choosing input valve

[SY]

If absolute fidelity is your goal (e.g., a fast efficient car), why use valves (e.g., T-Bird)? Answer: fidelity has surprisingly little to do with enjoyment...

For some, perhaps, which explains the popularity of effects boxes posing as amplifiers (and there's solid state effects boxes as well). For me, no, I want fidelity in my electronics and I have no problem achieving that with tubes.

 

[Merlinb]

For me, no, I want fidelity in my electronics and I have no problem achieving that with tubes.

But no valve circuit will acheive the <0.001% THD, noise, power, efficiency, damping, and low cost of even a trivial SS amplifier. Hmm... so what you're saying is, you want some fidelity, but it doesn't have to be the most. You're happy with second best, just as long as it's not third or fourth best. What a peculiar philosophy...

 

[flatheadmurre]

But no valve circuit will acheive the <0.001% THD, noise, power, efficiency, damping, and low cost of even a trivial SS amplifier. Hmm... so what you're saying is, you want some fidelity, but it doesn't have to be the most. You're happy with second best, just as long as it's not third or fourth best. What a peculiar philosophy...

Let´s not start a war here .

If we wan´t to compare let´s start with a zero feedback pre amp.

When it comes to current amplification the SS wins end of story !

 

[SY]

If THD is the sole metric, that's indeed true. If you take the attitude that below a certain point, it's not particularly meaningful in an auditory sense (that, after all, is the purpose of these components), then that's not very important and other factors rule.

 

[DF96]

But no valve circuit will acheive the <0.001% THD, noise, power, efficiency, damping, and low cost of even a trivial SS amplifier.

If you use the word 'trivial' with anything like its correct meaning then I would suggest that a trivial valve amplifier will always outperform a trivial SS amplifier in terms of listening quality. To improve SS you have to make it much more complicated. To improve a valve amp you only have to make it a little more complicated. You can eventually get the SS to surpass any valve amp in measured performance, but by this time both amps are well past the point where you can actually hear any difference with just your ears.

 

[kstagger]

what, no love for the pentode driver?

Anyway, there seems to be two camps of thoughts when it comes to tube amps. I've heard both results to varying degrees. A 300B amp that was so colored - it must have been pumping out just gobs of 2H distortion - that the character was evident with every recording. The owner thought it was the bees knees. Or a certain manufacturer whose tube gear might as well be solid-state. At least personally, I like to find a happy medium.

 

[Merlinb]

If THD is the sole metric, that's indeed true. If you take the attitude that below a certain point, it's not particularly meaningful in an auditory sense, then that's not very important and other factors rule.

But if you ignore distortion, what other factors are left to measure fidelity?

If you use the word 'trivial' with anything like its correct meaning then I would suggest that a trivial valve amplifier will always outperform a trivial SS amplifier in terms of listening quality. To improve SS you have to make it much more complicated.

I dunno... it takes fewer than ten transistors to make an SS amp with gobs of feedback and outstanding measured performance. I consider that trivial.

 

[DF96]

With two valves you can make a trivial amplifier which needs no feedback provided you don't drive it too hard.

Ten transistors is not trivial. Four or five maybe would be.

 

[Merlinb]

With two valves you can make a trivial amplifier which needs no feedback provided you don't drive it too hard.

Ok, but exactly the same is true for transistors- e.g., Nelson Pass. I suspect the transistor version would have the lower distortion, since it would presumably have more gm available to be wasted in local degeneration.

 

[DF96]

Local degeneration raises output impedance, the opposite of what you want in a trivial amplifier.

The problem you always have with BJTs (and to a smaller extent with FETs) is that they are essentially small-signal devices. A bare BJT will give you 1% THD with 1mV peak input voltage.

 

[flatheadmurre]

The distorsion figures of a SS similar to a pentode is more of a problem if you run them zero feedback.

Not THD but the curve.

I now use Vfets for ouput that has a more triode like curve.

 

[Merlinb]

Local degeneration raises output impedance, the opposite of what you want in a trivial amplifier.

So use local shunt feedback -same as is built into every triode. In any case, why should a high output impedance matter if the distortion is ultimately lower than the valve version?

The problem you always have with BJTs (and to a smaller extent with FETs) is that they are essentially small-signal devices. A bare BJT will give you 1% THD with 1mV peak input voltage.

So attenuate the input signal.

This trivial valve amp is still likely to be much more expensive to build, and less efficient than the transistor version. In any case, people on this forum don't seem too concerned with circuit complexity- they are quite happy to add elaborate CCS loads and shunt regulators to their valve circuits. I just haven't figured out why they bother, if they are truly chasing fidelity. The same money and effort could have been spent on transistor design that would have been commensurately even better. Why claim to be chasing better and better fidelity, always trying to lower the distortion by another decimal point, and then choose to use valves that everyone knows won't hold a candle to a similarly complex transistor design?

I maintain that the reason we use valves is because they sound good despite their obvious technical inferiority. If you choose to use valves, its because you want an effects box. Fidelity is not required for the best listening experience, and therefore we are perfectly justified in deliberately choosing a 'coloured' input valve.

 

[DF96]

In any case, why should a high output impedance matter if the distortion is ultimately lower than the valve version?

Because nonlinear distortion is not the only problem. You also need a fairly flat frequency response, which for most speakers means a lowish output impedance.

So attenuate the input signal.

Have you ever heard of boring details like signal-noise ratio?

I just haven't figured out why they bother, if they are truly chasing fidelity.

That is true, you haven't figured it out.

Why claim to be chasing better and better fidelity, always trying to lower the distortion by another decimal point, and then choose to use valves that everyone knows won't hold a candle to a similarly complex transistor design?

It is mainly the SS people who strain for another zero after the decimal point, not us.

I maintain that the reason we use valves is because they sound good despite their obvious technical inferiority. If you choose to use valves, its because you want an effects box.

No, not me. If you look at some of my posts you will see that I often criticise that view, especially when people ask for 'tube sound'.

 

[SY]

But if you ignore distortion, what other factors are left to measure fidelity?

Quite a few (eg., overload margin and recovery, signal to noise, frequency response...). But once you get well beyond the performance needed to be audibly transparent, the measurement is not very important, it's a matter of choosing what you'd rather "drive." I choose tubes (well, hybrids, really- I get a lot of grief for the electronic miscegenation that characterizes my designs).

 

[rongon]

I may be wrong, misguided, or just plain ignorant, but I thought tubes are best at voltage amplification (can swing lots of volts), while transistors are better at current drive (can swing lots of milliamps). Something like a triode followed by an emitter follower makes a nice low distortion amplifier without any loop feedback. Can even do a good job driving a giant triode and an output transformer.

But I may be wrong, misguided or just plain ignorant.

 

[rmb]

Rongon: I have enjoyed tube audio for years. One advantage of a single ended tube versus a single ended SS BJT is no "pinch off" voltage..which may be on the order of 0.2 volt for germanium and 0.6 volt for silicon. The "pinch off" may be avoided depending on where the device operating point is set. You have Miller effect regardless of whether a tube (especially a triode) or a BJT or FET is used.

As far as "standards" go, look at the EIA standard for broadcasting...the

 

[rmb]

Rongon: XP glitched and posted my reply prematurely..
As far as standards go, check out the EIA standard for broadcasting...the frequency range is 30 to 15,000 Hz. The Williamson and other "ultralinear" amplifier designs had much wider frequency range and the primary limiting factor was the output transformer. I never liked single ended power amps; impedance coupling was necessary to keep the DC plate current out of the output transformer primary. DC in the output transformer primary may create core saturation with adverse effects on frequency response and dynamic range....

As far as preamps go, the op amp NE5534 is a good one....this one is widely used in commercial equipment.

 

[Merlinb]

Quite a few (eg., overload margin and recovery, signal to noise, frequency response...).

By overload margin you mean dynamic range?

Signal to noise: again nothing special about valves there. SS can do as well or better.

Frequency response: this is just a form of 'linear' distortion. Again, nothing to recommend valves over transistors here...

 

[Merlinb]

I maintain that the reason we use valves is because they sound good despite their obvious technical inferiority. If you choose to use valves, its because you want an effects box.

No, not me. If you look at some of my posts you will see that I often criticise that view, especially when people ask for 'tube sound'.

I know you (and practically everyone on here) claim not to want an effects box, but I have yet to hear a rational argument why you would deliberately handicap yourself by choosing to use valves rather than transistors. (And then often use transistor CCSs to try and scrub away all the pesky nonlinearities of valves!)

 

[SY]

If the imperfections are below audibility, how is that "handicapping"?