SET amp owners thread

[DaveC]

I have noticed this largely with pro drivers or ultra-high sensitivity cones (as you say AER seem to be an exception). My Supravox 215-2000 are very stiff (doped accordion style surround) and hardly move and with a Qts or only a bit over 0.2 they would likely sound overdamped with SS.

IME it's Supravox that are going to be the exception. Omega, Feastrex, AER, Vox and pretty much every other modern single driver has a suspension that's very flexible.

It could be modern high strength motors have made highly damped drivers unnecessary, or maybe it was still a valid approach even with old field-coil motors, I'm not sure but imo it does lead to better results vs a highly damped driver.

With modern woofers it depends on design approach, heavy/stiff cones will require a more substantial suspension. For a dedicated woofer I prefer a rigid cone, that leads to a more damped suspension, a larger motor and a big SS amp. Unless you have the space and cash for multiple large drivers and/or bass horns.

 

[Atmasphere]

Ralph,

The "power paradigm" is nothing new - the rules of power transfer are studied in basic electronics.

Yes. I've made that point myself many times.

IMHO your paradigm page needs some links to the sources you refer. Surely just suggesting, YMMV.

I agree- its been needing that for some time. I've been working on updating it.

What you are saying is incorrect Ralph, the fact that a SET might have higher 3rd harmonic than a PP amp is not relevant it is the relationship with the other harmonics that are far relevant (i.e. the pattern of the distortion). This was long ago recognized by engineers in the BBC and later Jean Hiraga that a monotonic harmonic distortion pattern is more pleasing to the ear because that is more typical of what is observed in nature. A PP amp cancels even order harmonics and this is perceived as unnatural.

There was a test performed by Keith Howard that demonstrated, by digitally adding distortion to recordings, that the undistorted recording was best; however, of distortion patterns that typical electronics make the next best sounding to undistorted was a monotonic decay pattern with an exponential decay from 2nd to 3rd to 4th etc. What sounded distinctly worse was an all odd pattern of 3rd, 5th, 7th etc. that is typical of push/pull amps where even harmonics get cancelled. Somewhere in between was an all even pattern of 2nd, 4th 6th etc. which is a highly unlikely pattern to get as there is not an obvious mechanism. This was with all patterns decaying exponentially, which is not the actual case in a lot of amps where the pattern look more like a picket fence out to 20th order.

So, please don't oversimplify the situation with some easy heuristics that 2nd and 3rd are the same as if they are ever in isolation in an amp. The overall pattern is very important (apparently the phase of the harmonics as well, which is not obvious from a FFT spectrum).

I agree and I'm not. I'm simply pointing out that the 2nd and 3rd are treated the same by the ear. Harmonics above the 4th are treated very differently as we both know!

I should point out that the BBC tests were not of an amplifier that was fully differential- they tested amps that had single-ended inputs and push-pull outputs (even though they may have had balanced inputs; that was done via an input transformer). Most of them were also pentode or tetrode based and employed feedback (which, being insufficient to do the job properly, added higher ordered harmonic distortions of its own). IOW, apples and oranges. A push-pull amplifier with a single-ended input will have a bit of a 5th that to most SET owners sounds out of place, but it also isn't there at that level in a fully differential amplifier. IOW the topology of the amplifier is playing a role here, and the tests to which you refer ignore that.

In a zero feedback amplifier that is fully differential, the 3rd harmonic is the predominate distortion component, but some 2nd is unavoidable and the even orders above that are present too. What is different is that an amplifier that has primarily a cubic non-linearity (3rd harmonic) as opposed to a quadratic non-linearity (2nd harmonic) will have lower distortion across the board as its distortion components drop off more rapidly as the order is increased. The reduced amount of higher ordered harmonics, masked by the lower orders, results in an amplifier that is less harsh and smoother in sound, despite also being lower distortion

No amplifier I'm aware of makes only the odd ordered harmonics; I do agree that such an amp would be unpleasant. Keith Howard's test was a bit too simplistic in that regard or put another way, simply doesn't relate to actual amplifiers as real world amps are a bit more complicated than that.

 

[morricab]

Yes. I've made that point myself many times.


I agree- its been needing that for some time. I've been working on updating it.



I agree and I'm not. I'm simply pointing out that the 2nd and 3rd are treated the same by the ear. Harmonics above the 4th are treated very differently as we both know!

I should point out that the BBC tests were not of an amplifier that was fully differential- they tested amps that had single-ended inputs and push-pull outputs (even though they may have had balanced inputs; that was done via an input transformer). Most of them were also pentode or tetrode based and employed feedback (which, being insufficient to do the job properly, added higher ordered harmonic distortions of its own). IOW, apples and oranges. A push-pull amplifier with a single-ended input will have a bit of a 5th that to most SET owners sounds out of place, but it also isn't there at that level in a fully differential amplifier. IOW the topology of the amplifier is playing a role here, and the tests to which you refer ignore that.

In a zero feedback amplifier that is fully differential, the 3rd harmonic is the predominate distortion component, but some 2nd is unavoidable and the even orders above that are present too. What is different is that an amplifier that has primarily a cubic non-linearity (3rd harmonic) as opposed to a quadratic non-linearity (2nd harmonic) will have lower distortion across the board as its distortion components drop off more rapidly as the order is increased. The reduced amount of higher ordered harmonics, masked by the lower orders, results in an amplifier that is less harsh and smoother in sound, despite also being lower distortion

No amplifier I'm aware of makes only the odd ordered harmonics; I do agree that such an amp would be unpleasant. Keith Howard's test was a bit too simplistic in that regard or put another way, simply doesn't relate to actual amplifiers as real world amps are a bit more complicated than that.

D.E.L Shorter was one of the early pioneers in psychoacoustics with regard to making an effort to correlate the harmonics an amp makes with a general sound quality assessment "The influence of high order products in non-linear distortion". He came up with a harmonic weighting factor, which took the square of the harmonic order/2. It appears to not have been a heavy enough weighting as was found by Cheever and Geddes, which have much stronger weighting as the order increases.

I have seen many amp measurements that were predominantly odd order with only a bit of 2nd at the beginning of the sequence and virtually nothing beyond 4th or 6th...the rest odd (most amps have harmonics all the way to 20th and beyond that rises above the noise floor).
https://www.soundstagenetwork.com/measurements/atmasphere_ma1_mkii2/
This example of your work has harmonics that strong up to about 7th and then it drops of suddenly. A tiny bit of 9th and then some odd harmonics popping up further out (can't say exactly what but likely odd order as the evens have dropped out completely by this time). Interestingly, the 4th is just as strong as the 2nd.

The problem is that if psychoacoustics are weighted very heavily on the order of the harmonic (higher order gaining increasing weight) then the drop off with increasing order has to be something like 20db per harmonic... so -40db 2nd, -60db 3rd, -80db 4th, -100db 5th etc. Very few amps do this.


Cheever also found that it was level dependent...so as you turn up the wick the sensitivity to these harmonics gets lower. So soft sounds and wide dynamic range music are more affected by this than constant loud music.

https://www.soundstagenetwork.com/measurements/tenoraudio_75wi/

This OTL interestingly has a similar distortion pattern as yours, despite using 6C33Cs in the output stage.

https://www.soundstagenetwork.com/i...ac&catid=97:amplifier-measurements&Itemid=154
Well known and respected brand with predominantly odd order (even is there but greatly diminished).

 

[morricab]

I am not discussing technically at all. You should stay away from one that brings out the fact you can't listen to acoustats and apogees clipping with SETs

Again and again you mistake your assumptions for facts...

 

[Atmasphere]

D.E.L Shorter was one of the early pioneers in psychoacoustics with regard to making an effort to correlate the harmonics an amp makes with a general sound quality assessment "The influence of high order products in non-linear distortion". He came up with a harmonic weighting factor, which took the square of the harmonic order/2. It appears to not have been a heavy enough weighting as was found by Cheever and Geddes, which have much stronger weighting as the order increases.

I have seen many amp measurements that were predominantly odd order with only a bit of 2nd at the beginning of the sequence and virtually nothing beyond 4th or 6th...the rest odd (most amps have harmonics all the way to 20th and beyond that rises above the noise floor).
https://www.soundstagenetwork.com/measurements/atmasphere_ma1_mkii2/
This example of your work has harmonics that strong up to about 7th and then it drops of suddenly. A tiny bit of 9th and then some odd harmonics popping up further out (can't say exactly what but likely odd order as the evens have dropped out completely by this time). Interestingly, the 4th is just as strong as the 2nd.

The problem is that if psychoacoustics are weighted very heavily on the order of the harmonic (higher order gaining increasing weight) then the drop off with increasing order has to be something like 20db per harmonic... so -40db 2nd, -60db 3rd, -80db 4th, -100db 5th etc. Very few amps do this.


Cheever also found that it was level dependent...so as you turn up the wick the sensitivity to these harmonics gets lower. So soft sounds and wide dynamic range music are more affected by this than constant loud music.

https://www.soundstagenetwork.com/measurements/tenoraudio_75wi/

This OTL interestingly has a similar distortion pattern as yours, despite using 6C33Cs in the output stage.

https://www.soundstagenetwork.com/i...ac&catid=97:amplifier-measurements&Itemid=154
Well known and respected brand with predominantly odd order (even is there but greatly diminished).

+1 Thanks- I didn't know Geddes was in on this!

I've been pulling for a weighting system for harmonic distortion for years. If the 2nd is a one then the 7th is a 1000; that sort of thing. But at the same time I'm not holding my breath; the last thing the industry wants is a spec sheet that lets you know how an amplifier would sound!

I think Bascomb had one of the speaker terminals at ground in that set of tests. That causes the drive to be asymmetrical, and so changes the distortion signature significantly! That was also done 21 years ago; we've done a lot to reduce distortion since then, in the output stage, in the driver and found ways to reduce IMD as well. Since the newer amps are noticeably smoother, you can understand that the higher ordered harmonics have been reduced.

I often harp about the fact that the ear is keenly sensitive to the higher orders because it uses them to sense sound pressure
(with 103dB range- it has to be *very* sensitive). And the fact that the ear assigns a tonality to all forms of distortion. In this regard it sounds like you and I are on the same page.

 

[Alrainbow]

I have noticed this largely with pro drivers or ultra-high sensitivity cones (as you say AER seem to be an exception). My Supravox 215-2000 are very stiff (doped accordion style surround) and hardly move and with a Qts or only a bit over 0.2 they would likely sound overdamped with SS.

This was mostly due to the very poor linear xmax of vintage drivers , exaggerated by poor damping factor of SET and tubes in general ..

Stiff was in ...!

 

[bonzo75]

This was mostly due to the very poor linear xmax of vintage drivers , exaggerated by poor damping factor of SET and tubes in general ..

Stiff was in ...!

? It's an advantage, not a disadvantage, that you can get good bass by running them with low watt SETs. Those drivers were better. The investment in them and the collective learning of the large firms was much higher than today. Bell labs, Altec which later morphed to JBL and whose engineers then went to Pioneer/TAD had so much collective learning and so much backing with their speakers used in every walk of life from cinema onwards, that such incentives and collective learning to produce speakers will just not exist (except for miniature speakers which have a similar big market)

 

[Alrainbow]

? It's an advantage, not a disadvantage, that you can get good bass by running them with low watt SETs. Those drivers were better. The investment in them and the collective learning of the large firms was much higher than today. Bell labs, Altec which later morphed to JBL and whose engineers then went to Pioneer/TAD had so much collective learning and so much backing with their speakers used in every walk of life from cinema onwards, that such incentives and collective learning to produce speakers will just not exist (except for miniature speakers which have a similar big market)

Not saying one way or the other has an advantage , I'm simply stating , With the very high non linear distortion of vintage drivers they had very min linear max by today’s standard, so to keep distortion down you had to have a very stiff spider to limit movement.

Making this also necessary were the low damping factor of SET amps, Due to the limited Xmax , you also needed a very large driver so 15 inchers were not uncommon for good bass , bass less than the typical modern 10 inch high linear xmax woofer of today.

So yes you have to match amp to load , SS amps should be sized to speakers with the same effort as For an SET. BTW i was told Low QTS woofers are a necessity if doing a proper vented cabinet.


IMO , A modern low QTS woofer will produce good bass on SET’s, with less distortion than vintage woofers ..

 

[Alrainbow]

BTW i love the sound of a good SET on matching speakers, but its not for everyone ..

 

[bonzo75]

BTW i love the sound of a good SET on matching speakers, but its not for everyone ..

Most speakers people use them on, it doesn't match with

 

[Alrainbow]

Hahaha , yes ...!

 

[KeithR]

? It's an advantage, not a disadvantage, that you can get good bass by running them with low watt SETs. Those drivers were better. The investment in them and the collective learning of the large firms was much higher than today. Bell labs, Altec which later morphed to JBL and whose engineers then went to Pioneer/TAD had so much collective learning and so much backing with their speakers used in every walk of life from cinema onwards, that such incentives and collective learning to produce speakers will just not exist (except for miniature speakers which have a similar big market)

but no computers. a piece of software can design a horn much better than a few dozen engineers and slide rules.

 

[Solypsa]

but no computers. a piece of software can design a horn much better than a few dozen engineers and slide rules.

Crossovers too. Yet somehow (and this applies only to best-of-breed) those old RCA, Western Electric, Siemens etc hold their own. For some listeners maybe are unsurpassed.

My take is that those talented slide rule jockeys must have had a really good idea of what they wanted out of those designs...even if they were working somewhat in the dark...

 

[bonzo75]

but no computers. a piece of software can design a horn much better than a few dozen engineers and slide rules.

Horns are simple. Once you have those high quality drivers and core design which has been proven over decades nothing fancy is required The fanciness is only for new manufacturers to try something distinct, which is usually they screw up.

 

[Alrainbow]

but no computers. a piece of software can design a horn much better than a few dozen engineers and slide rules.

We have heard those , computers like them ..!

 

[Duke LeJeune]

I didn't know Geddes was in on this!

It's called the Gedlee Metric.

It's based on two papers published by Earl Geddes and Lydia Lee:

http://www.gedlee.com/Papers/Distortion_AES_I.pdf

http://www.gedlee.com/Papers/Distortion_AES_II.pdf

As he was finishing up writing the papers, Earl remarked to me: "Now I know why you and your friends like tube amps."

 

[KeithR]

Horns are simple. Once you have those high quality drivers and core design which has been proven over decades nothing fancy is required The fanciness is only for new manufacturers to try something distinct, which is usually they screw up.

if you don’t think hifi has progressed since the 40s, then we can agree to disagree. The irony is you want Be Radian tweeter horns in your diy job.

We have *much* better materials science and can eliminate distortion to a vastly superior degree than before. Iirc, JBL themselves spent several million developing the M2 horn.

 

[Solypsa]

Be diaphrams have been around 40 years. Of course some still prefer phenolic from 75 yrs.

Certainly cnc manufacturing and other controls have improved. Certainly dsp enables certain perfections. Certainly computers crunch big numbers. Certainly there are smart designers pushing the state if the art. My perspective is that the best of them are both smart and sensible. Just like their engineering forefathers that produced lasting work.

 

[bonzo75]

if you don’t think hifi has progressed since the 40s, then we can agree to disagree. The irony is you want Be Radian tweeter horns in your diy job.

We have *much* better materials science and can eliminate distortion to a vastly superior degree than before. Iirc, JBL themselves spent several million developing the M2 horn.

Radian engineer came from Altec...JBL older horns would be better and the new M2 are sh*t.

 

[morricab]

Not saying one way or the other has an advantage , I'm simply stating , With the very high non linear distortion of vintage drivers they had very min linear max by today’s standard, so to keep distortion down you had to have a very stiff spider to limit movement.

Making this also necessary were the low damping factor of SET amps, Due to the limited Xmax , you also needed a very large driver so 15 inchers were not uncommon for good bass , bass less than the typical modern 10 inch high linear xmax woofer of today.

So yes you have to match amp to load , SS amps should be sized to speakers with the same effort as For an SET. BTW i was told Low QTS woofers are a necessity if doing a proper vented cabinet.


IMO , A modern low QTS woofer will produce good bass on SET’s, with less distortion than vintage woofers ..

A whole bunch of modern pro drivers use very stiff suspensions still with hardly any throw. The dynamikks speakers we sell use such drivers. Even the 10 inch woofer in our Athos 10 is very stiff but produces deep and powerful bass while looking like it isn’t moving at all! It is a speaker using modern pro drivers.