Tube hi-fi advice?

[Steve Conner]

Hi guys

I fancy building myself a tube hi-fi amp sometime soon. I don't much like giant horn speakers or expensive triode tubes, so I guess I'll go for something fairly high-powered using ordinary tubes, that can drive ordinary speakers pretty loud. It would probably end up more or less a Dynaco Stereo 70 clone.

Anyway, the question is, what happens if I use other tubes than EL34s in the ultra-linear output stage? For instance the Sovtek so-called "5881s" that you can get really cheap nowadays. Does it still stay nice and ultra-linear or will I get some sort of weird communist distortion? ;-)

I have a stash of Mullard and Philips EL34s, but I'd rather save them for my guitar amps. I get the feeling a stereo amp is going to get left on longer and eat tubes faster.

steve

[Ray Ivers]

Steve,

Are you still using the Toaster? Seems like a mighty fine hi-fi platform to me - PS regulation, active Bax EQ, AC balancing, etc. - just add the UL OT and you'll be there.

Ray

[Shea]

I'm certainly not an expert on this, but here are my lowly two cents.

Other tubes besides the EL34 were used during the golden age of hi-fi. Fishers (not really a top-shelf line, but pretty good) used 7591s. McIntosh still makes tube hi-fis, and they use KT88s. In the Norman Crowhurst book that I have, the 5881 (the real one) is used in nearly all of the examples he gives of different power amp circuits.

I've read that beam forming tetrodes are more prone to oscillations than real pentodes, but there are ways to deal with that.

The Sovtek 5881 might work pretty well. I think it sounds a little dull for guitar, but that might be a good sign for hi-fi purposes. It's rather rugged as far as new-production tubes go.

Shea

[soundmasterg]

A lot of people feel the 6L6 and 6V6 family of tubes sound bland for hi-fi, especially in the mids. Thats not to say that they aren't/weren't used as the Mcintosh MC240 used 6L6's among others. EL34's and EL84's are very sweet sounding with hi-fi stuff. Check out Triode Electronics site and the Dynaco stuff they offer.

[Enzo]

And keep in mind that in guitar amps we run the piss out of tubes. Your hifi will not. It is exactly the sort of application the tubes were intended for in the first place. They should last longer in the hifi than the guitar amp. Yes, even if it is on most of the day.

And tone is another issue. Guitar amps are PART of the tone, they are not simple reproducers. But the hifi really is just a reproducer. Guitar amps are not flat nor are they intended to be, while hifi amps are flat - or supposed to be. SO tonal infuences of tubes on a guitar amp don't necessarily translate to hifi directly.

[Steve Conner]

Hi guys, thanks for the replies! It sounds like it's well worth trying with those Sovtek beam power tubes. I'll check out Sowter Transformers and see how much a PT and pair of UL OTs would cost. I guess before I choose an OT I need to decide whether I'm running in Class-AB or Class-A, and whether I'm using two or four tubes. Am I right in thinking the Stereo 70 was Class-AB but biased pretty hot?

Ray: The Toaster is all well and good but it's not stereo

[Ray Ivers]

Steve,

No - sorry! - I didn't mean actually converting the only existing Toaster amplifier - you're currently using it for bass, correct? - but rather using the original Toaster design (X 2) in a new build along with a UL OT, which should make for a stellar medium-power hi-fi rig.

Did I read in another post that you tried KT88's in the Toaster and got B+ sag? I take it you've eliminated the active B+ regulation, then? If you didn't like the way the regulated supply sounded, you could always add a (switchable?) build-out resistance right after the current-limiter sampling R, and have the best of both worlds.

Oh, BTW - I think I might have found why you were getting voltage-reference failures in your regulator circuit early on, I think I E-mailed you about it a long time ago.

Ray

[Steve Conner]

Hi Ray!

Oh, I get it! Well TBH, I thought about the regulator thing a lot, and I'm not going to use it again. I've since experimented with another 50 watt amp with just an old-fashioned CLC filter power supply, and there just isn't any audible hum at all. I guess there's intermodulation by ripple at high volumes, but that's probably part of the "Vintage tube sound", so I want it for my hi-fi too. ;-) The one good thing about my regulator was the screen protection circuit, but I can get the same effect with something like a 50mA fuse in series with each screen resistor.

I still have the regulator in the Toaster, but the KT88s draw so much current that the unregulated rail sags right down and the regulator runs out of headroom and drops out. That's what I think is happening anyway.

The reference would fail in my circuit because I was using resistors with too low a voltage rating in the sensing chain. They arced over internally and blew the TL431 to hell. Since I fixed that, the amp has run with the regulator for about 6 years now, without any failures.

[kg]

honestly steve i would seriously consider a znfb output stage. that means triodes (or triode connected multigrid tubes) to keep linearity. if you want power, parallel 'em up.

i think a quartet of 6550s in triode push pull would be nice for each channel. give yourself a nice b+ and a stout driver circuit and you'd still net at LEAST 50wpc with excellent linearity.

as a side note, over the years i have learned a couple of tricks for triode connecting multi grids that can improve their sound. i can go into more detail if you're interested.

[Ray Ivers]

Steve,

I've since experimented with another 50 watt amp with just an old-fashioned CLC filter power supply, and there just isn't any audible hum at all.

Yes, it's possible to get unregulated supplies extremely quiet and also to regulate fairly well; its getting them to do both simultaneously that can be (IMO) more costly and time-consuming than just regulating and being done with it. Just FWIW, of the nine regulated supplies in my amp, only three were regulated specifically to eliminate hum/noise.

The reference would fail in my circuit because I was using resistors with too low a voltage rating in the sensing chain. They arced over internally and blew the TL431 to hell. Since I fixed that, the amp has run with the regulator for about 6 years now, without any failures.

I saw that in your original Toaster notes, and I only mentioned this because the simulator showed a spike of full raw B+ appearing across the TL431 for about 50uS after turn-on, the time it took for the associated MOSFET's to conduct; I added 13V Zeners across the TL431's just in case, and used two series 350V resistors as you did as well.

[Steve Conner]

Hi KG, thanks for the advice! I'd definitely check out ZNFB, but it always kind of annoys me to pay $$$ for big power tubes and then cripple their output by running in triode mode. When I was looking at building a bass amp, I nearly crapped a brick when I saw the prices of "SVT Sets" of six Russian 6550s. It's not that I can't afford them, I just think they're overpriced. The Sovtek tubes are great value for money and I remember you testing the 5881WXT and finding that it could take a sh** load of power.

I guess I could try UL with no feedback, and pretend those Sovtek 5881s were ghetto 6550s UL is halfway to triode after all. If it sounded bad I could always rewire it in triode mode or add feedback. I imagine the main problem with UL and ZNFB would be too high an output impedance that would make modern speakers sound boomy in the bass end, since they're designed with high damping solid-state amps in mind.

A philosophical point: Triode mode is feedback from the plate to the screen directly. UL is about half as much feedback from the plate to the screen through part of the output transformer. Ordinary feedback is feedback from the plate to the control grid through the OT and driver tubes. When does feedback start becoming evil? I don't really see any distinction between the three kinds of feedback, except ordinary feedback is easier to screw up the design of >_<

I have some GEC KT88s that would be great, but unfortunately only three of them

Ray: Did you actually use my regulator circuit? That gives me a nice feeling, I thought nobody else was ever interested in it. I'm not sure about the spike thing, but I'd imagine it was an artifact of the computer model.

[Ray Ivers]

Steve,

Did you actually use my regulator circuit?

Use it? I designed my entire HV PS section around it!

I made a few tweaks here and there, not addressing problems but just optimizing it for my particular situation, available parts, etc. The hi-B+ regulator simulates a rock-solid 360mA @ 550V, at 108VAC line voltage (adjustable from 240V to 610V), and I made the current limiter adjustable using a Ge sensing transistor, 555-based time-delay turn-on, etc. Now to get it to make some noise...

I totally second Ken's UL suggestion. UL puts out practically the same power as pentode, with about the same Zout w/o global feedback as the pentode stage with a medium amount of NFB - you just need more grid drive to make it happen, and preferably tubes like good KT88's w/low internal rs, so they can more easily drive the UL taps. It seems like a lot of the used GEC KT88's I see on eBay are in the U.K. - you might find one more for a good price (set up your 'My eBay' to auto-search for one, you might have one in a week or two!).

And here's my 'NFB evilology'.

1) Worst: NFB across several reactive devices (e.g., the 5F6A)
2) "Better": NFB across a single reactive device (e.g., RC-coupled local GS NFB, UL - UL's really a piddling amount of NFB, BTW)
3) Acceptable: NFB across a non-reactive device (e.g., cathode/plate degeneration, direct-coupled local GS NFB)
4) Best: Zero NFB
5)... and don't forget about PFB!

Ray

[Steve Conner]

Yeah, I mean if negative feedback is evil, positive feedback must be good, and the more of it the better, right?

Anyway, it's looking like I'd get a pair of good quality Williamson style OTs from Sowter or whoever, designed for a pair of KT88s in ultralinear running about 70W output. I'll try and complete my set of KT88s, and if that doesn't work out, I guess I can use two 5881s or EL34s in place of each KT88?

steve

[Arthur B.]

Most of the original UL designs used KT66s for the output stage, so you can try those out.

[capnjuan]

Hi guys

I fancy building myself a tube hi-fi amp ... something fairly high-powered using ordinary tubes ... drive ordinary speakers ... a Dynaco Stereo 70 clone. . I get the feeling a stereo amp is going to get left on longer and eat tubes faster. steve

Hi Steve; can't answer on UL v. 5881s but two things; assuming good ventilation, reasonable volume, and attention to bias (a stock ST70 provides test points), EL34s can run for quite some time. I had a pair of ST70s that I ran strapped and, for the money, nothing better.

If you're looking for a project, that's one thing; if you're looking for gear, an updated ST70 w/ a regulated supply would still be, dollar-for-dollar, the best money you could spend; terrific transformers.

JH