Experiments with tube audio amplifiers

Tube talk

Experiments with tube audio amplifiers

Postby Doug Coulter » Thu Nov 18, 2010 1:10 pm

(note, this is a stub till I redraw some schematics to post with this)'

As an audiophool (in the past), I've long preferred solid state amps that are "a wire with gain" and will eventually get around to posting my improvements on the Dan Meyer "Tiger .01" circuit.
As a musician, I like tube amplifiers, mainly, as the job is different there -- you are producing, not reproducing sound, it's just a different game. My family (all engineers, some now deceased) were big tube buffs as well, so my first electronics were all tubes -- that was all there was when I began in the late '50's, after all. This eventually led to some pretty good "audiophool" class designs I will be sharing, the main goal of which was the old "wire with gain" standard, and the limitations were usually in the output iron -- then and now, things like UTC linear standard transformers don't grow on trees, and even those have issues and limitations, so I began to focus on how to overcome some of them.

The results were a couple of DC coupled tube amplifiers, right up to the transformer, long tailed differential throughout, with some feedforward compensation for the transformer characteristics. This was an interesting journey to say the least, as tubes do have some drift, and the "hard" low resistance transformer windings don't like DC imbalances, nor are they friendly to output tubes that decide to have a lower plate voltage than the supply -- things get hot! My first attempt, which a lot of people really liked, used two 12ax7's and two 6v6's as the outputs. All the "defects" like THD and IM distorion were below the (rather high) noise floor, and in tests, you couldn't tell which channel of a stereo amp had the crummy radio shack output iron, and which the linear standard, so it was a success by most measures. I will try and recreate the schematic from memory, as I gave it to a family member who since died and the people cleaning up the estate took it. So I won't remember all the resistor values and so on, just the basic wiring and ideas.

My other more recent attempt is sitting across the room for me, and was custom designed for a current tube-freak audio guy (he got the nice looking version, I kept the prototype). This one is audbily a tube amplifier, or as he says, sounds "tube-a-licious". I may have to pull it out and try it with my acoustic guitar at some point to see if it's any better than my excellent Fishman Loudbox (bi amped solid state, with nice features, the 8" woofer one that sounds better than their larger one). The customer insisted on triode outputs, which made it a lot harder to get in DC regulation, but eventually it worked, and we found some interesting things along the way. This one used nuvistors for the front end, then 6au6's for voltage gain, followed by the dual triode 3c33 for output. Doubtless few tube guys have even heard of this number -- it was low production right at the end of the tube era, used to drive mag amps for military servo/avionics things, but it's a decent size all-glass dual triode with nice curves, and I just happened to have about the world's supply of them on hand as new old stock. So, why not? Being exotic just gave better bragging rights to the eventual owner, which I suppose is important to someone paying that much for an audio amplifier.

Both of these designs were very unlike something Joe Sousa would normally do -- I didn't take any thought to making them minimal-component whatever, they were more "no holds barred" or "heroic" attempts to do the best at any cost or complexity, though they did come out simpler than some designs I've seen. Being DC coupled, both needed a lot of total supply voltage, which I got simply by making the usual full wave rectifier used in a tube amp into a bridge, to get plus and minus voltages. Further, I did DC supplies for all the tube filaments, to get rid of hum, which was largely successful. In general, most of the complexity actually wound up being in the power supplies, rather than the audio chain itself, since I needed some fairly good regulation on some odd voltages here and there to deal with the drifty tubes well. My only concession to solid state here was rectifiers, and in one case, a zener diode, but mostly I used the neat old gas tube regulators because hey, they are just cool and visually attractive. For the rest of the parts, the best available were used -- polypropylene or silver mica capacitors, 1% mil spec resistors and so forth -- that stuff is now cheap, and no reason to skimp when someone is crazy enough to pay you $10k/channel for a good amplifier, and yes, those guys exist (not many, though).

One of the interesting things we found (and Joe saw this during a vist here) was that the 7586 nuvistors I used as the first stage of the triode amp were temperature sensitive enough that touching one would cause a crazy amount of DC imbalance in the input stage. I wound up fabricating a custom "heat sink" that clamped over both tubes to keep them equal in temperature, a kind of 'tube integrated circuit" to deal with that, which had the side effect of looking pretty cool too, which again, is important at the rarified high end of audio-foolery. Being small, the response was fairly quick too -- a second or so, at this made me really do my homework so that there weren't conditions where one side dissipated more power than the other under normal conditions of audio inputs, a little bit of a challenge. The same issues appear in the article in this thread for solid state opamps, actually. The truth is, all these different technologies inform one another, and we all stand on the shoulders of the people who went before us. And it can sort of travel in time in either direction. Few were still working with tubes when that article was written, but the findings there informed more advanced tube designs just the same.

(I can see now I'm going to have to start another thread about tube amps for musicians, that have deliberate distortion and similar topics -- it's a lot of fun to play with changing the even and odd order distortion content on the fly in a design and in use. I am guessing that as my super video amp gets progress, it will go somewhere else, as it's not a legacy kind of thing, it's a modern thing where tubes just happen to be the best devices to use, still, since I need real high voltages at the output of that and without starting high, the magnetics problems become insurmountable.)

I'll get back here soon with some schematics of those amplifiers -- both turned out pretty well if you like that sort of thing. I'm not a very neat drawer of such things, but they should be readable, anyway. I don't think this should become "lost technology" which is the motivation here. You never know when the old will help out with the new, but I've been around the wheel a few times and it's happened enough to be worth paying attention to as a concept.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: Experiments with tube audio amplifiers

Postby vmike » Fri Nov 19, 2010 9:48 am

I too have spent years playing with tube and solid state amplifiers. I think I've built and listened to most every configuration of tube driver/phase-splitter/output sections I could think of. It is amazing the difference a capacitor can make. Over the years I too have noticed that power supply changes seem to have more significance than the majority of the others. My control amps are stock Eico EF-86, Dynaco ST-70 Mac 240 power amps hooked either to a Paragon System E, Audio Research SP-3A1 pre-amps or my Harmon Kardon A-500 integrated amp. I've found the stock configurations very listenable through both the Klipsh cornwalls and the Bozak B-210's I have in my media room. I've also been addicted to modifying Dynaco solid state amps to make them sound as good as tube amps. It can be done. As far as the Tiger amps go, they needed all the help they could get, just to keep the high frequency oscillations from making a snack out of the output transistors. All this being as it is (in this abridged version of history) I have Fusors and cyclotrons to play with, as well as my many other interests. So, I'll just listen to what I have built up and go play. :) Fun stuff though!

mike
I tried to contain myself, but I escaped.
User avatar
vmike
 
Posts: 30
Joined: Wed Sep 15, 2010 4:20 pm
Location: Stillwater, Oklahoma

Re: Experiments with tube audio amplifiers

Postby Doug Coulter » Fri Nov 19, 2010 10:16 pm

Yeah, you've obviously "been there and done that" too. Nowadays, I just don't care so much, and have another way to hear perfect Fi if I want -- pick up an instrument and play it for one thing.
I also have some good material from the days I ran a recording studio, pre processing.

We once did an interesting test at a college party. We put all our tweaker stereos in different rooms, let people choose what they wanted to hear. And this went from table radios to super crazy stuff (large electrostats, rabco tonearms etc, from a GE clock radio to the best there was then). The rather interesting result was that all the hot chicks (and therefore most of the guys) wound up in the table radio room, because it wasn't distracting from social intercourse....hmmmm. Mostly the lonely geeks wound up in the rooms with the truly good stuff. At any rate, there's so little good quality recorded stuff around, there's not so much point as you would hope, but it is fun to play a tape I've made that has things like a dropped quarter -- and see if everyone looks to where it was dropped when I recorded it (in the same room), or thunder -- I've had people run out to roll up their car windows on perfectly clear days.
The RIAA shot themselves in the foot with that one, nearly all their output is so bad no one can tell when it's been further mucked up by going to mp3....we warned them and complained since the 60's, but...Now they are getting what they deserve. Fine me another Beatles if you think you're owed the right to coin money off me, not some rapper using autotune for crying out loud.

I had the same issues with tigers, but found ways to fix that. But yes, once they oscillated, the old 802's and 4502 outputs couldn't turn off quick enough, and poof.

My revamp of that used FETs in the outputs, and more or less tweaking and no holds barred improvement throughout, down to matching parts and so forth (and a neat trick to match N and P feets via transconductance reduction in the N fet only). Real current sources instead of resistor, where that mattered, real voltage regulation on the front end, all that. I'll have to put the schiz up here for anyone who wants it -- they go to DC now, at full gain, so also make decent servo amps.

They are the hands down winners in all listening tests, and actually quite difficult to measure with confidence -- 20::21 KHz 1::1 at full power nets .0027% IMD, harmonc isn't measurable at all to 110 db down. The go to 4 mhz flat, so I had to add a 60 khz lpf RC to the PCB layout to keep from making radio transmitters by accident. Neat little thing. Never did get that good with tube stuff. But once one understands the issues that are different for sound production vs reproduction, tubes begin to shine again.

I did a demo with some audio editing software I wrote that took some tube dudes to tears -- I was able to simulate the "tube sound" on all solid state gear so well they couldn't tell, it really upset them. I had it so I could simply input some tube curves and adjust the output transformer simulation, and bingo -- very useful in the recording studio. But for those guys, it ruined the "magic". Sigh.

I'm all the way there with "other more-fun things to play with" now. Still it was fun today to turn on the stereo down in the lab (which has a 16 foot straight TL subwoofer built into the ceiling rafters) and watch the new assistant who started today look out the window every time a recording (from the radio) had a truck going by in it -- the guys doing the recording couldn't hear stuff like that on their monitors, so it made it through the chain, and by golly a woofer that's flat to 15-16hz -- for real -- will convince anyone it's a sound from nature and not the stereo. I wanted (and still do) to publish some of my more successful designs to that people who are still way deep into that can have them if they want. You never know, and why not share? What goes around, comes around.

So I got the new guy oriented (a neighbor kid with a pregnant girlfriend who is real motivated just now) today, which is why I've not been up here much till now today. He'll be real helpful in the shop/lab I think - and cost effective. Got a lot done today, did a nice fusor run (silver to 1100 cpm in 5 min run) so we have a baseline we believe in before we tear the whole thing down to move things from port to port, put in new grid designs, and generally get ready for the next stage of improvements. Still lots to do there, but I like to be ready before we break the door open, so we don't have the system up to atmosphere for long. We've learned prep is a good idea, so once you're in "ship in a bottle mode" it goes quick.

I'll try to get those schematics up. I don't think anyone has done quite what I did with tubes, and it's interesting anyway. Closest would be tektronix vert amplifier with real power output.
And 3 NFB loops -- two inside the transformer, one around the whole thing, and one feedforward comp for the transformer issues (self resonance etc).

Wasn't that slew rate TIM distorion bull crap funny when everyone was talking it and blaming it on NFB when they didn't understand slew rates funny? Every real engineer outside the audio business knew just what that was.

And yes, caps do make a difference -- and one clue is what works well for a sample and hold -- dielectric hysteresis is the culprit on the bad ones. Even Dan Meyer knew that. That's not some thing anyone ever had to guess about (not that it stopped anyone) -- it's easy to measure. Just as obviously, if there's zero signal drop across a cap, it doesn't matter, but no one else seemed to get that -- a big flat coupling cap (oversized) could be a crummy electrolytic and make no difference whatever. But mylar etc are poison if used in things where there is audio "across" them like in a tone control. And guess what, you can't do better than 10-12 bits goodness using them as the caps in a sample hold, either. Duh.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: Experiments with tube audio amplifiers

Postby Doug Coulter » Sat Nov 20, 2010 1:21 am

Oh, BTW, I loved the Dyna ST70's and had my share of fun with them too (and Scotts and Fishers) -- and I even made some of their preamps sound pretty good. I sort of liked their transistor amp design insofar as the trick drive-current limiting, but never really did like how they sounded that much, nor did DaleJ who worked here and earned a set of Magnepans on a job we did for an audio guy once he heard some better stuff on them (he liked his ST70 but it just didn't have enough stuff to drive them well). The outputs (2n3055?) were just too slow and they slew rate limited to easy. My "good" stereo uses Joshua near field monitors, a big sub and amp, and PJ Snyder's old leaf tweeters for the super top end, with my FET amp design (kind of a tiger clone with numerous improvements). That one's scary-good, but it's in a building I rarely frequent now. The two in this building are not as good, but as you say, good enough. It feels kind of weird to realize that I'm not now listening to speakers I built -- but plain old commercial ones, I'd never have guessed, I was big on that and designed some wild stuff, which all got sold along the way. The only homebrew speakers now are the sub woofers, and some huge PA speakers (for my old band) I built. Heck, I inherited some top end spkrs recently that are still in storage, but unless I put them in the bathroom or something - no place to put them at all!

My background music is from an old pioneer tuner-only and a pair of Roland studio bi amped monitors -- ok but nothing really special. Down in the shop I'm using (gasp) 4 radio shack speakers and a homebrew TL subwoofer that is simply amazing, and even drowns out the machines as wanted -- with of all things, an Onkyo receiver. Again, nothing special but outside my property, about the best in the area, just the same. As you said, good enough to not distract one from the content.

Does anyone else find that if they play music they really dig during work, it slows them down? I like most kinds (lucky that way) but if I play my faves, I'm boppin to it and not thinking or working as well as if I just put Classical on as the background. It keeps my music sense happy without distracting my mind as much as something more "active" does.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA


Return to Legacy (Tubes!)

Who is online

Users browsing this forum: No registered users and 3 guests

cron