Originally Posted by arnyk
Sounds like double talk to me. Please provide a practical example of this, including frequency response and distortion measurements with loudspeaker or simulated loudspeaker loads.
Why not just build the amp with solid state devices?
Tubes have been obsoleted in audio for all but niche boutique equipment.
Whoa there! I don't know how much you studied about choosing operating conditions for triodes way back when. I thought what I was talking about was really elementary stuff. Load lines and all that. I mean, I'm no EE or anything. But I'm honestly trying to be intelligent about this stuff.
A *triode* can be very linear when you put a gigantic (theoretically infinite) load in its anode (or "plate", which is analogous to the drain in an FET). You'd also need to buffer its high impedance output to drive any solid state stages downstream, but that's not a problem if driving a vacuum tube (high impedance) circuit in the next stage.
Have you ever looked at a triode's characteristic curves? Notice how different they are from a BJT's, or an FET's? They are a different kind of device. If you draw a load line for a triode with the B+ in the lower right going to the max anode current in the upper left, you'll notice that the lower the value of the anode load resistor, the steeper the load line. (Look for Norman Crowhurst on load lines, an article from way back when). The steeper the load line, the higher the distortion (the exact same is true for transistors too). The less steep (more 'flat'), the less distortion (within the linear window of operation, of course).
Now, if you could put an infinite value anode load resistor in there, you could get a flat loadline, which should result in very, very low distortion. You can do that with a cascoded pair of high voltage, depletion mode MOSFETs as a constant current source (or CCS). (That's the same kind of thing you'll see as a constant current sink in the shared sources of an FET differential pair, or as the source load of a source follower/buffer circuit). This CCS acts like a resistor with a value of something like 3M ohms, which gives the nearly flat load line we want, but only has to drop like 50V or so across it (although can drop much more if you use the proper devices). This was not possible without contemporary high voltage-capable depletion-mode MOSFETs, because if you tried to use a 3 megohm resistor as an anode load with only 5mA going across it, it would drop 3,000,000 x .005 = 15kV !!!! OK, that's not possible in the real world. But it can be done with a transistorized CCS as the anode load.
Look at the figure on page 16 of this article about tube load lines: http://valvewizard1.webs.com/Common_Gain_Stage.pdf
See how flat the 220k load line is compared to how steep the 47k load line is? Now imagine a 5kV B+ (horizontal axis) with a 3M resistor and the load line that would make.
Does that make any sense to you?
Originally Posted by arnyk
If tubes had some kind of special advantage for low distortion, why do tubed amps generally have more
distortion than many SS amps?
If tubes had some kind of special advantage for low distortion, why is all modern distortion analysis equipment, which justifies having very low distortion based on solely solid state devices?
I ROTFL at tubed equipment designers who use all-SS gear as their reference - as their distortion analysis equipment.
OK, first -- Vacuum tubes as triodes have really low self-noise compared to transistors, so are good with very low-level stuff. Note that I do not mean pentodes! They can (note I wrote "can") have quite low distortion if used correctly. So you can have a really low noise, low distortion amplifier stage with no negative feedback loop around it. Transistors just about always require a negative feedback loop around them to be acceptably linear input-to-output. There's nothing wrong with NFB *when applied carefully and correctly*. But that ain't easy, and isn't always done right. So that is one advantage of a simple vacuum tube gain stage. It *can* be super-simple, linear and all without negative feedback to make things complicated. Not that there's anything wrong with NFB, mind you... But it is also true that many people just like the sound of a clean, simple triode gain stage that's done right. It's a nice thing, that's all.
Second, push-pull power amplifiers are NOT where tubes are at their best. The typical guitar amp tubes used in even the most expensive power amps are pretty bad. That includes all the popular types -- 6L6GC, EL34, KT88, 6550A, etc. even as triodes. All of those have pretty high 3rd order distortion, which does not cancel out with push-pull operation. They are high impedance devices, so need to be used with output transformers (OPT's) to match to 4 or 8 ohm impedance loudspeakers, which are really *expensive* to make right. You want a big OPT for high power, and it should have high primary inductance for good bass response *open loop* (before negative feedback is applied). But if you make the OPT really big and with high inductance primaries, you get bitten by the increased capacitance between windings, which degrades high frequency response all the way down to the point where it usually affects the high audio frequencies. It's a lose-lose situation there, yes it is. So yes, a 35W per channel tube power amp made with EL34's is going to perform very poorly compared to a very competently designed solid state power amp of similar power. Start looking at 100W per ch amps and there's simply no comparison any more. I won't argue that point with you, because that's the way it is.
But a phono preamp, or an output stage for a DAC? OK, I can argue that you can make a line level output stage from a triode that has (worst case) only about 0.5% second harmonic distortion and much lower 3rd HD, and only one-tenth of that in 5th HD, *without any negative feedback loop applied*. That's pretty darn linear. And it can even sound really good. I know it's silly, but I find the "sound" I get from a simple circuit with no negative feedback (NFB) loop around it to sound "relaxed" and "musically involving." If that can be made linear (low distortion) then I see no problem with doing that, if that's what some people like.
Also, there are so many bad sounding solid state audio circuits out there. For instance, I have a Panasonic SA-XR75 receiver which has what I think are fabulous sounding DAC's and power amp stages, but its analog line level inputs are ghastly sounding (and its FM tuner is miserable). It's obvious where Panasonic spent their money making this receiver, and for the asking price I think they made fine choices. But, its analog line level inputs still sound wretched. One *can* make bad sounding solid state audio circuits, even now.
Finally, yes, a typical tube power amp is a hobbyist's exercise, especially now that many relatively inexpensive HTR's (AVR?) can sound so good. If you're not building your own amps, I'd say it's best for rational people to buy as good an HTR as they can afford and put the big money into speakers. Or for a small stereo, there's gotta be a good, cheap integrated amp out there somewhere.
Have a look at Morgan Jones "Valve Amplifiers" 3rd edition and/or 4th edition for a nice introductory discussion of this stuff from a contemporary angle. It's very sensible and reasonable, and has lots of great [objective, measured] info on the current thinking in tube amp design (lots of transistorized support circuitry involved, and not just in the DC rectification circuits). It is NOT one of those silly 'audiophile boutique hobbyist' books with lots of arm-waving about the superiority of 12" fullrange speakers with alnico magnets! Really.https://www.google.com/search?q=morgan+jones+valve+amplifiers
I'm trying to show that there isn't a black-white divide here. Tubes are not all awful for everything. Solid state is not 100% superior in every way there is (at least when it comes to low frequency amplification, aka "audio"). But to make a good audio power amp to drive real world loudspeakers for little money? Heck yeah, transistors rule for that. No doubt about it. So OK? Can we be friends?
PS - What was the OP anyway? Oh yeah -- Tube friendly speakers? And this is for a push-pull 6L6 amp, with a paraphase inverter and a high impedance power supply, right? The amp's damping factor will be very low. It's output stage will have relatively high 3rd order harmonic distortion, and the paraphase inverter will probably have high 2nd harmonic distortion (but that could still sound "good"). Look for a nice sounding speaker that has a reasonably flat frequency response, reasonably wide frequency response, high sensitivity (at least 90dB/1W/1m) and a flat impedance curve (preferable at 8 ohms, not 4 ohms). If you find such a speaker at a somewhat reasonable price, PLEASE TELL ME ABOUT IT, because I don't know of one.
--Edited by rongon - 11/23/13 at 9:32pm