watts?

[teemuk]

The reactive nature of a speaker load does not alter the clipping characteristics of different circuits. The change in signal amplitude you can measure is due to interaction of amplifier’s output- and speaker’s impedance. If only small amounts of negative feedback are used the amplifier actually has more voltage gain with higher impedance loads – such as at resonance peaks of the speaker or at higher frequencies where speaker load’s inductivity becomes more prominent. Tube amplifiers have this type of characteristic inherently due to their high output impedance and low amount of global NFB. Emulating this characteristic in a solid-state amplifier is extremely simple. So simple that maybe about 90% of modern SS guitar amps incorporate it. It does not account for loudness although it has a distinct effect to frequency response and to amount of overdrive/clipping at different frequencies.

Yes. Some circuits could have a wider non-linear (saturating) region of output signal swing. After the amplifier’s linear region is exceeded most solid-state amplifiers just hard clip to rails (in other words, saturate) and their peak output power remains about the same as what it was when the amp was “clean”. However, from that point on some tube amplifiers have a broader transition to full sturation, thus they only slightly begin to compress the signal and at best they can even double their peak output power before the signal begins to show signs of abrupt clipping and “flat topping”.

While this may seem like a big deal it only accounts for a difference of about 3 dB – which is not much loudness-wise. And even before that took place we would have been able to clearly perceive that the signal has been distorted.

Yes, this smoother compression allows boosting the lower-amplitude signal portions while keeping the signal “seemingly clean”, as the signal lacks the harsher, annoying clipping artefacts that are results from the “hard knees” in the clipped waveform. This boosting thus creates a sense of increased loudness, although the peak loudness level hasn’t really increased notably. You can hear examples of this all the time in TV commercials and on radio stations. But this is really not something that you couldn’t do with an ordinary compressor/soft clipper in front of the power amp as well. Today many SS guitar amps are finally beginning to have one.

And all the time we must remember that not all tube amplifiers are even capable of this kind of smooth compression. In most cases the power supply can’t keep up with the increased current demand and the rail voltages sag causing earlier clipping – which then also makes the clipping more abrupt given that the amplitude of the driving signal does not change. Then we have the issue of different circuit topologies and tubes clipping differently. So, not all tube amps clip “heavenly”, “softly” or “transparently”. I have heard plenty of tube amplifiers that sound downright horrible when they clip. They really weren’t any better in that regard than generic SS amps.

With bass amplifiers the detrimental effects of clipping are even more prominent than with guitar amps, which shows that this whole toleration of clipping is just a subjective issue and pretty much tied to specific genres of music or types of musical instruments. We can’t talk objectively about subjective issues. That became obvious already in Steve Conner's comment about Hendrix. ...And as much as I like his guitar tone I wouldn't want to sound like him when playing classical guitar.

I still think that this whole tube watts are louder –thing is just a big myth that has been created by misunderstanding the processes involved. All these years of talking, yet no one has proved it to be true. ...Perhaps that tells something?

[Enzo]

On the street, who knows, but in tech circles I don't think the claim is that tube watts are louder. The claim is that a tube amp of a given power rating will be louder than a SS amp of similar rating. That is not the same thing.

[loudthud]

The reactive nature of a speaker load means that the load does not demand peak current and peak voltage at the same time as a resistive load would. This allows a tube to supply more voltage to the load because the current is low at the voltage peak. It is true that high impedance portions of a speaker's impedance have the same effect. The low instantaneous current demand of the load allows the plate voltage to swing closer to ground than it can with a resistive load. More voltage across a speaker means more volume.

Solid state amps attempt to emulate this behavior with current feedback. This only goes so far. What is needed is a bulk resistance in series with each output transistor (or just the speaker) to create the same effect as the "saturation" of tubes. But, this would reduce the amount of power available to a resistive load and thus the power rating.

I have a Fender tweed Deluxe. Into an 8 ohm resistor it can supply about 10VRMS, 14V peak. Into an 8 ohm speaker it can supply over 20V peak. It would take a 24W solid state amp to play that speaker as loud (+/- 1dB) as the 12W Deluxe.

There are many factors that influence how a tube amp sounds when overdriven. And some of these make an amp more suitable for one type of music over another. There is no amp that every player would prefer over all others. There are many solid state amps that meet the needs of the players that own them. But that's not the debate in this thread. I stand by my statement that (in general) a 50W tube amp will sound louder than a 50W solid state amp because the tube amp can deliver more voltage to the speaker.

[Steve Conner]

I stumbled on a circuit the other day that might be relevant. It's basically a Champ-like SE EL84 amp that drives a transistor output stage through a driver transformer, like a hybrid version of the old Thomas Vox things. It clips softly and has a high output impedance, which makes it deliver more power into reactive loads as was mentioned.

My prototype measures at just 30w into a dummy load, but it really wails. I don't think it's much different to my 40w tube amp through the same speaker.

[kg]

The reactive nature of a speaker load does not alter the clipping characteristics of different circuits. The change in signal amplitude you can measure is due to interaction of amplifier’s output- and speaker’s impedance. If only small amounts of negative feedback are used the amplifier actually has more voltage gain with higher impedance loads – such as at resonance peaks of the speaker or at higher frequencies where speaker load’s inductivity becomes more prominent. Tube amplifiers have this type of characteristic inherently due to their high output impedance and low amount of global NFB. Emulating this characteristic in a solid-state amplifier is extremely simple. So simple that maybe about 90% of modern SS guitar amps incorporate it. It does not account for loudness although it has a distinct effect to frequency response and to amount of overdrive/clipping at different frequencies.

in general i agree with this paragraph. we're basically talking about tube amps having higher output z than typical ss amps.

i also agree that the voltage gain of a non zero output z amp is effectively higher as the load impedance goes up.

however, the large and incontrovertible disparity between the power amps occurs not so much in the active devices themselves, but in the transformer coupled output stage.

let's take an extreme case--open circuit a tube amp's output transformer and run it at full tilt. how much voltage is seen at the traffo secondary? a hell of a lot more than the "rated power output" would have you believe.

do the same thing to the transformerless SS amp. how much voltage is seen there? i guarantee no more than the rail voltage.

so yes, as the load z of a higher output z amp goes up, more voltage is observed at the terminals--whether the amp is tube or SS. however, the capacity of the two different amps to supply that voltage is highly dependent on the presence of an OPT.

I still think that this whole tube watts are louder –thing is just a big myth that has been created by misunderstanding the processes involved. All these years of talking, yet no one has proved it to be true. ...Perhaps that tells something?

do you have a tube power amp? and a ss power amp with the same rated output power?

which one sounds louder when you disconnect the test gear and actually pump some music into speakers through them?

that's all the "proof" anybody needs.

[teemuk]

let's take an extreme case...

Perhaps... But if we limit this discussion to realistic circuits the tube amplifier’s output signal swing is still limited by the rail voltage and/or the capabilities of the power supply. If it wasn’t we wouldn’t even fuzz about things such as “power amp distortion” or “tube clipping”.

which one sounds louder when you disconnect the test gear and actually pump some music into speakers through them?

In equal condititions, both driving an identical speaker system with an identical signal (which by default means the same amount of output power), they are equally loud. Anyone should know and agree with that. Anything else is apples to oranges, highly subjective and depends on the design and the many, many variables introduced by it. I’m sure that I – or anyone - could find examples of both scenarios (where tube amp is either louder or quieter than the SS amp) if I would care to look into it. Truth is - I don’t. This loudness issue is not a general rule – no matter how much you people would want it to be.

---

Don't get me wrong here, though. I actually agree with many things that have come up in this thread. - It's just that I fail to see how they would make a generic rule or how they could not be applied to SS circuits if the various processes that create the sense of loudness are understood and replicated. There are various examples of this. In some cases these things won't be very easy or simple implementations, though.

---

Steve Conner, would you care to share more details of that hybrid circuit you mentioned? You got me interested.

[kg]

Perhaps... But if we limit this discussion to realistic circuits the tube amplifier’s output signal swing is still limited by the rail voltage and/or the capabilities of the power supply. If it wasn’t we wouldn’t even fuzz about things such as “power amp distortion” or “tube clipping”.

what do you mean by the phrase, "realistic circuits?"

have you seen the impedance curve of a 12" full range driver? maybe plotted one yourself?

have you contemplated what the reflected primary impedance is in that case?

the tube amps output signal swing IS NOT LIMITED by the rail voltage. it is perfectly capable of generating plate voltage swings many multiples higher than the rail voltage.

it does this by virtue of the inductive load presented by the output tranny. unless you put an OPT in a SS amp, this WILL NEVER HAPPEN.

In equal condititions, both driving an identical speaker system with an identical signal (which by default means the same amount of output power), they are equally loud. Anyone should know and agree with that.

is that even in question here?

Anything else is apples to oranges, highly subjective and depends on the design and the many, many variables introduced by it. I’m sure that I – or anyone - could find examples of both scenarios (where tube amp is either louder or quieter than the SS amp) if I would care to look into it. Truth is - I don’t. This loudness issue is not a general rule – no matter how much you people would want it to be.

i don't "want anything to be." i know what i know from decades of experience with audio. this is hardly the first time i've met an "engineer type" that can't see the forest because of the trees.

i replaced a 1.2kw ss amp with a handmade 600w tube amp. the 600w tube amp blows the ss amp completely out of the water. i did this more than 10 years ago. have you tried such a comparison?

[teemuk]

With the term “realistic circuit” I mean circuits that you encounter in real life. Not hypothetical ones (e.g. tube amps running without load etc.)

And yes, I have plotted impedances of many speakers. I have even made Thiele-Small-based SPICE models for quite a many drivers based on their datasheet information. I am perfectly aware of their impedance characteristics. Yes, at some cases the reflected impedance can be - and is - very, very high.

However, you surely are not claiming that the primary voltage could - or even should - swing an unlimited amount? If so the output transformers would be failing constantly due to arcing and there would be no point in “power scaling” circuits, which deliberately reduce B+ to achieve earlier clipping. While the inductive characteristics can allow quite high voltage swings the output signal swing, in practice, still has a limit and the B+ voltage is one of the factors defining it. This is what I was implying in the first place.

I replaced a 1.2kw ss amp with a handmade 600w tube amp. the 600w tube amp blows the ss amp completely out of the water. i did this more than 10 years ago. have you tried such a comparison?

Frankly, as far as I know, there is no scientific quantity known as “blowing something out of water”. If you want to make a point you can talk about concrete, measurable things such as watts fed to the speaker system, waveform behaviour during overdrive, distortion specs etc. Things that actually have a meaning. Now, all I know is that you simply had two amps that performed dissimilarly (not that surprising) and that you just decided that one of them was better than the other. Now the big question is, why did you prefer the tube amp and could the reasons for your preference perhaps be replicated with solid-state circuitry? (If yes, how can you be sure someone hasn’t already done it in some designs?) Answer that and we may have actually learned something aside from the fact that you preferred one amplifier over another ten years ago.

Sure, I have done similar amp comparisons – and so have my musician friends, relatives and various members of various forums that discuss guitar amps and these sorts of things. One thing in common for most of these comparisons is that they have always been very case-specific and rather subjective. The huge amount of different variables affecting these comparisons often makes them “apples to oranges” as well and therefore... well, not meaningless but you certainly can’t form a “valid rule” out of these various experiences. Any meaningfully conducted comparison will pretty much reveal things that have been documented already. So far I nor anyone else has been able to quantify any property that would make tube amplifiers inherently any better or louder than some well-designed/built SS amps. If tube amps are really so much louder (per watt) or ultimately so much better than solid-state amps then I don’t know how you explain that some of the reputedly loudest guitar amps in fact have been solid-state (e.g. Legend hybrids or 60’s Jordan amps that reputedly outpowered the Marshall stacks of The Who) and that numerous solid-state amplifiers have been used in recording guitar tracks of many highly respected records. How do you explain the numerous occasions when people tell that “this SS amp was equally loud (or louder) than a tube amp having the same amount of output power”? I could list many more of these "mysteries" but I'm sure you got the point already...

If nine apples out of ten are green it does not mean that “all apples are green” and it certainly does not mean that “all apples taste good”. It’s simply beyond my wildest imagination why so many fall to following this kind of erroneous logic when they discuss about concepts such as “tube watts”, “tube loudness” and “tube sound”.

[Steve Conner]

Well, much as I respeck the BAGA, I can't help thinking teemuk has a point. It probably all comes down to euphonic clipping and low speaker damping. I'm pretty sure I could modify a 1.2kW PA amp so that you couldn't tell it apart from the BAGA in a double-blind test at any volume.

Bob Carver did this with stereo amps. Guitar amps are harder because you'd have to model the clipping behaviour too, but I doubt it's impossible.

http://en.wikipedia.org/wiki/Bob_Car...ifier_modeling

I also think voltage spikes from the OT are a non-issue. In a single-ended design, they are important: in fact, half of the output waveform is made of these spikes, delivered from stored energy in the core. But in a push-pull, the energy storage in the ungapped OT is pretty small. Well at least you hope so in a good quality OT, and I know the BAGA has about 40lbs of Hammond's finest.

The remaining objection is that tube amps don't clip hard against the rails because they're inefficient, the tubes dropping quite a lot of voltage even when they're full on. So there is room for higher output voltages into higher impedances.

Well, you can make transistors do that too. I guess it's rarely done because watts are a marketing quantity, so the designer of a commercial amp won't want to throw any away. And not all tubes do it, either. I've seen some beam tetrodes that clip so hard they turn the peaks of the waveform inside out!

BTW, my new hybrid is the result of 10 years puzzling over the problem of the hybrid guitar amp. I've actually got quite fond of transistors, and think they have something to offer to music. I don't see anything wrong with mixing them with tubes to create new tones. I'm keeping the design secret for now

[kg]

With the term “realistic circuit” I mean circuits that you encounter in real life. Not hypothetical ones (e.g. tube amps running without load etc.)

And yes, I have plotted impedances of many speakers. I have even made Thiele-Small-based SPICE models for quite a many drivers based on their datasheet information. I am perfectly aware of their impedance characteristics. Yes, at some cases the reflected impedance can be - and is - very, very high.

i'm glad you have that experience, because that means we're on the same page.

However, you surely are not claiming that the primary voltage could - or even should - swing an unlimited amount? If so the output transformers would be failing constantly due to arcing and there would be no point in “power scaling” circuits, which deliberately reduce B+ to achieve earlier clipping. While the inductive characteristics can allow quite high voltage swings the output signal swing, in practice, still has a limit and the B+ voltage is one of the factors defining it. This is what I was implying in the first place.

of course everything has a limit.

you seem to be totally missing the implication of having an output transformer between the power tubes and the load... the supply rail no longer limits the output voltage swing.

this is utterly, completely, fundamentally different from any OTL design (which, coincidentally, describes 99.999% of the SS amps ever made). with an OTL amp, you hit the rail, you're DONE... there is NO MORE voltage available.

a ss amp rated for 100wrms into 8r will net you ~29vrms, or ~40vpp. that's all you get, even if the load Z goes above 8r. put a peak hold voltmeter on the speaker leads and monitor it. tell me when it reads 50vpp.

a 100wrms tube amp can and will generate more than 40vpp as the load Z goes above 8r (and we BOTH agree that it does). that same peak hold vmm across the speaker leads will show it to you as plain as day. 50vpp? no problem.

you NEVER SEE THIS on the test bench because we all use (basically) flat Z dummy loads when we test the amps. you need the complex load of the speaker to observe it.

Frankly, as far as I know, there is no scientific quantity known as “blowing something out of water”. If you want to make a point you can talk about concrete, measurable things such as watts fed to the speaker system, waveform behaviour during overdrive, distortion specs etc. Things that actually have a meaning.

leave it to the engineer types to discredit subjectivity... even though it is the only thing that really matters!

Now, all I know is that you simply had two amps that performed dissimilarly (not that surprising) and that you just decided that one of them was better than the other.

correct. it was not a difficult decision, even though it was a completely impractical one.

Now the big question is, why did you prefer the tube amp

flat out sounded better. wider frequency response--crisper highs, better low end. the amp was more vicious, thrashy, punchy, it had better impact, rattle your guts and shake your pant legs more balls to it.

i don't expect an EE to intuitively understand "amp balls," but i do expect a musician to.

and could the reasons for your preference perhaps be replicated with solid-state circuitry? (If yes, how can you be sure someone hasn’t already done it in some designs?) Answer that and we may have actually learned something aside from the fact that you preferred one amplifier over another ten years ago.

i have no doubt it COULD be replicated with a SS amp having a rated power output in the 2kw range.

having neither the money nor the desire to acquire such an amp i can assure you i will not be testing the hypothesis any time soon.

Sure, I have done similar amp comparisons – and so have my musician friends, relatives and various members of various forums that discuss guitar amps and these sorts of things. One thing in common for most of these comparisons is that they have always been very case-specific and rather subjective.

thank god! call me a snob, but i really hate objective musicians.

The huge amount of different variables affecting these comparisons often makes them “apples to oranges” as well and therefore... well, not meaningless but you certainly can’t form a “valid rule” out of these various experiences. Any meaningfully conducted comparison will pretty much reveal things that have been documented already. So far I nor anyone else has been able to quantify any property that would make tube amplifiers inherently any better or louder than some well-designed/built SS amps.

so because you can't find it, or understand it, it doesn't exist?

c'mon.

go play an ORIGINAL ampeg SVT (rated at 300w) and compare it to any other 300w rated ss amp.

if you can't hear/feel a difference, then there is nothing more to discuss.

it is no coincidence that the current-production ss version of the svt is rated for 450w.

If tube amps are really so much louder (per watt) or ultimately so much better than solid-state amps then I don’t know how you explain that some of the reputedly loudest guitar amps in fact have been solid-state (e.g. Legend hybrids or 60’s Jordan amps that reputedly outpowered the Marshall stacks of The Who) and that numerous solid-state amplifiers have been used in recording guitar tracks of many highly respected records.

well, first off, having a 600w tube power amp myself, i can tell you that it is one heavy son of a bitch, and extraordinarily expensive to build and maintain.

there isn't a doubt in my mind that INTO A SPEAKER LOAD it is far louder than an equivalently rated 600w ss amp. it's not even remotely close.

is is possible to build a ss amp that has greater headroom and produces the same loudness? sure it is. on paper, it will be far larger than 600w. in my experience, it would have to be around 2kw.

the fact is the tube amp works, and works well, and is easier to design/build/modify. i collected the parts for a monster ss amp that i was going to build myself but the project lost steam--there was no longer a reason to build it.

How do you explain the numerous occasions when people tell that “this SS amp was equally loud (or louder) than a tube amp having the same amount of output power”? I could list many more of these "mysteries" but I'm sure you got the point already...

are you really trying to say that there are just as many "laypeople" (ie non-engineers who aren't fettered by preconceptions) who think SS amps are louder than tube amps?

does anybody on this bbs think that's true? that, rated watt for rated watt, ss amps are louder than tube amps? that ss amps have bigger balls than tube amps? that ss amps watts are hairier than tube amp watts?

(keep in mind that the largest commercially available tube amps peak out around 400wrms, and the vast majority are <120wrms, whereas many ss amps, especially for bass guitar use, are far, far higher.)

frankly i have never met someone who thought that--watt for watt--ss beats tube.

If nine apples out of ten are green it does not mean that “all apples are green” and it certainly does not mean that “all apples taste good”. It’s simply beyond my wildest imagination why so many fall to following this kind of erroneous logic when they discuss about concepts such as “tube watts”, “tube loudness” and “tube sound”.

dude, it has been explained MANY times already, in engineering terms that should be stimulating all the proper receptors in your brain, but you just don't want to listen:

1) output transformer's inductance allows plate voltage to swing symmetrically around B+ (unlike OTL ss amps)

2) higher output impedance allows greater output terminal voltage to be generated against higher impedance loads (unlike typical low z OTL ss amps)

3) minimal or zero global negative feedback around the output stage allows graceful clipping characteristics and recovery from overdrive (unlike high GNFB OTL ss amps)

4) typical tube amp power supplies are of higher impedance than typical ss amp power supplies allowing transient power output to more greatly exceed steady state output. since the majority of music program material has a very high peak/avg ratio this can have a profound effect on perceived loudness.

========================================================

on a side note, it is funny that the very same discussions occur in the gearhead world, where i spend some of my time.

basically the argument is that HP is HP.

of course, scientifically speaking, it's perfectly true: one HP is a very strictly defined unit of power, and it can have only one meaning.

given two otherwise identical cars, you would expect the one with higher peak HP to be "faster" than the one with lower peak HP.

put the cars on the dragstrip where both versions can be allowed to stay in their peak powerband, and the hypothesis is true.

however, put the two cars on a road race circuit and you may be surprised. now, the SHAPE of the powerband becomes a much larger issue, because rpms will not be contained within such a narrow range. suddenly the car with 20 less hp but an extra 1000 rpms to the powerband becomes much faster, and a clear winner.

in fact, this is exactly what happens with "turbo" hp and "na" hp. HP for HP, the NA car WILL BE FASTER AROUND A RACETRACK. it's not even up for discussion... it just IS.

the reason is because while HP is a very useful metric of performance, it is far from the ONLY metric.

steady state amp output power in watts into a non-reactive dummy load suffers exactly the same fate. it is but one facet among many that define the performance envelope.

[kg]

Well, much as I respeck the BAGA, I can't help thinking teemuk has a point. It probably all comes down to euphonic clipping and low speaker damping. I'm pretty sure I could modify a 1.2kW PA amp so that you couldn't tell it apart from the BAGA in a double-blind test at any volume.

well i think you should do it, steve, and sell it. you'd make a lot of money!

Bob Carver did this with stereo amps. Guitar amps are harder because you'd have to model the clipping behaviour too, but I doubt it's impossible.

http://en.wikipedia.org/wiki/Bob_Car...ifier_modeling

you DO mean the same bob carver whose pinnacle of amp design was a TUBE AMP (the silver seven)?

i wonder why someone so well versed in transfer functions would ever intentionally design a tube amp in this day and age?

I also think voltage spikes from the OT are a non-issue. In a single-ended design, they are important: in fact, half of the output waveform is made of these spikes, delivered from stored energy in the core. But in a push-pull, the energy storage in the ungapped OT is pretty small.

you may want to rethink that assumption. as you know there is no fundamental difference between a SE opt and a PP opt. they both work on amp-turns, and the magneto-motive force behaves the same way. just because one has an unbalanced net flux in the core (and thus an unbalanced net energy) doesn't mean they don't store and release energy the same way.

the ability to store/release energy in the form of flyback voltage is a matter of primary inductance, and nothing more.

Well at least you hope so in a good quality OT, and I know the BAGA has about 40lbs of Hammond's finest.

The remaining objection is that tube amps don't clip hard against the rails because they're inefficient, the tubes dropping quite a lot of voltage even when they're full on. So there is room for higher output voltages into higher impedances.

that's not the objection... the objection is in the flyback energy! i can't stress how important the presence of the OPT is in this discussion. to my mind it is the singularly most important factor.

building a SS amp that used a beefy OPT would get you a LOT closer to the mark of emulating a tube amp.

manufacturers don't do that (and likely never will) because the OPT is the single most expensive component in the amp! getting rid of it is the WHOLE REASON ss is so attractive.

i have long believed that if there is any "magic" in an amp, it lives in the interstage and output transformers. there is a reason the highest quality trannies demand a king's ransom.

[teemuk]

At the risk of beating a dead horse…

1) output transformer's inductance allows plate voltage to swing symmetrically around B+ (unlike OTL ss amps)

I agree that due to OT the voltage swing doesn’t have similar mechanism of “brick wall limiting” as in OTL designs; in fact, the output signal measured from the plates – like you say - rides a top the B+. This is especially prominent in SE designs, where theoretically the amplitude of the voltage swing at each half wave can even be two times higher than the B+ voltage. But still, I think there is no question that the B+ voltage wouldn’t define a limit for the voltage swing.

Because of that limit we don’t see high voltage tube amps powered by a low voltage supply (which would be logical) but we do see designs that have variable B+, which allows dialling in earlier clipping at the power amp circuit.

Although the limiting mechanism in OT amps is different than in OTL amps the outcome is essentially about the same and at least to me that’s all that matters.

go play an ORIGINAL ampeg SVT (rated at 300w) and compare it to any other 300w rated ss amp.

if you can't hear/feel a difference, then there is nothing more to discuss.

This is the problem: Difference in what? In loudness, distortion characteristics, frequency response, behaviour of the tone control circuitry, interaction with speaker loads, etc, etc. There is no reference amplifier that would have an identical architecture but instead of tubes would be using transistors. Comparison to solid-state SVTs is totally useless as well since they differ a lot. These kinds of comparisons can only provide subjective results; you cannot make any objective rules based on them. About the only result I could likely get from such a comparison is that the SVT sounded different in many respects than the other amp, which was unidentical in the first place. How is that any useful?

is is possible to build a ss amp that has greater headroom and produces the same loudness? sure it is. on paper, it will be far larger than 600w. in my experience, it would have to be around 2kw.

No disagreement in that. To generate the larger region of non-linear “soft clipping” area before the “brick wall –type” clipping occurs the amplifier naturally must be able to push out more power. There is no device that could magically get this power from a thin air so it must come from the supply and the amplifier must be able to handle it. (Although this handling is not linearly following the input anymore). Because of that soft clipping behaviour many tube amps inherently have more power reserve after the point where output power is defined at certain % of THD whereas many SS amps would just hard clip to rails. Theoretically, if you would (or even could) linearize the tube amp further (e.g. with more NFB) its output power (at that certain amount of THD) would increase. Naturally this would also mean that the clipping would become more abrupt and the behaviour would begin to resemble that of a generic OTL SS amp. There is no reason why this very same soft clipping characteristic could not be taken into account when designing SS amps. In fact, it has been done several times already so these ideas do not just exist "on paper".

2) higher output impedance allows greater output terminal voltage to be generated against higher impedance loads (unlike typical low z OTL ss amps)

What is typical? Nearly all modern, SS, OTL - guitar - amps I’ve seen behave exactly this way as well due to current feedback. HiFi amps naturally do not skew their frequency response on purpose.

3) minimal or zero global negative feedback around the output stage allows graceful clipping characteristics and recovery from overdrive (unlike high GNFB OTL ss amps)

Yes, but this does not happen in virtually every tube amplifier. If the tubes by themselves clip abruptly – or if the circuit introduces poor clipping recovery characteristic - the lack of feedback does not - and cannot - make the clipping any softer or nicer. If there is a high amount of feedback (which admittedly is quite rare due to phase shifts caused by the OT) the clipping is abrupt just like in any high feedback amp. Now, it makes me wonder, are we comparing NFB versus non-NFB or solid-state and tube? Furthermore, the abrupt rail clipping of SS amps can be prevented by softly limiting the input signal (for example, by output signal driven limiters, or like in Transtube, or Valvereactor and so on), by fitting clipping diodes to the feedback loop (e.g. Vox Busker), or simply by using a soft-clipping voltage amplifier stage (e.g. G-K's SC & GT series, Ampeg SVT hybrids with variable B+) and so on. Clipping recovery characteristics can be enhanced by Baker clamps (or other clamps). And there are many, many more techniques and examples - all taken from commercial designs. Why are you automatically assuming that a SS amp follows some basic and simple topology?

4) typical tube amp power supplies are of higher impedance than typical ss amp power supplies…

I am not completely following you on this one… As far as I know, high power supply impedance limits the amount of power the supply can deliver, thus phenomenons such as “sagging” occur more prominently in tube amps. Not all amps have tube rectifiers either so I don’t see how the power supply is really that different in these cases. But as I said, in this one I’m simply not following you; could you – or someone - elaborate this more?

building a SS amp that used a beefy OPT would get you a LOT closer to the mark of emulating a tube amp.

manufacturers don't do that (and likely never will) because the OPT is the single most expensive component in the amp! getting rid of it is the WHOLE REASON ss is so attractive.

They don't, huh? So how do you explain that there are plenty of examples of SS amps with output transformers, e.g. various SS amps from the early sixties, Pignose 7-100, some Gallien-Krueger (or was it GMT) models I can't remember and of course the very new Warwick "Hellborg" amps. Sure it is expensive, but you should be careful when saying "never".

[Steve Conner]

Wow, this is turning into a quality thread

I still think energy storage in the OPT is a dead horse. If the OPT stored enough energy to make a difference, then the old trick of using a 220 ohm resistor across the speaker output to protect against an open speaker wouldn't work. That resistor has to absorb the stored energy. Also, amps with the flyback protection diodes across their output tubes would sound different to amps without them, as these diodes disable the energy storage mechanism by recycling the energy back into the B+ rail.

Many years back I did a series of tests to try and debunk some myths about tube amps with open speakers. I found that the resistor trick works, and the voltage doesn't rise much with it in place. Basically, I'm not convinced by the whole OPT thing, and I doubt you could tell the difference between, say, the Warwick Hellborgs and a suitably modded OTL power amp in blind tests.

Now, on to the issue of higher impedance power supplies. This may or may not be true. Tube amp supplies used to need a lot of resistance to get a decent life from the rectifier tube, but they got pretty stiff when silicon diodes came in, once the stock of old tube power transformers ran out. If it is true, then you could design your SS amp the same, with a high rail voltage but an undersized power transformer. So it can deliver high output voltage, but sag down to the same average power as a tube amp. I don't imagine the supply in Ken's BAGA sags any more than a SS amp, though.

Clipping recovery: I messed around with a few solid-state power amps this year, abusing them with a booster pedal and a power soak, so I could hear how they clipped without going deaf. Surprisingly, the old Maplin/Hitachi MOSFET power amp clipped very gracefully. I could really blast hell out of it and it never sounded outright bad.

I think this is because it doesn't have much in the way of negative feedback. Bad clipping comes from lots of loop gain at low frequencies, which makes the amp "wind up" like an integrator and stay stuck to the rail after the transient is over.

When it comes down to it, I think it's all about marketing. The big amp makers probably figured out all the stuff we're debating 10 years ago, and then did some market research and gave the market the amps that it wanted. The SS and hybrids are for the low end, and they sell like hotcakes, probably accounting for most of the amps sold. The all-tube amps are for tone purists who think solid-state is inferior. If they see transistors or chips in there, they'll convince themselves that it sounds "solid-state".

So a high-end hybrid amp wouldn't position well in the market, even if you could make it sound identical to a plexi on heat, which I don't think would be all that difficult.

My own feelings are mixed. From all the research I've done, I don't think tubes have any magical qualities that can't be emulated by solid-state circuitry. But somehow I still love them. They look great, they sound great without even trying, and they even smell good.

Oh, and finally, didn't Carver only make the Silver Seven so he could make a solid-state clone of it?

[kg]

so the consensus is that horsepower is horsepower, eh?



ok, let's get into the magnetics...

both of you seem reasonably well equipped to handle the fact that an inductor can step UP voltage, as is seen in virtually every SMPS out there.

you should also both know that the ratio of flyback voltage to supply voltage depends on and is proportional to the Q of the inductor. high Q means high flyback, and vice versa. (for those following along, Q is the ratio of reactance to resistance.)

when the secondary of the OPT is loaded by a nominally rated impedance, the reflected primary Z swamps the effect of the flyback voltage. this is because the reflected primary impedance is much lower than the inductive reactance.

as the load impedance across the secondary of the opt tends to infinity, the inductive reactance will dominate, and the Q will dictate how much primary voltage can and will be generated.

given an OPT with sufficiently high Q, you CAN swing BELOW ground with one end of the primary! since the plate voltage swings symmetrically about B+, it is also observably true that you CAN also swing higher than 2x B+!

in fact you can actually get to Q * B+.

the Q of the vast majority of guitar amp OPTs isn't very high. most of them have fairly miserable primary L, which of course limits Q, and thereby limits the flyback voltage accordingly, but they still have Qs greater than 1. some OPTs are orders of magnitude better than this.

(my daughter is having a meltdown so i'll have to continue this later...)

[loudthud]

It's been known for years that there is a poor correlation between lab measurements and perceived sound quality in the HiFi realm. The same is certainly true for guitar amps. So it can be said that a musician's preference of one instrument or piece of equipment over another may not correlate to any objective evaluation. However, if there is any doubt in the engineering community that the typical tube amp can deliver more voltage to a speaker than the equivalently rated solid state amp, run a SPICE simulation. If the models used come anywhere close to real components, I have no doubt in the results. This is not to say that a solid state amp could not be constructed to closely mimic a tube amp's behavior. I have spent many hundreds of hours and at least 10 Grand on the project. I submit the attached schematic, somewhat simplified, as an example of such a power amp intended for a 2 ohm load.