SS Sound vs. Tube Sound (for the umpteenth time)

[DonH50]

I refer you to Reference Data for Radio Engineers, RF Radio Concepts, Vacuum Tube Transmission Systems, or any one of a number of other texts describing the series. The factorial terms cause the distortion to decrease more rapidly than an exponential series (I believe the terms are in the denominator). We had to derive it when I was in school (decades ago); I can look it up again but do not have it memorized, just the results. I have presented it elsewhere, not sure I ever put the equations here on WBF, but it can be found many other places. Been a long time since I did a Taylor series expansion (or McLaurin? Way too long...) The actual distortion depends upon many other factors, natch.

The accuracy of the transformer, which includes the usual nonlinear B-H curves and hysteresis, is a separate issue, and I already noted that it is (IMO) the leading cause of distortion in power amps. I am familiar with YIG and other oscillators though that is not my field of expertise. IIRC, part of the tuning issue is the characteristics of the amp, not just the transformer, as their capacitance and gain is nonlinearly dependent upon voltage. The hysteresis of transformers is well-known (and quite annoying).

There are places for tubes in the world, just as there are for SS devices.

I said my preamp had very low distortion, and I am already on record numerous times stating I can appreciate the sound of a tube amp even knowing it is inaccurate. Not sure why you thought I was arguing the point, nor why the rant?

 

[valkyrie]

Well if the expansion is in fact in the denominator then the tubes will have a lower distortion. My point was that a factorial expansion expands much more quickly than an exponential expansion. The Taylor and McLaurin series expansions are rather trivial to do - as far as I can very dimly recall they proceeded in terms of fractional coefficients of sines and cosines(?). But such is not my point or my concern.

Yes transformers are very non-linear. But who can argue with the sound of a nicely working tube amplifier in full voice? Simply glorious. But there is a price to be paid for such a beast - almost like some law of Thermodynamics it seems that the euphonic distortions introduced by tube power amplifiers result in a less than stellar performance in other areas of the reproduction effort.

I myself swear by tube pre-amplifiers - low power, which seems to be the forte of tubes and no transformers to deal with. All the benefits with far fewer of the costs.

As for my rant? Well sometimes, like all people I just feel like ranting. Seems rather human that need. However nothing personal was intended or desired. If I offended you in anyway please accept my apology. Having been an engineer for many years I am used to finding fault with any position presented in public - especially any that contradict my knowledge of mathematics.

But in the case of this particular rant I am struck by the number of folks who seem to build their system on a set of mutually contradictory voicings - bright beaming speakers coupled to soggy sources and rolled off amplifiers. What really gets me going in when they claim these systems as representing truth and beauty. Maybe they have some beauty - but there is no truth in them. At least not to my ears.

As for YIG oscillators - I once had an HP engineer (back when HP built really great equipment) explain to me that to find ONE YIG CRYSTAL, that performed ideally, required testing of 50 pounds of such crystals. Course this may have been nothing more than an apology for the really high prices they got for their YIGs. Think stereo equipment is expensive? At least you get something large with blinking lights and all kinds of flashy jewelry. That little YIG, which would easily have fit in my pocket cost about $30,000 in today's dollars. Jeez Louise

 

[DonH50]

Trivial probably, but I have not had to sit down and do one (an expansion) in decades. They are based upon derivatives of a function; Taylor is about a defined point (e.g. the DC bias point of a device) while MacLaurin is about 0 (so a Taylor series about 0 is the same as the MacLaurin series). The terms in the expansion depend upon the starting function and it's derivatives, which differs for tubes and transistors (natch). That's about all I recall...

Manufacturing has improved so YIGs are not quite so pricey, but they still ain't cheap. My (very limited) experience with them is as oscillators in mw and mmw radar systems. I am no expert in them.

 

[microstrip]

(...) Yes transformers are very non-linear. But who can argue with the sound of a nicely working tube amplifier in full voice? Simply glorious. But there is a price to be paid for such a beast - almost like some law of Thermodynamics it seems that the euphonic distortions introduced by tube power amplifiers result in a less than stellar performance in other areas of the reproduction effort.

I myself swear by tube pre-amplifiers - low power, which seems to be the forte of tubes and no transformers to deal with. All the benefits with far fewer of the costs. (...)

I have supported this view for long time : using a tube preamplifier with a SS power amplifier - it sounded good ans seemed logical. But the late enhancements in modern tube power amplifiers and in SS preamplifiers have changed my mind. To the point that I now favor using either a full SS system or a full tube system, source included. The much greater lack of coloration, lower perceived noise and better dynamics of such systems exposes their ultimate sonic signatures and makes the crossed pair less "complete", and leaving a lot to the "pure" pairing. All IMHO.

 

[mep]

Too bad my LS17 and VS115 didn't sing together. The VS115 sounded much the better with the Krell KBL.

 

[opus111]

Yes transformers are very non-linear.

I can't agree with that as a blanket statement about all transformers. Did you really mean that ALL transformers are very non-linear or were you referring to a restricted set (for example, output transformers)? Seeing as you later went on to talk about preamps without transformers then probably you weren't only thinking of those.

 

[A.wayne]

Hmmm... Both tubes and SS generate odd and even harmonics. The harmonic structure is probably determined more by design and biasing than device; although there are intrinsic differences. Expanding the series for each device, you'll find bipolar transistors have an exponential characteristic, and tubes factorial, meaning fundamentally tubes have lower distortion.

Oh - sorry Don but a factorial characteristic expands to much larger values, much more quickly as "n" increases than ANY exponential series. This is a basic fact of complexity analysis in terms of defining problem limits. If in fact your "expansion of the series" for each device does yield a "exponential expansion" for solid state while tubes provide a factorial expansion then such would indicate ABSOLUTELY that the tubes are much higher in distortion. Which btw, they in fact ARE.

The real argument for employing tubes is NOT their accuracy as amplification devices - for they are not extremely accurate in a 10 octave bandwidth amplifier especially when the very non-linear coupling transformers are employed. However tubes do seem to produce a sound that human hearing (in all its glorious non-linear and completely misunderstood behavior) enjoys greatly. So the purpose of a sound system - regardless of cost - is ultimately human enjoyment. So rather than argue about the "accuracy" of a "hollow state" device vis a vi its solid state brethren it would seem far more appropriate for euphonic tube lovers to praise the characteristics of the bloated, ringy sound that tubes provide.

Something that I learned many years ago about transformers and their characteristics is that transformers are very non-linear depending on whether the field is being discharged or charged (collapsing or expanding the field). This learning was done during development of software for a certain type of missile that used a YIG (Yttrium-Iron-Garnet) oscillator - this particular crystal would oscillate with a frequency that was proportional to the B (field density) of a set of coils surrounding the crystal - compact little device about as large as the smaller (B4 I think) C ration cans. What we found was that to tune the oscillator "up" (or increase field density) in frequency was an order of magnitude faster than tuning the frequency down (decreasing field strength). In fact to "down tune" we had to create timing loops that would provide us with the "wait" factors necessary. In fact no coil (be it transformer or choke) is linear in terms of this charge/discharge phenomena.

Of course the "tube" types never mention this behavior - when in fact it is exactly this non-linearity that provides a tube amplifier with its "soggy, ringing" sound. But this gets, shall we say, glossed over in glowing descriptions of "sustains" and a "full sound".

I often hear so-called "high end" systems which generally in this era comprise a set of screechy, lean, bright, brittle speakers (sometimes costing as much as a ordinary house). This "high end" system is then coupled to soft, bloated and ringing tube amps and fed a generous diet of rolled-off vinyl to calm the result down to something approaching listenable. For if you play digital on this monstrosity (expensive monstrosity but still a monstrosity) it will take the tartar off your teeth at 10 paces.

Geez guys - accuracy? You expect accuracy from an active device that starts decaying the moment power is supplied? You expect accuracy from an amplifier that has to rely on huge piles of windings to couple to a real world load?

Yes you may get a much more enjoyable sound from your hollow state wonders - no question there. But it is NOT ABOUT ACCURACY.

Well unless you use them in pre-amp's where there are no transformers ...

Now there's a thought !!!! .......

PS: you should get out more often and stay away from screechy stereos with rolled off vinyl.....!!!! ..

 

[DonH50]

All transformers are nonlinear... If used beyond their linear operating range. Same as for any other component.

 

[opus111]

I agree - which isn't quite the same as saying they're very non-linear without qualification. So valkyrie - why single out transformers?

 

[DonH50]

Actually, if we want to be really pedantic, everything is nonlinear, just far below the level we usually notice. Except my teens; quite noticeable.

Don't many (most? all?) amps take feedback after the transformer, thus linearizing the output despite the transformer? I think that's the only way to get the output impedance as low as it is in tube amps...

 

[opus111]

Linear is the exception yes. Which is why I rather like the analogy I heard from some 'non-linear' systems theorist - 'I went to the zoo and noticed there were a lot of non-elephants there.'

 

[Atmasphere]

Perhaps, I can't offer an educated opinion based on the above "noise floor" criteria you stated.
tb1

That's OK- its not a matter of debate anyway. This has been known since the 1950s- Norman Crowhurst has a lot to say about it.

You should read the SS guy's version first , before commiting all to memory, there are oddities carefully left out ....

FWIW, I am not aware of anything that was 'carefully left out'.

/QUOTE]

Forty separate wires? I wonder why they do that? If they want separate wires there's always litz. Or are there 20 valves involved here?

Yes, 20 dual triodes, each wtih its own connection to the speaker terminals. The idea is to minimize the effects of the internal wiring.

Excellent summary. May be we can consider that tubes are large, expensive and generate a lot of heat; transistors are small, inexpensive and efficient - these practical factors also favor simple circuits with few components in tube designs and complicated circuits with many transistors in SS designs.

Some designers enjoy breaking my previous rule - just for example, Ralph Karsteen designs a differential preamplier with many tubes and Nelson Pass has designed SS amplifiers with few components.

FWIW, the MP-1 has three stages of gain between the LOMC phono inputs and the main outputs. Yes, it takes a few tubes- its all done differentially with 2-stage vaccum-tube CCS circuits, and a number of paralleled tube sections for lower noise/lower distortion/eider bandwidth.

The real argument for employing tubes is NOT their accuracy as amplification devices - for they are not extremely accurate in a 10 octave bandwidth amplifier especially when the very non-linear coupling transformers are employed. However tubes do seem to produce a sound that human hearing (in all its glorious non-linear and completely misunderstood behavior) enjoys greatly. So the purpose of a sound system - regardless of cost - is ultimately human enjoyment. So rather than argue about the "accuracy" of a "hollow state" device vis a vi its solid state brethren it would seem far more appropriate for euphonic tube lovers to praise the characteristics of the bloated, ringy sound that tubes provide.

It is not distortion if our ears do not see it that way. IOW, our ears are the most important aspect of audio. As proof, I am sure that most (except those that troll for entertainment) would agree that we would not mess with high end audio if we had no ears. They are pretty important! So really, its about obeying the *human hearing rules* rather than creating specs on paper that have no real meaning to the real world (e.i. our ears).

Of course the "tube" types never mention this behavior - when in fact it is exactly this non-linearity that provides a tube amplifier with its "soggy, ringing" sound. But this gets, shall we say, glossed over in glowing descriptions of "sustains" and a "full sound".

I often hear so-called "high end" systems which generally in this era comprise a set of screechy, lean, bright, brittle speakers (sometimes costing as much as a ordinary house). This "high end" system is then coupled to soft, bloated and ringing tube amps and fed a generous diet of rolled-off vinyl to calm the result down to something approaching listenable. For if you play digital on this monstrosity (expensive monstrosity but still a monstrosity) it will take the tartar off your teeth at 10 paces.

Geez guys - accuracy? You expect accuracy from an active device that starts decaying the moment power is supplied? You expect accuracy from an amplifier that has to rely on huge piles of windings to couple to a real world load?

Yes you may get a much more enjoyable sound from your hollow state wonders - no question there. But it is NOT ABOUT ACCURACY.

Of course I take umbridge to this (it is *indeed* about accuracy) but I do have similar things to say about transformers.... that is why we make OTLs

 

[DonH50]

Not sure what happened, but the quote above attributed to me is not mine...

Is it possible to get a graph of the output impedance of one of your amps? Privately if need be, I am just curious and would love to piddle with an OTL model. We can talk NDA if required.

Ever curious,
Don

 

[opus111]

Yes, 20 dual triodes, each wtih its own connection to the speaker terminals. The idea is to minimize the effects of the internal wiring.

That makes sense to me, given there are that many valves. I sometimes use a similar approach when I have multiple DAC chips in my DAC designs - each one has its own individual connection to ground.

 

[Atmasphere]

Not sure what happened, but the quote above attributed to me is not mine...

Is it possible to get a graph of the output impedance of one of your amps? Privately if need be, I am just curious and would love to piddle with an OTL model. We can talk NDA if required.

Ever curious,
Don

The graph is a line that looks a lot like the power bandwidth curve with different numbers. This is not true of all OTLs but ours have a direct-coupled output. As far as the actual value, depends on if you subscribe to Voltage Paradigm convention or Power Paradigm. The way output impedance is measured is quite different! We use a simple mathematic formula found in the Radiotron Designer's Guide; due to the age of the tome I would say it reflects Power Paradigm thought, which is a bit less complicated than the Voltage Paradigm approach.

 

[hvbias]

I love tubes, but two problems for me- the electrostatic speakers I like have a wildly varying impedance curve from 20-20 Khz thus they like amps with low output impedance. And secondly I always plan to have a large dedicated listening room with the equipment rack off to the side near my listening chair so I don't screw up the imaging/soundstage depth. This means a preamp that can drive long lengths of cable (like 25+ feet), so it needs to have a low output impedance as well. IMO the differences between tube gear with output Z's similar to good solid state gear (with little/no NFB) are minimal. So I've left this tubes vs SS quandary in the past and just go with what sounds best to me

 

[microstrip]

I love tubes, but two problems for me- the electrostatic speakers I like have a wildly varying impedance curve from 20-20 Khz thus they like amps with low output impedance. And secondly I always plan to have a large dedicated listening room with the equipment rack off to the side near my listening chair so I don't screw up the imaging/soundstage depth. This means a preamp that can drive long lengths of cable (like 25+ feet), so it needs to have a low output impedance as well. IMO the differences between tube gear with output Z's similar to good solid state gear (with little/no NFB) are minimal. So I've left this tubes vs SS quandary in the past and just go with what sounds best to me

Look for systems using Atma-sphere and electrostatics in audiogon. Forget about the classical recipes - ask your speaker manufacturer or good dealers for advice, see what other happy users are listening to.

 

[Gregadd]

Briian at Esential Audio is a good source for Atma-Sphere Soundlab advice.

 

[A.wayne]

The graph is a line that looks a lot like the power bandwidth curve with different numbers. This is not true of all OTLs but ours have a direct-coupled output. As far as the actual value, depends on if you subscribe to Voltage Paradigm convention or Power Paradigm. The way output impedance is measured is quite different! We use a simple mathematic formula found in the Radiotron Designer's Guide; due to the age of the tome I would say it reflects Power Paradigm thought, which is a bit less complicated than the Voltage Paradigm approach.

Whats the output impedance on your OTL's ...?

 

[Atmasphere]

I love tubes, but two problems for me- the electrostatic speakers I like have a wildly varying impedance curve from 20-20 Khz thus they like amps with low output impedance. And secondly I always plan to have a large dedicated listening room with the equipment rack off to the side near my listening chair so I don't screw up the imaging/soundstage depth. This means a preamp that can drive long lengths of cable (like 25+ feet), so it needs to have a low output impedance as well. IMO the differences between tube gear with output Z's similar to good solid state gear (with little/no NFB) are minimal. So I've left this tubes vs SS quandary in the past and just go with what sounds best to me

We make tube preamps that can drive cables of over 100 feet with ease. Further, all ESLs have widly variable impedance curves- that comes with the territory. But that does not seem to be a problem with tubes so long as the impedance does not get too low (below about 1.5 ohms @ 20KHz). See http://www.atma-sphere.com/Resources/Paradigms_in_Amplifier_Design.php for why.

Whats the output impedance on your OTL's ...?

You might want to look at the link above too. That will make it easier to understand the answer, as the Voltage Paradigm has a different definition for the term 'output impedance' than seems to be used elsewhere in electronics. It is also measured in a different way. In the Power Paradigm, the output impedance is the actual impedance of the output section and FWIW what we measure usually falls within 5-10% of what the formula predicts. So with the S-30 its about 7 ohms, the M-60 about 4 ohms, the MA-1 about 2.3 ohms, the MA-2 about 1.75 ohms and the MA-3 about 0.8 ohms.

Negative feedback is supposed to lower output impedance, but that is an example of a myth. It has no effect on actual output impedance at all.

Here is why: If an amplifier has an output impedance of X, then it can drive a certain load that will be X times Y. If the output impedance is cut in half, the amplifier output impedance would then be 1/2X and could drive a load that is 1/2(X times Y) with the same power. However in reality we find this is not true. If you have a tube amp with an output transformer that can drive an 8 ohm load with 35 watts, if you then connect it using the same tap to a 4 ohm load it will be seen to have less power. If you add feedback the power level will not increase to the level it was with the 8 ohm load, although steady-state distortion may drop.

The only way to get more power into that lower impedance is to have a different tap, or more power tubes, bigger power supplies, more heatsink area, more output transistors etc. There is something called Kirchoff's Law, which is a basic principle of electricity, which cannot be violated and is why you can't simply add feedback to get more power into a lower impedance! Thus we see that the Voltage Paradigm is using a different definition (whether admitted or not) than is seen elsewhere in the electronics industry. I use the word 'Paradigm' as those schooled in its principles will attack anything outside that platform of thought as heretical. We see that all the time in this industry- it is the source of much debate- tubes/transistors, objectivist/subjectivist, etc.