Tube power vs solid state power

[teemuk]

Ah, this age old discussion always following the same paths.

Myth #1: Tubes clip softly.
Some tubes do, some tubes do not. Overall it also depends very much on the design. Few minutes spend with some classic tube amps and an oscilloscope will pretty much prove that tube amps do not really clip all that softly.

Myth #2: Solid-state amps clip hard and sound harsh.
Isn't always the case either. Soft clipping solid-state amp designs have been around at least since the mid 1960's (e.g. Thomas Organ Vox amps) and they are increasing by amount all the time. Today, with a decent design in hand, you really can't tell a difference to a tube amp in a blind test.

Myth #3: Tube amps are louder than equally rated solid-state amps.
Some are, some are not. As much as I've heard this cliche repeated, I have heard people saying that solid-state amp x was as louder (or louder) than an equally rated tube amp. Enough to to be certain that you really can't regard the claim about tube loudness as universal truth. So, it depends on design, speakers, etc. Naturally. Similarly, some equally rated tube amps can be louder than some other tube amps. It depends.

Myth #4: The harmonics thing.
I'm surprised that no one seemed to mention it yet in this thread. Of course the entire tale about tubes and solid-state amps producing different harmonic content is nothing but hogwash.

[Old Tele man]

...McDonald's chicken and amplifiers are kinda/sorta related:

• CHICKEN: "...parts is parts..."
• AMPLIFIERS: "...watts is watts..."

[sprinter]

What a variety of great answers. Thanks.

Judging from these posts, plus the mere fact that there has been so much debate among experienced guitarists for so long, and that there are so many variables in amp design, so far I take it that it all comes down to one thing: Personal Preference. What a concept!

I have heard of some blind tests (maybe not very scientifically done) that suggest that it's difficult or impossible to tell the difference, but even if it's only a matter of subjective perception, as it is often said, "Perception is Reality".

[Steve Conner]

Hogwash? I think that's a little unfair. You can design a SS amp to have the same harmonic spectrum vs. Overdrive characteristic as a tube one. But many of the early SS amps weren't done that way. I once took apart an old CBS Fender London Reverb, and for the dirty channel they just stuck in Craig Anderton's circuit with the hex inverter chip. Something like Peavey's Transtube system is a lot more sophisticated, but the damage was already done by the previous generation of amps.

Also: tube amps look like they clip harshly on a dummy load, but have you ever got your ear muffs on and tried it with a speaker load? The tube amp clips by running out of current. The SS one hits the rail and runs out of voltage.

[olddawg]

Fet version of the Marshall Superlead Plexi 40W Here, figure it out for yourself.

As far as I can tell, it's normal Class-AB. The guy developed a MOSFET building block that emulates a tube, and he uses one for each tube stage in the 1959SLP. It even has an output transformer.

He's either a genius or completely insane, or both.

It's not really a new thing. I have an 80s vintage Lead 100 Mosfet Marshall amp. It was supposed to be the SS answer to the JCM800. It is a decent amp and f'ing loud as f. I actually like it better than the more modern Marshall SS amps, but it doesn't sound like a good all tube amp. It is a decent clean pedal platform with a usable reverb. I gutted the box and built an 18 watt Marshall variant/clone into it. The Mosfet amp is on a shelf in a box in the garage.

[Steve Conner]

Those Marshall MOSFET things are an example of what I'm talking about. They're just a bunch of op-amps in a box with no attempt at emulating tube behaviour at all. The fact that any ever sold is proof to me that the Marshall tone is actually in the speaker. Or possibly the badge.

KMG's Plexi thing is altogether more advanced, each stage of the circuit is a reasonable emulation of the corresponding tube stage in a Plexi.

[teemuk]

Also: tube amps look like they clip harshly on a dummy load, but have you ever got your ear muffs on and tried it with a speaker load? The tube amp clips by running out of current. The SS one hits the rail and runs out of voltage.

Nevertheless, both clip. The harshness just isn't so much revealed by a tube amp running to a reactive load because the non-linear frequency response (generated by the interaction of high output impedance and speaker's non-linear impedance) won't allow the amp to output "square waves" like it would do with a resistive load. The underlaying knee of clipping doesn't, however, change one bit. The tubes still clip hard but the output waveform is such that it doesn't reveal it as evidently.

And... You can get the same type of behaviour from a solid-state amp that has a high output impedance. The modern Transtube Peavey's being a good example, Vox Valvetronix amps another one, Randall's new hybrids third, Hughes&Kettner's Dynavalve bass amps fourth, and the list could be continued with lots and lots of more examples. You get the point. The different behavior to reactive load isn't just a "tube thing", many solid-state (instrument) amps are designed to do it as well


.

[Steve Conner]

So do these new amps actually allow the speaker to swing outside the supply rails?

If not, then they would require a bigger power supply and more transistors/heatsinking per "marketing watt", because extra headroom would have to be left to generate the spikes seen in the above pictures. Do the manufacturers really make that sacrifice?

As far as I know, they don't, and that is how the tube watts thing came about. Tube amp designers are forced to make the sacrifice by the inefficient nature of tubes.

[Enzo]

But the Peavey transtubes are just in the preamps, the power amps driving the speakers are plain old conventional push-pull transistor affairs.

And when you say

tubes still clip hard but the output waveform is such that it doesn't reveal it as evidently.

Isn't that saying that there is indeed a difference in performance?

[teemuk]

Do the manufacturers really make that sacrifice?

Yes, some of them do. Naturally not all of them.

But some do.

Of course, it only makes sense in the instrument amplification scene because HiFi amps with high percentage of distortion, poor headroom, low power and non-linear response are not, uh, very HiFi.

But the Peavey transtubes are just in the preamps, the power amps driving the speakers are plain old conventional push-pull transistor affairs.

No they are not and you should know it.

Yes, the output stage may be a generic, conventional circuit element (there’s not that many ways to actually make a solid-state amp), but everything else is not that typical for a solid-state power amp: It incorporates a soft clipping circuit with crossover bias shift scheme, as well as a (global) current feedback loop to make that stage respond like it was a high-impedance amp driving the inductive load of a loudspeaker. It’s clearly a design that is aimed to produce the “flaws” of tube power amplifiers. Several elements of the Transtube power amp design are also patented. It’s not a conventional solid-state power amp design by a long shot.

Edit:
This is one example of a "TransTube" series power amp. I chose it because it simplifies things by using a LM3886 integrated power amp chip for voltage and current amplifying purposes. Many other higher power Peavey's use a discrete circuit instead. Anyway, that is a generic solid-state power amp block. BUT everything else you see is also part of the power amp - important part - and basically makes the circuit as unique as it is.

[loudthud]

Also: tube amps look like they clip harshly on a dummy load, but have you ever got your ear muffs on and tried it with a speaker load? The tube amp clips by running out of current. The SS one hits the rail and runs out of voltage.

This is the most intelligent thing I've seen in this thread so far. The answer is that a tube amp driven to clipping will apply quite a bit more voltage into a speaker than it will a dummy load. Connect the output of a tube amp to an A/B switch with a dummy load and a speaker. Observe the output of the amp with an oscilloscope. Put in your ear plugs and slam a tube amp with a power cord. Into the speaker the peak to peak voltage will be something like 1.2 to 1.4 times what you see when the resistor is connected. That's 2 to 3 dB. With most solid state amps, the increase will only be 5%. More voltage applied to the speaker means it louder.

[Wilder Amplification]

This is the most intelligent thing I've seen in this thread so far. The answer is that a tube amp driven to clipping will apply quite a bit more voltage into a speaker than it will a dummy load. Connect the output of a tube amp to an A/B switch with a dummy load and a speaker. Observe the output of the amp with an oscilloscope. Put in your ear plugs and slam a tube amp with a power cord. Into the speaker the peak to peak voltage will be something like 1.2 to 1.4 times what you see when the resistor is connected. That's 2 to 3 dB. With most solid state amps, the increase will only be 5%. More voltage applied to the speaker means it louder.

Yes but isn't that increase you see the back EMF from the speaker? I would think so as a DC coupled SS amp would have a much better damping factor to dissipate the back EMF from the speaker than a transformer coupled valve amplifier would.

[loudthud]

Yes but isn't that increase you see the back EMF from the speaker? I would think so as a DC coupled SS amp would have a much better damping factor to dissipate the back EMF from the speaker than a transformer coupled valve amplifier would.

I'm not sure exactly what "Back EMF" is or if there is a universally accepted definition. I think of things in the Time Domain. In a reactive load, peak current and peak voltage do no occur at the same time. This allows the voltage on a tube's plate to swing closer to ground (than it will with a resistive load) because the load does not demand much current at the voltage peak.

I have seen speakers generate voltage spikes of hundreds of volts when driven by an amp that has a crossover dead zone. The bias has shifted so far that the amp is operating class C where for a short time both output devices are biased off. These spikes are of short duration. They are blamed for damage to output transformers and tube sockets where arcs breakdown the insulation. I don't believe these spikes contribute to tone or volume but I could be wrong.

Regardless of the reason, 20% to 40% more voltage at the speaker terminals means more volume.

[teemuk]

Regardless of the reason, 20% to 40% more voltage at the speaker terminals means more volume.

Loudspeakers are not voltage-driven devices, they are current-driven. While the voltage at speaker terminals may have huge peaks due to reactive nature of the load, the current doesn't. The total power being generated and driving the speaker speaker isn't following the voltage waveform, like it would do with a resistive load, so in the end there really is no 20 to 40 % of increase in anything meaningful considering overall speaker drive (speaker sensitivity is not rated dB/V but dB/W) . Those same reactive characteristics that make the voltage waveform spiky also reduce the current and level out the output power during those peaks.

[Steve Conner]

(speaker sensitivity is not rated dB/V but dB/W)

Ah, but it actually is dB/V. This is one of the great mysteries (scams, even? ) of the speaker industry. The "1W" they use for sensitivity and frequency response characterisation is actually 2.83V delivered from an amp with a low output impedance. That is the voltage that gives exactly 1W into an 8 ohm resistive load.

However, because the impedance of the speaker is not 8 ohms, the power actually delivered to the speaker during the test is not 1W either. It's much less at most frequencies. The speaker relies on this deficit of power to produce a flat frequency response: it's just another way of looking at what we call damping.

It also relies on the deficit of power to keep it in one piece. A 500 watt woofer is one that can withstand being connected to a solid-state amp set to deliver 500W of pink noise into an 8 ohm load, as set out in the IHF test procedure. It can't, and won't, actually absorb 500W of real thermodynamic power.

This applies especially to the kinds of subwoofers used in car audio, which are a strongly reactive load in the bass region, because of their powerful motors and incredibly poor matching to the air load. You can actually plug one into the wall, and the cone will flap like crazy, but the voice coil won't explode and burn like an 8 ohm resistor would.

Because of this, and the other points made above, I maintain that a tube amp actually delivers more real power into a speaker load, and to emulate this, a solid-state amp needs not just a high output impedance, but more headroom built into the rails, and clipping through current limiting, set to operate below that headroom. The large-signal behaviour has to be right as well as the small-signal stuff.

The extra headroom costs money, but gives no measurable increase in output power on a dummy load test, which is how power is measured for advertising purposes. So nobody does it. And this is the reason why "tube watts" are louder.

This graph was sent to me by one of the guys over at ssguitar.com. It shows how a current-drive power amp reacts to a speaker load vs. a dummy load. I guess this is only a small-signal test, though. If my argument were right, the peaks in the upper curve would disappear if he mashed the amp into clipping, assuming the frequency response could even be measured under those conditions.