Bored with Leak ST20 - ideas?

[andyjevans]

I've been modifying Leak Stereo 20s for what seems like about two centuries, and my jaded soul is now looking for more extreme mods.

Main condition is that from the outside it continues to look like a retro Leak - no horrible large caps or chokes in the middle of the amp. Valves must be nicely aligned on each side of the chassis to give the Leak "look". But the valves could be noval, octal or UX4/5, or indeed rimlock. One way of doing this is to have a small additional plate to cover up the noval sockets. The noval sockets are really the weak link in terms of more exotic ideas.

Second condition - sorry guys - is all triode operation with zero NFB. Goal here is serious audiophile quality.

HT can be about 320v with GZ34 or 385v with solid state rectification, in which case the rectifier supply of 5v at 2A would be liberated for use. Max existing heater capacity is 4.8A at 6.3v.

So far I've successfully built - conforming to above conditions:
1. ECC40 into EL84s in triode, zero NFB. That's easy and very good. Can be interstage coupled or cap coupled (Russian teflons). Balanced all through with XLR inputs.
2. ECC40 into parallel PP 6S4A. Rest balanced as above, except ss rectification. ECC40s operate at just over 5v AC off rectifier winding. I guess I should improve on this bit, but it sounds fine. In fact it's a lovely sounding amp altogether. Combination bias on the output tubes.

Now I want to go a stage further. Ideas are:

a) PP 2a3 amp, not sure of input - maybe try a two stage amp.
b) PP 2a3 amp with three stages, using ECC40 and one other
c) 1624 outputs or 46 outputs as above - idea here is that a couple of Hammond 2.5 + 2.5v transformers could go under the bonnet.
d) Indirectly heated triode outputs. Many possibilities here, but given the 4.8A heater max, we'd be looking at trioded 6L6 variants like the 807 or 5881, or indeed 6AV5 or 6FW5.
e) Double triodes as outputs - like 7236, 5998, 6AS7, 6080. Would fit into the heater requirements (just) but harder to drive and input tubes would need extra filament transformer. Not ideal.
f) Lower power option with 6AH4, 1626 or 6V6 (trioded) outputs, or parallel 12B4 outputs, or dissimilar triodes like 6EW7

I'm also considering DHT outputs like 71A but here we're into DC supplies which means real estate. I'd like to go all DHT in fact, which could be done with a SE to PP interstage. In this case it would be a three stage amp. Inputs might be 30, 31, or even a double triode like 1J6G. Something fairly slim at any rate.

So - over to your creative thinking. Think BIG - think audiophile quality!

Cheers, Andy

[Eli Duttman]

Andy,

IMO, you can have everything you wished for, except no loop NFB. The active circuitry will be just fine, without NFB. Unfortunately, the Leak O/P trafos need a bit of linearizing help.

You can follow in SHiFTY's footsteps and build an "El Leako". The Saratov made, TungSol labeled, 6V6 is a superb tube, when triode wired. Ask Poindexter. Our "Kiwi" friend put a big choke on his amp. To avoid that, SS rectify the B+ and stick a gyrator inside the chassis.

Surely, you have access to Mullard made 12AT7 equivalents.

[andyjevans]

Hello Eli!

Actually the Leak OPTs are fine - there's a Morgan Jones mod which rewires them and the sound is very good with zero NFB. That's what I'm using in my current version of the ST20 with ECC40 into parallel 6S4As, and no sound problems there.

I'm aiming high here, so though I have used 6V6 in triode and it's nice, as you say, I'm inclined to go with a 2.5v filament DHT like 2a3 or 46 or 1624

I've been measuring up the chassis, and one version which appeals to me is 1J6G direct coupled to 5687, transformer coupled with Hammond 124b into the 2a3. My client wants balanced operation.

This means losing the GZ34, but we can nevertheless put in a couple of smallish 10H chokes which would combine into a 5H choke at 200mA which is the max for the mains transformer. HT would be about 385v so with cathode bias on the 2a3 that gives plate to cathode voltage of around 320v for about 40 to 45mA. Is that 320v safe for Chinese tubes? Should be fine for Sovteks.

Gyrator would be good. I'd also have to find -15v for a couple of CCSs per channel.

I'm just wondering how to do the filaments of the 1J6G - would it be OK, I wonder, to use the 5v rectifier winding to make two seperate DC supplies for the two tubes?

Those are my present thoughts. Will have the amp on my bench tomorrow.

Andy

[jez]

Hi Andy,
It,s been a few years since I was last on the Leak Yahoo groups thingy.....
I would still stand by what I said then in as much as that the best stereo 20 I ever heard was one which I modded for a customer using totally regulated supplies...still ultralinear and NFB etc.
The thing here that I think is a salient point is what is the difference( both objectively and subjectively) between the earlier
Leaks with 100K anode load on first stage and the original type transfomers, and later ones which had 47K anode load (and some had a step network on the first stage) and different transformers ???. I've seen nothing said ever about this but surely with such big differences in the design there would be equally big differences in the sound? The transformers were wound in a completely different way according to Stephen Spicers book. It's quite possible that the performance of this particular sample was down to it having the optimum combination of output transformers, component values and NOS special quality Mullard valves.... and that the regulated supplies I fitted to it were just the icing on the cake!?
Maybe you could shed some light on this after experiancing many different examples of '20... I can say that with either Quad 57's or Spendor BC1's this particular sample sounded *&%& amazing!!!
much more so than any other I have heard. Maybe the OPT's were the opposite of Monday morning ones?
In many ways this is THE BEST sound I have ever heard
.. and I have heard a lot of high end hi fi in my time!
I think that this whole "no feedback, must be triodes" thing is almost a religous fervour for many (thats slimey!! so many cans of worms being opened!)
I'll leave it at that and await the hit-men at my door :-)
Jez.

[Eli Duttman]

I think that this whole "no feedback, must be triodes" thing is almost a religous fervour for many (thats slimey!! so many cans of worms being opened!)

Jez,

No "torpedoes" sent out on my orders.

NFB is a tool. Like all tools, it's subject to abuse. If the open loop circuitry is reasonably linear to begin with, a few dB. of NFB can make good better. OTOH, using gobs of NFB to bludgeon a fundamentally poor design into line rarely (never?) works well. Tool? Yes. Crutch? NO!!

[andyjevans]

Hi guys.

On my journey to zero NFB I created several feedback versions of the Leak as it gradually morphed into something better (to my ears). One major improvement was using the standard Mullard circuit - direct coupling stage one to stage two, with adjustable anode resistor on one side of stage two. Best valves tried for this (out of about 30 common types we tried) was 6N30 into 6N1P. This became known as the Russian Leak and several were made, delighting their owners.

I then by accident forgot to connect the feedback on one side. I couldn't believe the improvement - the music seemed to let go and lose its tightness. After that I did become, as you say, a believer. I made a couple of amps with ECC40 into EL84 in UL with no feedback, and they were better to my ears, though I had by then started using DHTs like the 2a3. That really settled it for me - I decided from then on to only use triodes and drop the feedback, and I've continued along that path. It's a trade off - pentodes and feedback give you more power and dynamics, but there's no doubt in my mind that triodes and zero NFB give better tone and timbre to the instruments, and that's really what I plan my improvements around. I'm an ex-pro musician and the sounds of acoustic instruments are so vivid in my head I try and re-create them as closely as possible. So that's my thrill and my yardstick - timbre. I'm well aware that others prioritise volume, headroom and dynamics. That's just how it goes. Still others prioritise soundstage, a road I don't go down at all. We all have our faves.

Andy

[jez]

Well I think that at the end of the day it's "each to his own" when it comes to hi fi, especially speakers but obviously amps too.
Personally I like an amp that neither adds anything nor removes anything from the original sound. This approach does not always give the most "pleasant" sound but at least it's accurate.....The rest is down to the recording quality, room acoustics etc.

Back to Stereo 20's though. Can you comment on any differences between standard early 3921 and later 8778 transformer equipped versions?

I am just finishing a rebuild on a late model for a customer so I should be able to compare this to my own early type over the next few days. Unfortunately it will not be too meaningfull a test as my own version has huge amounts of external reservoir caps and an external choke, whereas the customers is fairly standard.
Should be interesting though.....

Eli,
Using "just a few db" of feedback can actually make things worse!! There's plenty of literature out there on the subject
It tends to result in a small reduction in second harmonic distortion but a RISE in the nasty 3rd, 5th etc. I would guess that around 15db would be a minimum, much less than this and you may indeed be better of without feedback

[andyjevans]

<<Back to Stereo 20's though. Can you comment on any differences between standard early 3921 and later 8778 transformer equipped versions?

No idea. I rewire the OPT anyway. Have you tried this?

Andy

[jez]

I can't say I have tried that. In what way do you rewire the OPT?
I have certainly considered taking the NFB from the tap that's in use (usually 4R here) and making the required changes to compensation etc but have never got round to trying it.

[andyjevans]

Well - just got the amp. A quick look confirms that I can easily put in four UX4 sockets for 2a3s using the existing larger holes for the caps and GZ34. So I think that's a decision already.

So now it's a question of the input stages. I have three 9 pin sockets per channel available. If I use two for a couple of double triodes like 6N30 and ECC40 I have one left over.

could be:

- two times EZ81 for tube rectification.
- a couple of 105v damper diodes to give 210v for the driver.

could be a couple of 12b4 instead of one of the double triodes.

Just thinking out loud - needs to be balanced all the way through.

andy