends on the performance you want. Output transformer design is a matter of trade-offs. Among the things the designer will consider are the 'stray' and 'leakage' capacitances and inductances and these will be chosen to work with the external circuit impedances. Running a transformer with twice its design circuit impedances may not get the best out of it. It might be good enough though.

Another thing to worry about is phase-shift in the transformer. This can limit how much negative feedback and/or how wide a frequency range you can use.

Cheers,

GJ

[Radio Wrangler]

It isn't just the weight of your output transformer. Transformers for single-ended amplifiers usually have to be gapped to control the flux density created by the DC ampere turns of magnetomotive force being applied. The gap also brings down (and stabilises) the amount of primary inductance, but this isn't too bad because the lower Z that triode anodes are usually run at requires less L for the same low frequency roll-off point.

The methods of design and the resulting item for good quality push-pull and single-ended output transformers are quite different. Treat the single ended ones as rather specialist.

Not only do you have the overall non-linearity of the valve, you are also running your transformer offset a long way on its core's B/H curve. So it distorts a lot too. Differently to the valve.

David

[Gabe001]

"Not only do you have the overall non-linearity of the valve, you are also running your transformer offset a long way on its core's B/H curve. So it distorts a lot too. Differently to the valve."

I don't quite understand David. Does this statement apply to all single ended transformers, to paralleled tubes, to using triodes, or to using a higher impedance speaker than intended to increase the reflected load?

[ian_rodger]

Gabriel,

I have a complete transformer set (mains + 2 x o/p) for the World Audio Design 2A3 parallel single ended amplifier not doing much here. From memory the o/p transformers are 2k primary and 8 ohm secondary but I'd have to check. These are high quality parts. Presently I have them in an own design amp but would consider selling. PM me if interested.

[Radio Wrangler]

All single-ended transformer output amplifiers The valve needs to work in Class-A, so there must be a standing (quiescent) current big enough that the valve never cuts off even on + going peaks of anode voltage. This creates unidirectional DC magnetisation in the core. So the amplifier operates with the core far offset from zero flux as its rest condition.

Operation, power for power, goes further into the end curves of the magnetisation characteristic of the laminations material.

So the core is giving even-order distortion as well as the valve is.

David

[Electronpusher0]

Originally Posted by Radio Wrangler

So the core is giving even-order distortion as well as the valve is.

David

Ah, that magic "valve sound"

Peter

[Diabolical Artificer]

I've built parallel PP amps, there's not much to it though several things to bear in mind. In my case I used 6 EL34's so Rg1's value has to be reduced accordingly, in your case halved or use two of the same value in parallel. The driver needs to be relatively beefy as you have twice internal capacitance, here you can just strap a double triode together, win win. As regards the OPT it's pretty straight forward as mentioned in terms of the usual parameters.

Lastly don't forget the 2C34 and it's ilk which are very useful triodes and still available for a good price.

Andy.

[Valvepower]

Hello,

Just picked up on this thread…

I’ve recently had experience with a parallel single ended (SE) amplifier, all be it in the amplifier for the Clavioline keyboard.

The original Clavioline amplifier used just one 6V6 in SE configuration without negative feedback. The reason for using the SE 6V6 without feedback is because it’s part of the sonic recipe of the Clavioline in order to get the sound of the instrument.

Later versions of the Clavioline used two SE 6V6’s in parallel which were subsequently replaced with two 6AQ5 again in parallel. This was done to get a tad more power.

Strangely, from what I’ve found they didn’t change the speciation/part number of the output transformer and it stayed the same part throughout the life of the Clavioline. I’ve checked this by inspection of early and later units along with the diagrams and parts lists.

I was able to check this on an actual units, an early unit with a SE 6V6 and also a later unit with two SE 6AQ5 in parallel. I measured the anode impedance load three transformers and they and they all measured the same.

Thinking tangentially, and not changing the impedance of the primary of the output transformer when using it with two valves in parallel; it will effectively increase the anode load over of that of a single valve…, this [off the top of my head] will give higher second harmonic distortion at the cost of the maximum available output power using two vales in parallel over that of using a lower anode load with two valves in parallel…

Maybe this added to the sound...? As this is musical instrument amplification, it’s as much about sound as is the technical parameters.

Finally, this is from an article I’ve done about the Clavioline…
I injected 500Hz sine wave and monitored the speaker output with an oscilloscope and I was getting 6.2watts at 8.3% THD into 3Ω. Alright, it was not the greatest looking sine wave with 8.3% THD, but seeing the amplifier is voiced with a high level of distortion this appeared acceptable. With an HT of 334V and two 6AQ5 in parallel I expected to see slightly more output power, but I feel the anode load of 5200Ω could be affecting this. Maybe this was done to get the ‘sound’…

I’ve attached two pictures of the output transformer…

Sorry for the rambling nature of the post, but halfway through writing it, Caroline played Hawkwind, Master of the Universe off the In Search of Space LP – I just had to turn up the radio and take in the Space Vibes…

Terry

[Gabe001]

Thank you for your feedback.

Options for driver include a 6sn7 cascade which will provide a lot of gain and good headroom for some global neg feedback, or a 6sl7 mu follower which will provide a voltage gain of 50. The el34 in triode mode at 420v at plate will bias at around 34v and will need a good signal to drive it. I want to give it about 6-10 dB of feedback. Low enough not to have to worry about stability but enough to hopefully decrease noise to below 80db and improve the damping factor which according to my back of envelope calculations should be around 10 (equivalent to a Mullard 5-10)

The 6sl7 mu follower worked very well in my single ended el34s ulralinear amp but the el34 was running at 335v on the plate and therefore needed a smaller signal to drive it to full output.

Anyway so far these are musings,I haven't got far yet and I haven't modelled anything on Lt spice.


Ian, I've PM'd you

[kalee20]

Originally Posted by Radio Wrangler

It isn't just the weight of your output transformer. Transformers for single-ended amplifiers usually have to be gapped to control the flux density created by the DC ampere turns of magnetomotive force being applied. The gap also brings down (and stabilises) the amount of primary inductance, but this isn't too bad because the lower Z that triode anodes are usually run at requires less L for the same low frequency roll-off point.

That's certainly true, but at the small-signal roll-off point, the valve is being loaded with a lower impedance (71% of the mid-frequency value).

So wind up the volume, and you'll start hitting the threshold of distortion at a lower volume level.

I tend to aim for an output transformer roll-off at least an octave lower (half the frequency) of my intended roll-off, and then put the roll-off upstream in the amplifier so that the troublesome frequencies are sufficiently attenuated anyway by the time they hit the power stage. That way, the valve load doesn't drop below 89% of the mid-frequency value.

Originally Posted by Radio Wrangler

The methods of design and the resulting item for good quality push-pull and single-ended output transformers are quite different. Treat the single ended ones as rather specialist.

Definitely! A push-pull amplifier, with no standing flux, can be improved by exotic high-permeability alloys in the core, giving super-high inductance. And as long as there's sufficient turns to not saturate at the lowest frequency, it'll work well.

Whereas the effective permeability in the SE case is dominated by the air gap. Design methods are quite different!

If you use paralleled valves, it's worth giving each its own cathode resistor and bypass capacitor, to ensure current sharing as much as possible

[Gabe001]

Originally Posted by kalee20

If you use paralleled valves, it's worth giving each its own cathode resistor and bypass capacitor, to ensure current sharing as much as possible

Yes I read about this. The valves also ideally need to be reasonably matched.

The audio equivalent of selective foetal growth restriction syndrome

Thanks Terry for your post. I would have thought that the higher output load would decrease the second and total harmonic distortion, not increase it. I could be mistaken though

See table here and compare 2nd harmonic distortion for z of 3k Vs 5k for each tube at the same plate voltage

http://tubelab.com/designs/tubelab-sse/applications/

[kalee20]

Originally Posted by Valvepower

The original Clavioline amplifier used just one 6V6 in SE configuration without negative feedback.

...

Later versions of the Clavioline used two SE 6V6’s in parallel.

...

Strangely, from what I’ve found they didn’t change the speciation/part number of the output transformer and it stayed the same part... I measured the anode impedance load three transformers and they and they all measured the same.

Thinking tangentially, and not changing the impedance of the primary of the output transformer when using it with two valves in parallel; it will effectively increase the anode load over of that of a single valve…, this [off the top of my head] will give higher second harmonic distortion at the cost of the maximum available output power using two vales in parallel over that of using a lower anode load with two valves in parallel…

Originally Posted by Gabe001

Thanks Terry for your post. I would have thought that the higher output load would decrease the second and total harmonic distortion, not increase it. I could be mistaken though

I've been thinking about this! A well-matched output load (ie matched for maximum power) is arranged so that the positive half-cycles (on the output valve anode) go into clipping at the same signal level as the negative half-cycles. It makes the best use of the available voltage swing and the available current swing.

The positive half-cycles start clipping when the anode current just hits zero - it can't go below 0 no matter how much input is fed to the grid.

The negative half-cycles start clipping when the anode voltage hits the bottoming voltage (typically 25V). The valve can't pull its anode lower than this.

The clipping levels aren't 'hard' levels: as the valve approaches cut-off the Ia - Vg characteristic curves off progressively, so the output waveform will have a rounded top, not a sliced-off top. And the other clipping level - as the anode voltage gets lower, the valve becomes progressively less effective at pulling its anode down, so again the waveform is rounded, not sliced-off.

With both peaks rounded-off, the distortion is reasonably symmetric, therefore third-order (and higher, odd-order).

If you double-up on output valves, the quiescent current will be doubled. So that's double the reduction in current that's now available, as the grid goes negative and the anode goes positive.

But, the twin valves can't pull their anodes any lower on the opposite half-cycle. So at the onset of clipping, the anode voltage swing is still the same; the load impedance is still the same, so the output power is still the same. The difference is, the positive-going half cycle is now miles from cutoff, so there'll be no clipping here.

The clipping will thus be asymmetric, second (and higher, even-order) distortion, once the volume has been wound up such that distortion starts to rise.

[Valvepower]

Hi,

I was writing this before I saw kalee20 post and, I suppose what you said kind 'a matches with my train of thought that the ratio between the second [and all even] order distortion is going to change, and the even order distortion will be greater than the odd order distortion with two valves in parallel using the original high anode load.

I feel the original designers wanted to increase to power, so they connected two valves in parallel, but left the transformer load the same, so there would be a tad more power available using two valves in parallel, however, its driven harder in pursuance of this increase in power with a possible increase in distortion in pursuance of this increased power - if you see what I mean!

Maybe they wanted more power, but lowering the load impedance changed the sound too much [below clip], so they compromised on two valves in parallel however keeping the original load impedance? I wonder if this is reason for keeping the high [original] impedance of the output transformer in the Clavioline...

Again, I think I know what I mean

Terry.

[Valvepower]

Gabriel,

I can see what you mean looking at the table in your last post…
5K Load Second = 5.7% Third = 0.5%
3K Load Second = 7.8% Third = 0.4%

As regards making a parallel SE amplifier – and as already said – If you can get some sensibly priced output transformers it’s got to be worth a try. Roughly speaking and off the top of my head, an output transformer for use with one EL34 at 10 watts could be useful for two EL84’s in parallel – this transformer may be more easily obtainable at a sensible price.

The Clavioline is an extreme case plus its musical instrument to boot, so it’s not an orthodox application of a parallel SE amplifier.

I still have the amplifier from the ‘mule’ Clavioline, so I could get that on the bench and simply try it with a single or both output valves. At best it’ll need a re-cap to get it going.

As if I don’t have an enough things/projects to do

Terry

[Gabe001]

No need Terry, don't worry, I know you're busy enough especially with your nice haul from audiojumble! Im still a few months off from starting this project and have another amp to finish first.

I'll see if I can get the 2k ones offered earlier. If not, there's always China but I'm not sure I trust them completely with single ended transformers, especially for a parallel single ended design. Primary windings may do a custom one if necessary, but let's see. Usually with a bit of patience an opportunity will present itself.

Gabriel

[Avinunca]

I have a Border Patrol SE300b amplifier with a single 300b valve per channel. It sounds very sweet but is a little limited in power needing sensitive speakers. I have read reports that the company's parallel 300b version gains some heft at the expense of absolute fidelity.
Another place you might ask the question is the UK forum audio-talk https://www.audio-talk.co.uk/phpBB3/ They have an annual get together near Doncaster where they show off their latest creations to fellow builders.

[Gabe001]

Nice amp! There's one for sale on Facebook marketplace at the moment for £2k. Thanks for the link