Jeez,
What's happened to this thread by rellum, dedicated to listener evaluation of the practical DIY embodiment of my apparently iconoclastic design concepts, focussing on 'intelligent design' to achieve great Hi End audio at low cost.
More impressions please - only good ones of course, like Hughs links.
Cheers,
Greg
What's happened to this thread by rellum, dedicated to listener evaluation of the practical DIY embodiment of my apparently iconoclastic design concepts, focussing on 'intelligent design' to achieve great Hi End audio at low cost.
More impressions please - only good ones of course, like Hughs links.
Cheers,
Greg
AKSA said:
So if I may be a little presumtuous and for the moment sum it up by saying that some want their gain straight and some like it with a bit of seasoning, then for those that like a bit of seasoning, where is it most important? I have assumed that it's the mid's and highs that would benefit most if this is one's thing. Can one have their cake and eat it too, i.e., SS slam and impact from 200-300hz (or where?) on down and a little sweetness above? Even tube folk usually use SS on a sub (below about 100hz). Is that intelligent design or merely a practical concession?
Sheldon
Hi Sheldon,
I don't think that's presumptuous at all - you pretty much have it, in my opinion at least.
People love the thwack of viseral, tight bass.
Only SS can really do this, and it's really important for rock and orchestral music.
The midrange contains the all important vocals, and most of the musical instrument melody line is in this region. You want a bit of color to avoid the leanness so common in good, even high end, SS. Leanness does measure well, since there is usually no distortion (straight wire with gain) but it makes the music sound mechanical, as though the vocalists are tired and the musicians are indifferent. Paradoxically these presentations also tend to fatigue the listener.
The top end is interesting, since at high frequencies the harmonics may fall outside human range at high order, particularly from the fifth and beyond - the nasty ones. This band should be relatively clean, particularly as artefacts like the crossover event can extend down into the human range to cause listener fatigue. A little H2/H3 is OK, but definitely nothing higher.
The best topologies and devices for this are, of course, the conventional single ended circuit using the vacuum triode, since the transfer function is more benign (power 1.5, Childs Law) than the transistor (e[exp]x, exponential) and thus the artefacts are lower order. The absence of a crossover event is very important too.
When you realise that the human ear can happily tolerate up to 3% H2, you understand that all this folklore about 0.005% distortion is absurd, although the qualifier is that tubes often are used without global feedback, which prevents creation by intermodulation of higher order artefacts anyway. The horrendous non-linearities of transistors are compensated by high gain, so lashes of NFB reduces this to vanishingly low levels, but still creates a tiny spray of higher order distortions to which the ear is astonishingly sensitive. It is important to understand that only feedforward REMOVES distortion; feedback merely REDUCES it, but FFW is not well understood and seldom used. Of course, huge R&D has gone into reducing distortion to almost immeasureable levels, but the fact remains that to the human ear, such infinitesimal distortions as remain can be extremely damaging to the music.
The only viable conclusion I draw from the often surgical, acerbic sounds I hear from modern SS equipment is that very low distortion outcomes do not always make a better sound, and the correlations are vague at best. This is also compounded by the recording process, which is stripping away some of the lower order harmonics which make up live music, rendering them lifeless, and dead. This could be happening for a number of reasons, but one would be the very directional nature of modern mikes - after all, the best of them incorporate vacuum tubes. It is as true today as it was in the heyday of microgroove records that the sound recording engineer is pivotal in the quality of the recording - and that typically the 'best' recordings are those that sound 'natural', and 'rich', and 'satisfying'. Listen to a good Quincy Jones recording and tell me I'm wrong!! These observations are all indicative of higher levels of lower order distortion, which clearly relate to something primordial in the way the ear apprehends recorded music......
Remember, to those for whom this is a personal affront, it is only an opinion, and nothing is more dangerous than the opinion of one man. But I do admit to enjoyment of my rational heresy........
Cheers,
Hugh
I don't think that's presumptuous at all - you pretty much have it, in my opinion at least.
People love the thwack of viseral, tight bass.
Only SS can really do this, and it's really important for rock and orchestral music.The midrange contains the all important vocals, and most of the musical instrument melody line is in this region. You want a bit of color to avoid the leanness so common in good, even high end, SS. Leanness does measure well, since there is usually no distortion (straight wire with gain) but it makes the music sound mechanical, as though the vocalists are tired and the musicians are indifferent. Paradoxically these presentations also tend to fatigue the listener.
The top end is interesting, since at high frequencies the harmonics may fall outside human range at high order, particularly from the fifth and beyond - the nasty ones. This band should be relatively clean, particularly as artefacts like the crossover event can extend down into the human range to cause listener fatigue. A little H2/H3 is OK, but definitely nothing higher.
The best topologies and devices for this are, of course, the conventional single ended circuit using the vacuum triode, since the transfer function is more benign (power 1.5, Childs Law) than the transistor (e[exp]x, exponential) and thus the artefacts are lower order. The absence of a crossover event is very important too.
When you realise that the human ear can happily tolerate up to 3% H2, you understand that all this folklore about 0.005% distortion is absurd, although the qualifier is that tubes often are used without global feedback, which prevents creation by intermodulation of higher order artefacts anyway. The horrendous non-linearities of transistors are compensated by high gain, so lashes of NFB reduces this to vanishingly low levels, but still creates a tiny spray of higher order distortions to which the ear is astonishingly sensitive. It is important to understand that only feedforward REMOVES distortion; feedback merely REDUCES it, but FFW is not well understood and seldom used. Of course, huge R&D has gone into reducing distortion to almost immeasureable levels, but the fact remains that to the human ear, such infinitesimal distortions as remain can be extremely damaging to the music.
The only viable conclusion I draw from the often surgical, acerbic sounds I hear from modern SS equipment is that very low distortion outcomes do not always make a better sound, and the correlations are vague at best. This is also compounded by the recording process, which is stripping away some of the lower order harmonics which make up live music, rendering them lifeless, and dead. This could be happening for a number of reasons, but one would be the very directional nature of modern mikes - after all, the best of them incorporate vacuum tubes. It is as true today as it was in the heyday of microgroove records that the sound recording engineer is pivotal in the quality of the recording - and that typically the 'best' recordings are those that sound 'natural', and 'rich', and 'satisfying'. Listen to a good Quincy Jones recording and tell me I'm wrong!! These observations are all indicative of higher levels of lower order distortion, which clearly relate to something primordial in the way the ear apprehends recorded music......
Remember, to those for whom this is a personal affront, it is only an opinion, and nothing is more dangerous than the opinion of one man. But I do admit to enjoyment of my rational heresy........
Cheers,
Hugh
“”This is also compounded by the recording process, which is stripping away some of the lower order harmonics which make up live music, rendering them lifeless, and dead. This could be happening for a number of reasons, but one would be the very directional nature of modern mikes””
Perhaps another reason is that the medium (air) also produces some distortion (mainly 2nd harmonic), and since modern recordings mostly place the mikes in the near field, they do not capture these distortions - as they would be perceived in a concert hall.
BTW placing mikes near the instruments also alters the real frequency balance of the individual instruments in the orchestra, as I think. In a concert hall where you hear every instrument at –lets say- 20 or more meters apart, upper frequencies are somewhat dimmed (the air has greater resistance as the frequency rises), but in our recordings we mostly hear more upper frequencies – which is perhaps good for the ‘detail’, but mostly unnatural.
Regards,
Thalis
Perhaps another reason is that the medium (air) also produces some distortion (mainly 2nd harmonic), and since modern recordings mostly place the mikes in the near field, they do not capture these distortions - as they would be perceived in a concert hall.
BTW placing mikes near the instruments also alters the real frequency balance of the individual instruments in the orchestra, as I think. In a concert hall where you hear every instrument at –lets say- 20 or more meters apart, upper frequencies are somewhat dimmed (the air has greater resistance as the frequency rises), but in our recordings we mostly hear more upper frequencies – which is perhaps good for the ‘detail’, but mostly unnatural.
Regards,
Thalis
Hugh
Have you ever tried 'coloring' recorded material with low order distortion, the 'tube-like' one? I have to say that from my little experience and to my taste distortion (same as most studio effects) sounds better when applied to instruments or vocals individally. I mean better to add distortion to every track, possibly for various extend, and not to 'mix down'. Same as reverbs, compressions and so on. As little 'mastering' as possible.
Ergo, I don't understand why reproduction should be considered as good place to 'colorate' and as far as I remember Greg is more or less the same opinion.
What do you think?
regards
Have you ever tried 'coloring' recorded material with low order distortion, the 'tube-like' one? I have to say that from my little experience and to my taste distortion (same as most studio effects) sounds better when applied to instruments or vocals individally. I mean better to add distortion to every track, possibly for various extend, and not to 'mix down'. Same as reverbs, compressions and so on. As little 'mastering' as possible.
Ergo, I don't understand why reproduction should be considered as good place to 'colorate' and as far as I remember Greg is more or less the same opinion.
What do you think?
regards
DF,
I think I agree that coloring the recording at playback is not the best way to do it, for the reason that it has the potential to intermodulate with all the other 'mixed down' components of the music - particularly orchestral - to create unmusical 'chords' which will sound discordant.
However, if this is not done at mixing - and yes, it should be done at instrument level which forces an expensive recording process - then I see nothing wrong in doing it at the playback point IFF the added H2/H3 does not exceed around 0.04%.
These are the mixes I use in my designs and it works well. More, and things fall apart.
Thalis,
Thanks for your very positive post. I had not realised that the asymmetric compressibility of air itself is highly distortive, and that a near field mike placement, such as occurs on most electric instruments, militates against H2 pickup. Damn good point, and thanks for bringing it up!!
I have to admit I'm very surprised that no-one has come in (yet!) with a flamethrower!!
Cheers,
Hugh
I think I agree that coloring the recording at playback is not the best way to do it, for the reason that it has the potential to intermodulate with all the other 'mixed down' components of the music - particularly orchestral - to create unmusical 'chords' which will sound discordant.
However, if this is not done at mixing - and yes, it should be done at instrument level which forces an expensive recording process - then I see nothing wrong in doing it at the playback point IFF the added H2/H3 does not exceed around 0.04%.
These are the mixes I use in my designs and it works well. More, and things fall apart.
Thalis,
Thanks for your very positive post. I had not realised that the asymmetric compressibility of air itself is highly distortive, and that a near field mike placement, such as occurs on most electric instruments, militates against H2 pickup. Damn good point, and thanks for bringing it up!!
I have to admit I'm very surprised that no-one has come in (yet!) with a flamethrower!!
Cheers,
Hugh
Hi Crazyhub,
........"The main and uncommon qualities are:
- the SKA seams to have really NO distortion at all, giving a very soft listening: if you hear a harsh sound, it doesn’t come from the amp. The sizes of imaging are “true” it’s to say that they grow up very linearly with spl.
When you listen to the SKA, you think: this amp is the “righter” I have never heard because even if there is some lack of what I pointed out in previous posts, you are absolutely unable to find fault with its sound."...
What more could a designer ask for?
Hi Darkfenriz,
...."Ergo, I don't understand why reproduction should be considered as good place to 'colorate' and as far as I remember Greg is more or less the same opinion."
Agreed.
Having said that, I might add the SKA can be easily converted to gently 'colourise' the sound by simply removing the bootstrap capacitors, for those who wish to experiment - which is what DIY is all about!
Cheers,
Greg
........"The main and uncommon qualities are:
- the SKA seams to have really NO distortion at all, giving a very soft listening: if you hear a harsh sound, it doesn’t come from the amp. The sizes of imaging are “true” it’s to say that they grow up very linearly with spl.
When you listen to the SKA, you think: this amp is the “righter” I have never heard because even if there is some lack of what I pointed out in previous posts, you are absolutely unable to find fault with its sound."...
What more could a designer ask for?
Hi Darkfenriz,
...."Ergo, I don't understand why reproduction should be considered as good place to 'colorate' and as far as I remember Greg is more or less the same opinion."
Agreed.
Having said that, I might add the SKA can be easily converted to gently 'colourise' the sound by simply removing the bootstrap capacitors, for those who wish to experiment - which is what DIY is all about!
Cheers,
Greg
yggdrasil said:
You have to have actually met and spent time them to understand that.
I can say I have as a statement of fact and I suppose that is what makes the distinction between words and facts.
You can't discern too much on a forum.
Its a case of comparing of "how they think" versus "how they want you to think".
I have yet to meet Greg but have spoken at length with him on the phone.
Greg is definately the former and a very nice unassuming fellow.
I hope to complete my GB 150D sometime this week.
I enjoyed this thread a LOT. Mostly because I use myself (like Hubert) the Ellis 1801b.
So far I was not able to find a solid state amplifier that will please my ears as a tube amplifier.
The problem with Ellis speakers is the HUGE power that is needed (83 db/w/m). To make this speakers sing I am looking at 250W!!!
I Dynaco ST70 (30W) - perfect in the midrange and highs (measure going up to 50kHz!). The problem was the lake of power to give bass a slam (or oumph).
Anyhow - keep posting listening impresions about SKA - maybe there is a way to change a bit the sound to make it be more "tremolo".
So far I was not able to find a solid state amplifier that will please my ears as a tube amplifier.
The problem with Ellis speakers is the HUGE power that is needed (83 db/w/m). To make this speakers sing I am looking at 250W!!!
I Dynaco ST70 (30W) - perfect in the midrange and highs (measure going up to 50kHz!). The problem was the lake of power to give bass a slam (or oumph).
Anyhow - keep posting listening impresions about SKA - maybe there is a way to change a bit the sound to make it be more "tremolo".
Hi Bill and Macka,
Thanks for your support. I'm looking forward to your listening impressions. The best way to an objective assessment is through many subjective ones with different ears, acoustics and ancilliary equipment. Only then can we form a perspective.
Hi rmihai,
I have a fair idea of your need and my take is that these amps you talk of on your amp design thread are going to be worked quite hard.
These are inefficient speakers (at 83 dB/W) and you like it LOUD. I think a 300W capable amp (such as the larger version of the SKA) run derated at 250W/ 8 ohm or so would be best. Say a 500VA traffo and +/- 70V (48/0/48) and 0.4 deg C/W heatsinks each side. The Class A for 20W shouldn't be necessary in a well designed Class AB.
Good luck.
Cheers,
Greg
Thanks for your support. I'm looking forward to your listening impressions. The best way to an objective assessment is through many subjective ones with different ears, acoustics and ancilliary equipment. Only then can we form a perspective.
Hi rmihai,
I have a fair idea of your need and my take is that these amps you talk of on your amp design thread are going to be worked quite hard.
These are inefficient speakers (at 83 dB/W) and you like it LOUD. I think a 300W capable amp (such as the larger version of the SKA) run derated at 250W/ 8 ohm or so would be best. Say a 500VA traffo and +/- 70V (48/0/48) and 0.4 deg C/W heatsinks each side. The Class A for 20W shouldn't be necessary in a well designed Class AB.
Good luck.
Cheers,
Greg
Hi Kanwar,
Have you heard of the compound loop version of the SKA using a FET cascode chip front end driving a SKA?
Guess what - I've got them here (30 in fact) and THD 1.3ppm% on sim - way better than Halcro I believe, but SKA sonics.
That's 0.0000013%, can't measure it anyway. It wasn't the goal. Nor was the >10,000 damping factor or the DC - 3MHz response, but the 160 dB PSRR was.
Sorry, comp was very tricky, no schema or detail. Not even the module pics I have handy.
Cheers,
greg
Have you heard of the compound loop version of the SKA using a FET cascode chip front end driving a SKA?
Guess what - I've got them here (30 in fact) and THD 1.3ppm% on sim - way better than Halcro I believe, but SKA sonics.
That's 0.0000013%, can't measure it anyway. It wasn't the goal. Nor was the >10,000 damping factor or the DC - 3MHz response, but the 160 dB PSRR was.
Sorry, comp was very tricky, no schema or detail. Not even the module pics I have handy.
Cheers,
greg
amplifierguru said:
I know we have heard amplifierguru mention PSRR, power supply rejection ratio, a few times now.
With PSRR we can try to say, how sensitive an amplifier is
to disturbance caused by its power supply.
If power supply was totally clean all the time,
we would not need any high supply rejection ratio.
But when an amplifier works, there are all sorts of small signals
as well as larger ripples travelling at power supply rails
and also in the ground rail.
PSRR is an important factor!
Take a look at the elite top class OP-amps.
They may have different values in parameters like noise, slewrate, distortion and other things.
Even if these other variables are also at very good levels,
PSRR, of a top class Operational Amplifier, is however always very high.
This is true, almost without exceptions.
amplifierguru said:
Greg,
Is this the one with the OPA627 front end driving a discrete
inverting back end you spoke of recently?
WRT sim THD results, don't get too excited, there's been plenty
of sim results around here that would never be realised in
real life.
I have found for super linear power stages a good way to unsettle
them is drive a 1 or 2 ohm load with decent power at 10kHz. That's
usually enough to start seeing some harmonics.
A while ago I had an error corrected power OP stage on the bench.
I was doing a quick and dirty linearity check on it by measuring
the AC voltage difference between IP and OP of the OP power
stage.... sort of a unity gain null test.
It had about 2mV rms 'error' with 10V into 1.8 ohms at 10kHz
(55W)! Upon checking the 'error' waveform it was basically a sine
wave.
So even if the error wave distortion was 2% (easily seen on cro)
the THD would be less than 0.0005%, which is pretty good
for that kind of torture test.
Cheers,
Terry
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.