My first push-pull vacuum amplifier

[Miguel Lopez]

Hi all

The title of this thread express my intention to build a vacuum push-pull amplifier, more specifically, a mono Williamson type amplifier. I'm just in the study phase as I have never built something like that, so I will not begin to cut metal very soon.

I want to use a couple of 807s as output valves in class AB, avoiding the grid voltage to go more positive than the cathode voltage (not sure if AB1 or AB2).

It is not my intention to built some beast which can deliver more than 20W.

My first question here is not related with the output stage but with the preamplifier and phase splitter. I've been looking at the circuit at the national Valve Museum which uses a pentode and a triode as phase splitter and an EF86 as preamplifier.

http://www.r-type.org/articles/art-097.htm

I have a couple of EF86 (sent to me by forum members) and several Soviet 6F1P triode-pentode valves. 6F1P is very similar to ECF80. Not sure if it would be suitable for this circuit.

I would like to hear your opinions regarding this. Will be worth to try this circuit?
Should I try anything more simple like a common long-tail or a concertina?

Thanks in advance

Simple is best in my book, have you seen the Mullard amplifier book http://www.vvttransformers.co.uk/drb...b12919_dhn.htm not only the how but why.

[Radio Wrangler]

The triode-pentode splitter IS a long-tail pair.

Build a simple triode-triode pair and there will be capacitive feedback (Miller effect) from the inverting anode to the input grid, but the second triode acts as a grounded grid amplifier for the path to the non-inverting anode.

Consequently, in the triode-triode long-tail pair one path is effectively cascoded, the other path is not. The paths are unbalanced and there is Miller feedback spoiling the bandwidth and introducing phase shift.

Dr Arthur Bailey of Bradford University was doing design work for Arthur Radford when he invented the pentode-triode long tail pair. In this circuit the screen and suppressor grid of the pentode screen out the Miller feedback and make the pentode side look like a cascode. The circuit becomes better balanced, the phase shift gets reduced and the bandwidth gets increased. All good! This is a very important improvement in cirsuit design.

In the concertina phase splitter, the output impedances are unbalanced. One is a unity gain high output impedance amplifier sort of thing, while the other is a cathode follower output. Unlike the long tail pair types, there is low gain which needs to be made up for elsewhere.

The long tail pair could have been done as a pentode-pentode, but there was no dual pentodes available, but there was a triode-pentode and that was good enough.

Bailey's design allows more gain and more feedback to be used, and an ECF80 equivalent sounds suitable.

David

[PsychMan]

Originally Posted by merlinmaxwell

Simple is best in my book, have you seen the Mullard amplifier book http://www.vvttransformers.co.uk/drb...b12919_dhn.htm not only the how but why.

I'd heard good things about this book but had no idea it was so freely available. After an initial peruse it looks fantastic and has all the info you need to build one of these fantastic circuits. Thanks very much for sharing this !

In the concertina phase splitter, the output impedances are unbalanced. One is a unity gain high output impedance amplifier sort of thing

Not quite true, see...

http://www.valvewizard.co.uk/cathodyne.html

[ColinTheAmpMan1]

Miguel, I suppose I could be wrong, but I'm not sure that an amp with 807s is really what you want. These valves are quite big and can put out a lot of power, certainly over 20 W (Wikipedia quotes 120 W for a pair in AB1/AB2)! They are beam tetrodes and also not good for ultralinear output stages due to the low maximum screen voltage of 300 V (the max anode voltage is 750 V). Something like a couple of 6V6 or EL84 types would be more in keeping with your intentions. It shouldn't be difficult to find EL84s - I have loads that still have some life left in them.

I would also say that it is possibly a bit of a waste to use EF86 valves here, too. They are lovely little pentodes with fairly high gain, but do rather suffer from microphony. I would try to source some anti-microphony sockets for them, if I could.

Of course, it may be that you are limited in what you can obtain, so my comments are superfluous.

Whatever way you go, good luck.

Colin.

[Miguel Lopez]

Originally Posted by ColinTheAmpMan1

but I'm not sure that an amp with 807s is really what you want.

That is exactly what I want. I'm not interested in watts, but in watts using the 807s.

Althought I know I could obtain the same amount of power with a less complicated circuit and less complicated valves, such as the EL84, I would like to expose the reasons why I want to use the 807:
1. To learn new things.
2. To face a new challenge.
3. To honor my friend Andy (aka Dr. Wobble), who has been my patron on all of this vacuum adventures.
4. To honor all of the other forum members who have contributed to this.
5. To have fun. (don't worry, I have learned that high voltage on my fingers is not funny)

Originally Posted by ColinTheAmpMan1

I would also say that it is possibly a bit of a waste to use EF86 valves here

The waste would be to keep it in a box forever. If it needs something to avoid microphony. I will mount its socket on rubber washers, althought I saw nothing relevant on the scope this weekend.

The waste would be to keep it in a box forever

Oh yes! Good man. You may have to bias the 807s quite a bit negative for a low power amplifier, but that's half the challenge. I think you could do a very good amplifier with an EF86 as the first stage (as a pentode) then another EF86 wired as a triode for a concertina phase splitter and two 807s for the output. A bit of negative feedback and it will sound as good as anything else.

Seeing other examples of your work it will look good too. I also bet the 807s under-run will last for ages.

[Radio Wrangler]

807s are also VERY long-lived valves and should last a lot longer than EL84s.

MM, I think we'd better agree to disagree on that concertina phase splitter, yes I know that the input Z of the valve loading the cathode will modify what is provided to the valve loading the anode, but the effective source impedances seen by each valve are different. Arthur Bailey's analysis seemed reasonable to me. My preference is for cascoded diff pairs driving power devices via followers. I took apart the amplifier I'd built with TV line-output valves once I'd completed some experiments with it, but the transistor version has been sitting in my living room for thirty-five years, so I guess it's vintage, now.

David

[Diabolical Artificer]

Bless you Miguel, never been called a patron before and helping out you out is always a pleasure. In that light as I'm due to send you another parcel I have a few ECF82's as well as a lonely ECL86 I can send you.I guess you'd rather knock up some odd phase inverter though, as I know you like to experiment.

As you know you can use a triode as a Cathodyne splitter and take one end of the PP pair from the cathode, not very hifi though.

Andy.

[Miguel Lopez]

What do you think about inserting a tone control (Baxandall) network before the EF86? In this case surely I will need another valve as input and pre-preamplifier (if the later exist ), maybe an ECC83 or ECC81.

I was also thinking if the phase splitter output should has similar impedance to drive the 807 valves? I have seen that such power valves usually have about 200K on the grid, which is smaller than the usual 470K and could possibly load the phase splitter.
Would they need buffering? Maybe I'm thinking with a bipolar-transistorized mind, so I ask you for ideas.

[Radio Wrangler]

The Baxandall tone control is a feedback circuit, so it has to go around an amplifier of some sort. That amplifier could be an EF86, or any one of a number of types.
The input stage of a power amplifier is usually the place where any feedback from the final output is introduced so it needs to be a separate stage to one implementig a tone control.

Also the Baxandall circuit poses a bit of a variable load on the stage driving it, so having a buffer in front of it is a good idea.

So we arrive at the classic structure of a 2-valve preamp, then a power amp input stage, phase splitter and then power valves. It evolved for good reasons.

(If Andy's going to be sending a parcel across, is there anything you're on the lookout for? I can have a look in the attic the next time I'm up there.)

David

[Omegaman]

Hi Miguel,
Hope you are well?

I have attached a couple of Geloso amplifier schematics that use 807s.
Maybe of some help or give you inspiration.

Lots more here.
http://www.rsp-italy.it/Electronics/...loso/index.htm


Best regards

[Miguel Lopez]

Thank you very much for all your kindness.

Saddly, I will have to "mothball" the PP amplifier project. I tested the 807s this weekend and one of them seems to be dead. I fixed-biased them with 280V on anode and -28V on grid. Cathode fixed to HT-, and Screen to anode via a 1K resistor. One of the 807s emitted 30mA with -24V on grid, so this seems to be OK. The other required -7.5V on grid to emit the same amount of current so I guess this one is "kaput".

Now I'm thinking to use the PL36 or PL504 that forum members sent to me some time ago (via Andy), I tested a PL504 and with -28V on grid it emitted 120mA which is very good. As I have several valves of this type, I can use them without problem with spares.

I realized that I can not face a project without spares cause if some component fails (or I kill it), then the whole project (and work and time) get lost. I also have a couple of 6V6 which are good (I also tested them) but there is the same problem with spares. They are just two, so if any of them fails, well, the project is "kaput" too. I think that the PL504 or PL36 should be fine. I have to test the PL36 yet.

Then I'm thinking to use a couple of PY33 diodes (I have three in fact) that forum members sent to me some time ago (via Andy) in order to build my first all-vacuum amplifier. For this I will have to rewind and existing transformer as all of this valves requires 27V on heaters, so, it will takes a little more time, that's why I will make a stop by now.

This project is not ended but "mothballed", and I will resume it soon.

Bless you all.

[Diabolical Artificer]

Sorry Miguel, I didn't have a valve tester when I sent them to you- they were part of a job lot off Ebay. Anyhoo, all is not lost Miguel, as I'll send you a spare. Chris AKA Herald etc is sending me some 807's and a few bits and bobs which I'll forward to you, though obviously it may take some time for the bits to reach you. You have to wind the OPT anyway.

I wonder why Williamson chose the 807 as the OP valves in his original design. Were 6L6's and KT66's available when he drew up the design? Pretty sure they were.

Andy.

[k_yller]

Hi
PL504 will be fine, easily 25 W per channel
If you need diagram of commercial amplifier with PL504 (just for inspiration)
just let me know, I have it somewhere

[Miguel Lopez]

Any diagram is always welcomed. That helps to get ideas.

[Miguel Lopez]

Hello

It is time to resume this project. In fact I've been working on the chassis for this amplifier. I have a doubt that I like to clear with your comments.

As the 807 is a very tall valve, and I want to put the amplifier inside a cabinet as I don't want such dangerous voltages and temperatures exposed (my wife's duster is very restless and audacious), so I thought to mount the valves horizontally and then the doubt arose.

In the datasheet by RCA (USA), it is said that the valve can be mounted on any position, but in the datasheet by STC (Australia) it is said the following:

Mounting: The valve should be mounted in equipment si that it operates in a vertical position with the base downward. In exceptional circumstances it is permisible to mount the valve horizontally but only in such a way that the plane of the majot axis of the control grid and screen are vertical. The plane of these grids is correct if Pin 3 is on the vertical diameter of the valve base.

Does anybody have any experience mounting this valve horizontally?

[Herald1360]

No experience, but the description of grids long axis vertical makes good sense (I believe similar advice is offered for KT88/TT22 valves) since it reduces the likelihood of a hot grid sagging onto the cathode or other grid and causing fireworks. If the valves are run reasonably within their limits then problems would be unlikely. Full power continuous commercial service in a transmitter final might be a different story.

I believe RCA 807s (like the one I sent) are supposed to be the "gold standard" for these bottles.

[joebog1]

I can supply two brand new wax packaged Australian army 807's if you wish. Made by Radiotron Australia. ( EXCELLENT quality valves )
I also have CV428 brand new ex RAAF made in England by ITT. But they are loctal and run much hotter than 807's due to the very small bottle. I have ceramic bases for both of these valve types. PM me if you want a couple of each.

Regards
Joe