Valve amplifier design 1965

[Bobdger]

Valve amplifier design

I am after a copy of the article written by Bailey in 1965. There is a copy of the whole year on e bay, but I only want to read this one article.

Bailey published a new design for an amplifier that used ECC88 in cascode as input, 2 X EF184s as phase split, EL34s output and used a radical new system for feedback, two feedback paths, one that included the output transformer for low frequency and one for HF from the output anodes.

This was published in "Electronic Engineering" Dec 1965 (according to the HiFi history book)
I cannot find an electronic copy of this, WorldRadioHistory does not have 1965.

Many thanks.

Bob

[GrimJosef]

I don't have the EE '65 article Bob, but I take it you're aware of the STA100 documentation here https://www.radfordrevival.co.uk/literature/sta100/.

The small-signal valves are the same as you've described but it uses KT88s to get 100W out. The feedback from the OP transformer secondary only is conventional.

I've worked on one of these and found that it could perform very well indeed, but it was a bit dependent on the EF184s being accurately balanced (there's a pot for this) and, since the first stage is DC-coupled to the second, on the ECC88's parameters being close to nominal.

If there is a copy of the article available then I'd be interested to see it too.

Cheers,

GJ

[Bobdger]

Thank you for the Radford info. It's the feed back that I am interested in. I will let this request run for a bit, hoping that Edd or someone else may have this article.
Bob

[Ed_Dinning]

Hi Folks, I have the article and will try ans publish a scan.This article was the subject of a patent by Bailey, so that may be available from the patent office library.
The phase splitter was covered in WW 68, P411 (1962)

Ed

[Bobdger]

Hi Ed
Thank you, please can you PM me a copy.
Bob

[Ed_Dinning]

Hi Folks, attached some scans of the article, apologies for the poor quality due to the scanner and it being a bound volume. Not much detail I'm afraid, especially of the transformers, so a copy of the patent may be useful

Ed

[Ed_Dinning]

Hi Folks, an examination of the article in Sept 62 WW on the world radio history site shows no further details of the summation transformer, but does give references to earlier works.

It would appear to be a bi-filar wound 1:1 type and I would suspect it would us an Ni-Fe core of some type. If anyone can find the patent there may be more details there

Ed

[kalee20]

Thanks for putting up this one, Ed!

[GrimJosef]

Interesting article Ed. Bailey's clearly refined this circuit to quite a high degree.

I wonder how critical the wiring layout is ? I see he's fed a little bit of signal back from the anode of V2 to the anode of V1b, and thence to V2's grid, using a capacitor which he's specified as 0.5pF. In audio amp design you don't very often see 0.5pF caps used explicitly ! (Even at 100kHz it'll have a reactance of 3Mohm.)

Cheers,

GJ

[GrimJosef]

Or, thinking about it, I wonder if the 0.5pF, multiplied by the Miller effect, is there to tweak the load that V1b sees ?

Cheers,

GJ

[Ed_Dinning]

OK Guys, see what else you can glean from the other references

I have some Ni Fe lams if anyone fancies trying to spec the transformer.
It would also be interesting to know what his spec was for the OPT

I wonder what happened to his papers, I may query with the IET library

Ed

[Bobdger]

Hi Ed.
Thank you very much, more food for thought.

Bob

[Ed_Dinning]

Hi Folks, I've managed to pursued my scanner to output as .pdf's, so see if theses are any better.
If not could one of the recipients repost in a suitable format for all

Ed

[joebog1]

My PC still shows that as a JPG image Ed.

Joe

[Electronpusher0]

Ed kindly sent me his scans by e-mail to avoid the size reduction the forum software applies to jpg pictures.
I have cleaned them up and combined them into a single pdf, the resolution is really not bad.
I have also extracted the schematic at A3 size.

Peter

[Bobdger]

Hi Ed and Peter.
Thank you for the new hi res copy, I now dont need that new pair of glasses. Hopefully Grimjosef has also seen this as he was also afte a copy.
Many thanks.
Bob.

[Radio Wrangler]

Extremely interesting!

Traditional feedback arrangements from the output of the output transformer have limited the amount of feedback which can be employed. The loop gain and phase shift has to be controlled to the point where loop gain has sunk below unity (into the full specified range of loads) in order to ensure stability. This point will be well above the audible range in a good amplifier. Leak probably pushed the feedback factor about as far as it could be with the traditional topology.

Here, Bailey has taken it a step further using a crossover network to segue from output winding feedback to anode feedback, and he's provided a balun to cover both sides of the push-pull output stage.

Also in this article, he's gone to the full pentode-pentode long tailed pair, pointing out the limited drive capability of the simplified pentode-triode single bottle version he put in in the later Radford amplifiers.

Thinking of Radford, Arthur Radford did a lot of business in lab equipment with his Labpack power supplies. With the connection between the two Arthurs, was this article heading towards a lab amplifier as a Radford product?

I think I can make a good guess as to where the transformers were made I suspect the main output transformer could be an existing Radford EL34 type, possibly with different secondaries to handle a different selection of load impedances, but I suspect the sectional design remains.

The high band balun transformer for the feedback path has the crossover networks driving it. These not only limit the bandwidth the transformer has to handle on the LF side, they also shift the impedance regime the transformer must work in. In fact, there is a 68 Ohm resistor across the unbalanced output of this little transformer.

A little analysis of his crossover network will show where the 3dB LF rolloff is in the drive to the balun, so the inductance of each winding of the balun needs to be an order of magnitude more than 68 Ohms reactive at this frequency.

68 Ohms is an awful long way below the anode impedances EL34 amplifiers run at. Combine this with the higher LF roll off and that little balun transformer looks to be pretty easy. A transmission line transformer on a ferrite core might even be a possibility. The low inductance and the low number of turns and the low impedance mean the spurious resonances ought to be fairly easy to push well above the unity loop gain frequency.

It seems a shame this article has faded into the mists of time, unnoticed. This is the next step beyond the Radfords, Leaks and Quads which preceded it.

It is also from an era before the hifi world spawned its extremists and generated its hatred of feedback. It was just an innocent engineering technique back then and satan was not involved. This is an amplifier for engineers. I can understand it not catching on in today's world, but it dates from before the silliness started. It's the next step in engineering out any remaining traces of 'valve sound'.

David

[Electronpusher0]

A bit more research on the Radford revival site shows that Radford did indeed produce an amplifier using crossover feedback, the TT100!

https://www.radfordrevival.co.uk/literature/tt100/

The difference is that instead of using a transformer to invert the phase of the hf feedback from one anode and then sum the hf feedback from the other it uses a differential op-amp.
The LF feedback from the output transformer is applied after the differential amp.

Peter

[Bobdger]

Hi David.

I have just ordered some ETD29 cores from Farnell. I was going to run 2 x 100 turns with ct to see what I get. my thinking that the smaller range would work posiable that the 1;1 pulse transformers my be in the right range.

Bob.

[GrimJosef]

The STA100 that I worked on had a shorted-turn problem in the primary of one of its output transformers. In fact it worked surprisingly well, despite this - a testament to the robust design ! But unsurprisingly my customer wanted it working as well as it could. So we ordered two replacement OP trannies through Radford Revival, who had them wound to the original TT100 drawings, rather than the STA100 ones. These had the secondary optimised for a low-impedance load i.e. a conventional hi-fi speaker, rather than 100V line.

I also added another TT100 feature - a second LC smoothing stage on the main HT supply. This improved the hum performance markedly.

One final change was to modify the bias supply so we could run with KT90 output valves which have a little bit more power capability, and high-voltage durability, than KT88s (or, at least, modern ones). Squeezing 100W out of a pair of '88s at very low distortion is tough.

Despite all this it was a struggle to get to 0.1% THD at 100W CW, even using carefully selected valves and with the drive balance pot carefully adjusted. However at more realistic domestic powers - say 20W - the distortion was genuinely negligible, and that was without applying the clever feedback. I'll be interested to hear how your experiments with the transformers go Bob .

Cheers,

GJ