advertisement


What makes a great amplifier?

What makes a great amplifier is low distortion.

Amplifier distortion has been orders of magnitude below typical loudspeaker distortion for a lot longer than I’ve been alive. It is just not the bottleneck. A 1948 Leak TL12.1 gets its name from being able to deliver its (in most cases perfectly adequate) 12 Watts at less that 0.1% THD. Find me any loudspeaker regardless of price that can match that! Specs are only of use these days if you are trying to match speakers that are real outliers, and then you tend to be looking at power output, stability into specific loads, damping factor etc. Distortion was at inaudible levels 70+ years ago!

Basically amplifier design is a mature technology. The only real “gains” of recent decades are in cost-cutting, power efficiency, and ease of mass manufacturing. This to the point most sold in the consumer realm are now throw-away landfill tat that will be lucky to last a decade.
 
Amplifier distortion has been orders of magnitude below typical loudspeaker distortion for a lot longer than I’ve been alive. It is just not the bottleneck. A 1948 Leak TL12.1 gets its name from being able to deliver its (in most cases perfectly adequate) 12 Watts at less that 0.1% THD. Find me any loudspeaker regardless of price that can match that! Specs are only of use these days if you are trying to match speakers that are real outliers, and then you tend to be looking at power output, stability into specific loads, damping factor etc. Distortion was at inaudible levels 70+ years ago!

Basically amplifier design is a mature technology. The only real “gains” of recent decades are in cost-cutting, power efficiency, and ease of mass manufacturing. This to the point most sold in the consumer realm are now throw-away landfill tat that will be lucky to last a decade.

Even though amplifier distortion has been orders of magnitude below typical loudspeaker distortion we can still hear differences between amplifiers.
Besides, there are several types of distortion. Which "Distortion was at inaudible levels 70+ years ago!"?
 
0.1% is a reasonable figure for what should be the maximum acceptable distortion. There are loads of caveats to that though! If that is from a well designed class A amplifier with fairly low NFB (push pull of course!! Single ended is a non starter for power amps in many ways) then end of story... If it's from a high feedback, cool running, class A/B amplifier it may be unacceptable as a good part of that TOTAL Harmonic Distortion could be high order distortion which is vastly more disagreeable to the ear than low order distortion and noticeable at far lower levels.

Whilst amplifiers are a mature technology there are still plenty of new tricks from the old dog yet!
 
…Single ended is a non starter for power amps in many ways

Again, one has to view things from a full system context, and if one is partnering a high quality SET with real high-efficiency speakers (e.g. Avant Garde, Cessaro etc) chances are the overall system distortion is vastly less than a typical high-spec class AB amp driving a typical box speaker. There are countless perfectly valid ways to approach audio! I far prefer not to have a dog in any specific race and just enjoy what each design school has to offer. There is a whole world of amazing kit out there and I can find a lot to like in a lot of different places!
 
Isn't the job of an amplifier to increase the amplitude of the input signal?
If so then the idea of "a straight wire with gain" doesn't sound too off, if you accept that cables won't add nor remove anything from the signal.

The problem is that doesn’t tell you anything about any given amplifier... it might be an ideal scenario but it doesn’t exist.
 
Again, one has to view things from a full system context, and if one is partnering a high quality SET with real high-efficiency speakers (e.g. Avant Garde, Cessaro etc) chances are the overall system distortion is vastly less than a typical high-spec class AB amp driving a typical box speaker. There are countless perfectly valid ways to approach audio! I far prefer not to have a dog in any specific race and just enjoy what each design school has to offer. There is a whole world of amazing kit out there!

Push pull amps ALWAYS give lower distortion than single ended (check out the data sheet for any output valve). If one insists on triode valves in class A, and why not, (single ended has to be class A), <20W and zero feedback then typically the push pull will give about 0.8 to 2.5% THD and the single ended about 4 to 10% THD all else being equal. There are no down sides and for the same HT voltage the push pull amp will also give greater power. Add a bit of NFB to that class A triode push pull amp and voila... Leak TL12 and Williamson amplifier!

EG for 300B SE max power 7W and THD 5% @350V HT but for a push pull pair at the same HT voltage 20W and 2% THD. From Western Electric 300B datasheet.

There are many valid approaches to audio yes but some are intrinsically better than others and many dictated by fads and who's shouting loudest!
 
Push pull amps ALWAYS give lower distortion than single ended (check out the data sheet for any output valve). If one insists on triode valves in class A, and why not, (single ended has to be class A), <20W and zero feedback then typically the push pull will give about 0.8 to 2.5% THD and the single ended about 4 to 10% THD all else being equal. There are no down sides and for the same HT voltage the push pull amp will also give greater power. Add a bit of NFB to that class A triode push pull amp and voila... Leak TL12 and Williamson amplifier!

EG for 300B SE max power 7W and THD 5% @350V HT but for a push pull pair at the same HT voltage 20W and 2% THD. From Western Electric 300B datasheet.

There are many valid approaches to audio yes but some are intrinsically better than others and many dictated by fads and who's shouting loudest!

There is one problem apparently with push pull valve amplifiers that single ended amplifiers don’t have which was explained to me by someone who has designed his own transformers and valve amplifier. Apparently the metal used in even expensive amplifiers has non linearities in the output transformer B/H curve at low levels, this leads to higher distortion at lower levels similar to crossover distortion in transistor amplifiers. He kindly gave me the design for the transformers but because the metals are not standard the cost for a pair of output transformers is going to be over 1K as there is an MOQ of about 10Kg on the metal, I will probably get some of that back as he has promised to publish the design and the manufacturer has promised to reimburse me if he sells transformers using them. 1K is a lot of wedge for a couple of output transformers but a hobby is a hobby….
 
There are many valid approaches to audio yes but some are intrinsically better than others and many dictated by fads and who's shouting loudest!

I have no ego, ideology or agenda in this. I’m certainly not trying to sell anything (beyond maybe a greener and more sustainable way of looking at audio). I can say some of the very best systems I’ve ever heard are SETs into giant horns. Ok, they usually tend to cost as much as a half-reasonable house, but I can certainly see the good there and why so many view it as such a valid approach. As I can with pretty much every other topology if well executed in its right context, e.g. I love a really good solid state amp into good panel speakers too, which one could argue was the polar opposite. I’ve heard so many systems I’ve enjoyed over the past 45 years or so of such profoundly differing design ideology I now just refuse to be drawn to anything more than ‘I like that system’ vs. ‘that one wasn’t for me at all’. I stopped shaking my fist at clouds decades ago. Whatever the design topology if it is genuinely nicely made, serviceable, people like it, and will stay out of landfill for a good few generations then I wholeheartedly approve.
 
There is one problem apparently with push pull valve amplifiers that single ended amplifiers don’t have which was explained to me by someone who has designed his own transformers and valve amplifier. Apparently the metal used in even expensive amplifiers has non linearities in the output transformer B/H curve at low levels, this leads to higher distortion at lower levels similar to crossover distortion in transistor amplifiers. He kindly gave me the design for the transformers but because the metals are not standard the cost for a pair of output transformers is going to be over 1K as there is an MOQ of about 10Kg on the metal, I will probably get some of that back as he has promised to publish the design and the manufacturer has promised to reimburse me if he sells transformers using them. 1K is a lot of wedge for a couple of output transformers but a hobby is a hobby….

That may be apocryphal... Generally speaking SE output transformers have much more distortion than push pull ones. This is well known and there are many articles about it "in the literature".
The distortion figures from the 300B datasheet will be calculated and not include the distortion from the OPT... I'll guesstimate an extra 0.03% from the push pull OPT and 1-2% for the SE one. Both much worse at very low frequencies of course.
 
That may be apocryphal... Generally speaking SE output transformers have much more distortion than push pull ones. This is well known and there are many articles about it "in the literature".
The distortion figures from the 300B datasheet will be calculated and not include the distortion from the OPT... I'll guesstimate an extra 0.03% from the push pull OPT and 1-2% for the SE one. Both much worse at very low frequencies of course.


What do you mean apocryphal?, it’s what he has measured and calculated!
 
I have no ego, ideology or agenda in this. I’m certainly not trying to sell anything (beyond maybe a greener and more sustainable way of looking at audio). I can say some of the very best systems I’ve ever heard are SETs into giant horns. Ok, they usually tend to cost as much as a half-reasonable house, but I can certainly see the good there and why so many view it as such a valid approach. As I can with pretty much every other topology if well executed in its right context, e.g. I love a really good solid state amp into good panel speakers too, which one could argue was the polar opposite. I’ve heard so many systems I’ve enjoyed over the past 45 years or so of such profoundly differing design ideology I now just refuse to be drawn to anything more than ‘I like that system’ vs. ‘that one wasn’t for me at all’. I stopped shaking my fist at clouds decades ago. Whatever the design topology if it is genuinely nicely made, serviceable, people like it, and will stay out of landfill for a good few generations then I wholeheartedly approve.

I don't make valve power amps and so have no "dog in the fight".. and note that I agreed that there are many valid approaches. If, heaven forbid, you'd simply replied something like "Ah that was an interesting technical point there Jez. Yes the Western Electric figures support what you are saying"... I'd have checked it wasn't raining frogs:rolleyes:
 
What do you mean apocryphal?, it’s what he has measured and calculated!

Oh it could be correct yes but if so I've never seen such a thing reported before but have read many articles describing why SET OPT's have much higher distortion than PP ones... that's all I'm saying:)
The design of the very best OPT's has always been a bit of a "black art"!
 
If, heaven forbid, you'd simply replied something like "Ah that was an interesting technical point there Jez. Yes the Western Electric figures support what you are saying"... I'd have checked it wasn't raining frogs:rolleyes:

Obviously I understand that. I just view things in a wider context and fully respect freedom of choice, e.g. I grasp that partnered with say a huge pair of Western Electric cinema horns you’d have a system with a scale, dynamic range, life and vitality that some folk will value above all other priorities. Dynamic compression is distortion too, yet no one talks about that as all domestic audio that ends in little boxes does it. I’m certainly not egotistical or arrogant enough to try to tell them they are wrong. It is a perfectly valid system and as a jazz fan (i.e. exceptionally dynamic music) I totally get why people would chase that kind of presentation. It’s not necessarily for me even if I had the space to house it, and clearly not for you, but it is perfectly ok for folk to like.
 
Isn't the job of an amplifier to increase the amplitude of the input signal?
If so then the idea of "a straight wire with gain" doesn't sound too off, if you accept that cables won't add nor remove anything from the signal.

That is the job, but an amplifier can't be thought of in isolation, only in relation to the load it will be up against. Maybe there's a (somewhat tenuous) analogy to be made with what Mike Tyson said, about everybody having a plan until they're punched in the face. An amplifier design might be a straight wire with gain on paper, until it comes up against a real loudspeaker. That's when its character becomes apparent.
 
Wasn’t it Alan Shaw of Harbeth who said that any differences between similarly specified amplifiers were inaudible and put a pair of not inexpensive speakers on the line? AFAIK, no-one took him up on the challenge
 
Wasn’t it Alan Shaw of Harbeth who said that any differences between similarly specified amplifiers were inaudible and put a pair of not inexpensive speakers on the line?

It struck me (and many others) as being rather similar to the James Randi cable dem in that the criteria was highly nailed-down and the test methodology very rigid. IIRC they both wanted to vet what items they considered ‘similarly specified’ and I know Alan Shaw wanted to place his own switching electronics into the signal path.

I’m not one who argues that things that measure identically sound different. If I was in the market for a new amplifier I’d be listening to things like Leben, Sugden, Accuphase, Shindo, Conrad Johnson etc and I’d fully expect to be able to tell them apart as they exist in different design schools, use different technology and suit different speakers. I’d not be that interested in comparing say £500-800 class A/B integrated amps as I’d expect them all to be much of a muchness. My impression is that is what the Harbeth ‘test’ was, just as Randi’s cable “challenge” famously disallowed the cables Michael Fremer was confident he could tell apart. Basically file under ‘marketing’!
 
I take your point but what skin did Harbeth have in the game? Or did it reflect well on their speakers in some way?
 
I take your point but what skin did Harbeth have in the game? Or did it reflect well on their speakers in some way?

He sells loudspeakers, and argues that any decent amp is fine. It is clearly in his interest to argue that say a £3.5k pair of his speakers on the end of a £500 amp is a ‘balanced system’ as he gets seven eighths of that budget! It appears that his customer base don’t buy it as most Harbeths are partnered with similar price-band kit. I’m not even arguing he is wrong; his speakers are not hard to drive and sound great on a second-hand Quad or whatever. In fact I think he uses an old 606 as his design amp. I’ll always argue for synergy, and given amps are IMHO a mature technology as long as it is happy into the speaker load and can deliver the required power there is just no need for overkill.
 


advertisement


Back
Top