KT-88 from a book

[diesel_tech]

This schematic is from " An approach to auio frequency audio desgin"

Its a 30 watt KT-88, what do you think about it? Is it worth building?




I have a large transformer to use, outputs of 400-0-400 no load, and outputs for filaments. No real idea of current, but the core is pretty big 3 1/4"X4 1/4"X 3 1/2".

Other amps you have built, that sound goos, but are still rather simple?

Thanks
Jeremy

[Eli Duttman]

Jeremy,

That's Williamson topology. Only creme de la creme O/P "iron" works in that style of circuitry. Otherwise, the phase shifts will make your life miserable.

[JoshK]

Why is that, Eli? I don't see what in particular about this design would make that so (I am still learning).

[diesel_tech]

I thought a williamson amp used a single resistor on the cathodes of the O/P? This uses a seperate resistor on the cathodes 400 ohms bypassed by a 50 uF cap.

Or is my thinking wrong, it very well could be, i'm trying to learn.

THanks
Jeremy

[Eli Duttman]

Originally posted by JoshK
Why is that, Eli? I don't see what in particular about this design would make that so (I am still learning).

Josh,

Multiple caps. in the signal path, along with GOBS of loop NFB, make the design EASY to destabilize. The very last thing you want in an amp is to satisfy Barkhausen's criterion for oscillation. A 180 degree phase shift, unfortunately, does exactly that.

Notice that there are 2 signal path caps. in each leg of the amp. Right away, that's 120 degrees. Mediocre "iron" can EASILY provide the remaining 60 degrees.

Please examine the Mullard circuit. That topology has but 1 signal path cap. in each leg of the amp. So, it's MUCH more difficult to destabilize that topology. Cap. induced phase shift is only 60 degrees.

[JoshK]

Eli...thanks again! I thought it might have something to do with the caps and the amount of FB (which isn't obvious from that simplified schematic), but wasn't sure if that was what was it.

[richwalters]

Originally posted by Eli Duttman
Jeremy,

That's Williamson topology. Only creme de la creme O/P "iron" works in that style of circuitry. Otherwise, the phase shifts will make your life miserable.

That's unfair criticism of a circuit topology that's stood the test of time. I regulary design 4 stage amps with nom 20dB global feedback and can get til 36dB before instability sets in.
I disagree with you on the Mullard 20Watt amp. One of the reasons it's stable is the very low pole of the phase splitter mentioned in Morgan Jones amp book page 415-417. I've listened to the 20Watt amp and it sounds very flat.

I regulary use concertina phase splitter and there is absolutely nothing wrong with it. If the previous section is DC coupled to the concertina, then really it becomes a 3 stage amp in disguise. So many amps have 2 sets of interstage coupling caps and although you mention the theoretical phase shift disadvange, in practise the problems are eclipsed. Work the problem...

I regulary place orders for 150W o/p transformers based on Williamson design practises, although 18 sectioned and pricy, I consider them cremé-de-la-cremé .

richj

[Eli Duttman]

Rich,

There is absolutely nothing wrong with the Williamson topology. As you said, it HAS stood the test of time. Built with GOOD "iron", it gets the job done. I merely remind folks that Williamson used VERY special Partridge made trafos in his original. Don't build a Williamson amp with Hammond O/P trafos.

While I like Mullard's topology, I don't like their tube selection. Circuits that employ loop NFB should use high gm types in the small signal positions, to provide resistance against slew limiting. The 6SN7 Williamson used is reasonably high gm.

[richwalters]

Eli, what puzzles me is that earlier designs were never really sussed out for IM thd, and I found the driver stages were often blamed for poor performance. (preceeding stage is a concertina)
Enclosed is a modified driver circuit I regulary use,
typical of Williamson but seems never to have been followed up in other push-pull voltage driver designs. (RCA did it in a 50W juke box amp). The cathode/ g1 feedback on each side is a glaring obvious, and the common cathode greatly assists in stage balance and to top it, a pot is used in the anode loads to balance out thd. I find it a darned good stage and with 12BY7's as triodes, the b/w performance and output impedance (appro 1.5K) is exemplorary in parallel o/p stage p-p amps.

Strangely enough Morgan Jones doesn't even touch this.



The inclusion of a 10 Ohm resistor in each cathode leg of fixed bias o/p stage tubes is well known.
As with the above comments, I'm sure is a major contributer in the excellent sonic performance of my amps. This is tipping on a massive technical subject, but -> Has anyone done a computer simulation ?

richj

[ray_moth]

Circuits that employ loop NFB should use high gm types in the small signal positions, to provide resistance against slew limiting.

Eli, you've made this statement in several threads and I've been dying to ask what you mean.