and
by Post#6 (in the link):
and by Post#9 (in the link):
T5 ... how about you do your part reading what the OP suggests? I merely quoted/answered to his last&only post tonight. You tone to me is inappropriate, making hard to communicate with you. What kind of precision you demand from anyone if yourself do not read and try to understand?
seriously, why 199-200KHz? Not better 397-398KHz - and why if not?
Or, can we use the 97-98KHz to measure our more common amplifiers designed for 100KHz BW?
I am out of this path anyway. To start with, I do not have 2x precision sources at 200KHz with less than 120dB harmonics. Then, my audio designs involve LP filters everywhere (input, output) and removing them (for enabling such measurement) will make the designs immediately out of specs by default.
For curiosity sake, I would like to see practical examples for sources and also for amplifiers, then an explanation why all this effort is relevant for audio even if only at simulations levels.
For my curiosity, again, I wonder why a serious discussion on my proposal 10*THD +N (instead of THD+N) would not be more relevant in this topic.
Or, can we use the 97-98KHz to measure our more common amplifiers designed for 100KHz BW?
I am out of this path anyway. To start with, I do not have 2x precision sources at 200KHz with less than 120dB harmonics. Then, my audio designs involve LP filters everywhere (input, output) and removing them (for enabling such measurement) will make the designs immediately out of specs by default.
For curiosity sake, I would like to see practical examples for sources and also for amplifiers, then an explanation why all this effort is relevant for audio even if only at simulations levels.
For my curiosity, again, I wonder why a serious discussion on my proposal 10*THD +N (instead of THD+N) would not be more relevant in this topic.
Sorry if my tone seemed inappropriate. To be honest I joined this thread half way through and just skimmed the first few pages to get a grip on the original topic.
I will let all the stuff regarding IMD measurements at 199/200kHz be, it's not practical, most of us don't have neither sources clean enough at these frequencies (DACs mostly out of question, these have reconstruction LPFs ways lower in frequency). Same goes for audio ADCs. Not to mention that the whole chain would have to run at 768kHz to fulfill nyquist requirements. And after all these measurements would only show decreasing open loop gain of opamps (usually dropping at 20dB per decade), so it would be easier to measure IMD at 19/20kHz and infer that performance at 199/200kHz should be 20dB worse. To be honest: I'm still not getting why it would be important for your point you're trying to make.
Where does the -70dB noise floor come from? We should verify that in the first place, practically noise floor can be much lower in modern electronics. If you design the gain structure of your signal chain correctly it can be way below -100dB, even <-110dB! Then, from your theory we would need in excess of -130dB THD. Still I'm a little bit trying to guess what is actually your point here...
At least I will provide my opinion: If top-performing opamps are used and they are implemented in a way they can achieve, or at least near datasheet performance, why not use them? One less worry to have in your chain when opamps act as "wires with gain", or at least as close to it as possible.
You started out by asking about discrete opamps:
I tried to make clear that harmonic distortion and IMD is mathematically related (might call it "the same mechanism"). Think I made that clear enough, so I will tone down now.
Let's see where your theory and proposal for a "new metric" will go... All the best!
Thanks, I'll play the ball but I am not an expert.
My point is, -100dBu noise floor or perceived harmonics or IMDs (10*THD+N) should be enough for typical ears.
If this can be done also with zero-phase and micro-precision performance (such that all harmonics still recompose the signal correctly) then it is my definition of analytically perfect.
Is this 'perfect' an equivalent for 'good' or 'best' sounding: I do not know. Probably not.
From personal experience:
I am sure that I do not have the sufficient knowledge or I have not searched enough for solid states which resemble the feelings of valves.
Still, my visits to HighEnd München gave me so far no new results: solid states remain solid states, and valves remain valves.
The mystery is still there, for me.
Back to my point:
- if the abyssal THD+N on solid states did not solve the mystery, then why not make a step back (to normality): 100dB 10*THD+N is my proposal.
Else, please give me examples of solid states which measures AND sounds exactly like good valves.
P.S. in my office and in the lab I still have small old valve radios from Loewe!
- 70dB was just an example in this context. If one insists, the number came to mind inspired from typical magnetic tape performances: SNR 70dBrms-A, H3-3% (innocent for IMDs) but H2+IMDs should less than the tape noise floor.
- myself, I fail to see the point for 199/200KHZ IMD test as well. Precisely for the reasons you explained.
- when possible, I like to shoot for -100dBu noise floor and -120dBu any harmonics
- if I understood correctly learning from friends/musicians: for a pleasant clean tone (in the sense of a valve preamp or a guitar amplifier) the number of main harmonics is slightly more important for tonality than their order. That said, H2+3 or H2+4 are better sounding than H2+3+4, and all these are definitely better than H2+3+4+5+6+7+...
- the solid state answer for this are the offers with abyssal THD+N
My point is, -100dBu noise floor or perceived harmonics or IMDs (10*THD+N) should be enough for typical ears.
If this can be done also with zero-phase and micro-precision performance (such that all harmonics still recompose the signal correctly) then it is my definition of analytically perfect.
Is this 'perfect' an equivalent for 'good' or 'best' sounding: I do not know. Probably not.
From personal experience:
- the 40 years of memories of music + room or venues are ALL either live or tape or valve based. They are "young, alive and multidimensional".
- in contrast, the memories of music from reproductions with solid-state amps : neither one is remembered with the room or venue, all of them are 'dead', all are 'old' immediately after finish of the song. They simply don't penetrate or stick to my hypothalamus.
I am sure that I do not have the sufficient knowledge or I have not searched enough for solid states which resemble the feelings of valves.
Still, my visits to HighEnd München gave me so far no new results: solid states remain solid states, and valves remain valves.
The mystery is still there, for me.
Back to my point:
- if the abyssal THD+N on solid states did not solve the mystery, then why not make a step back (to normality): 100dB 10*THD+N is my proposal.
Else, please give me examples of solid states which measures AND sounds exactly like good valves.
P.S. in my office and in the lab I still have small old valve radios from Loewe!
This is going a bit off-topic but I think it is worth pointing out to answer questions which appear to be embedded in your post.
In case you are not aware tape saturation was (is) a real (and often deliberately exploited) effect. Here is a quick little blurb (from an AI enhanced search). What I am trying to point out is that the sonic characteristics of tape saturation (and valves) were (are) sufficiently desirable that there are digital plugins which seek to emulate the effect:
[This is also related to why musicians often love certain gear such as the vacuum tubes in guitar amplifiers and some microphones/preamps.]
Sure I know this detail a bit.
But there are not existing yet any emulators for the 'modulation noise'. That's also important for the tape-sound (not only the H3/K3 and smooth type of saturation).
IIRC, there are yet no scientific level emulators for music-on-tape.
So, no true tape emulators yet.
Besides, by principle of their inner working the emulators make the phase a mess (day to night difference in comparison with tape).
Same as the solid state vs valves: there are similarities. But one is like the pseudo-stereo/stereo enhancer/stereo wide effects, and the later is the true stereo. Maybe they are kind of similar on the short term (during listening), but not the same thing on the long term (memory).
I appreciate the thread title as really inspired and important, but the definition of 'good sounding' still not clear and precise enough.
But there are not existing yet any emulators for the 'modulation noise'. That's also important for the tape-sound (not only the H3/K3 and smooth type of saturation).
IIRC, there are yet no scientific level emulators for music-on-tape.
So, no true tape emulators yet.
Besides, by principle of their inner working the emulators make the phase a mess (day to night difference in comparison with tape).
Same as the solid state vs valves: there are similarities. But one is like the pseudo-stereo/stereo enhancer/stereo wide effects, and the later is the true stereo. Maybe they are kind of similar on the short term (during listening), but not the same thing on the long term (memory).
I appreciate the thread title as really inspired and important, but the definition of 'good sounding' still not clear and precise enough.
(Briefly) I view the problem broken into two pieces:
1. During creation of music various devices (tape, vacuum tubes, transformers, etc.) are carefully and deliberately used to craft certain desired harmonics, distortion, etc.
2. During recording and playback the chain (op-amps, DAC, ADC, DSP, etc.) are carefully used to record and then reproduce with the least modifications or additions (harmonics, distortion, noise, etc.)
So in this context I expect (in a manner of speaking) that the (perfect) op-amps don't have a sound (as much as possible) and simply serve their functions as precisely as possible. Ideally shouldn't the DAC/ADC (and involved op-amps, etc.) eventually have no describable impact on the sound?
1. During creation of music various devices (tape, vacuum tubes, transformers, etc.) are carefully and deliberately used to craft certain desired harmonics, distortion, etc.
2. During recording and playback the chain (op-amps, DAC, ADC, DSP, etc.) are carefully used to record and then reproduce with the least modifications or additions (harmonics, distortion, noise, etc.)
So in this context I expect (in a manner of speaking) that the (perfect) op-amps don't have a sound (as much as possible) and simply serve their functions as precisely as possible. Ideally shouldn't the DAC/ADC (and involved op-amps, etc.) eventually have no describable impact on the sound?
The answer was included here, intended to be read:
Your invitation is insidious. After "an amp", maybe comes "the amp", then the loudspeakers, then the room, then volume, then the music, then my ears, then my taste and finally me being different, then reissue : repeat everything due to a slight misunderstanding. Then another misunderstanding. Then switch to another "the amp". Etc. I am not born yesterday.
If Mr. Pass wants to display at HighEndMünchen this year, I will surely visit his booth.
Others have knowledge. I can only attend the place where they make their case in the best way possible.
Your invitation is insidious. After "an amp", maybe comes "the amp", then the loudspeakers, then the room, then volume, then the music, then my ears, then my taste and finally me being different, then reissue : repeat everything due to a slight misunderstanding. Then another misunderstanding. Then switch to another "the amp". Etc. I am not born yesterday.
If Mr. Pass wants to display at HighEndMünchen this year, I will surely visit his booth.
In most cases by most listeners (include me) remains this impression after listening test (e. g. in hifi shops) of various solid state amps vs. various tube amps.Thanks, I'll play the ball but I am not an expert.
So far, these are just clarifications as I see them (but I could be wrong).
- 70dB was just an example in this context. If one insists, the number came to mind inspired from typical magnetic tape performances: SNR 70dBrms-A, H3-3% (innocent for IMDs) but H2+IMDs should less than the tape noise floor.
- myself, I fail to see the point for 199/200KHZ IMD test as well. Precisely for the reasons you explained.
- when possible, I like to shoot for -100dBu noise floor and -120dBu any harmonics
- if I understood correctly learning from friends/musicians: for a pleasant clean tone (in the sense of a valve preamp or a guitar amplifier) the number of main harmonics is slightly more important for tonality than their order. That said, H2+3 or H2+4 are better sounding than H2+3+4, and all these are definitely better than H2+3+4+5+6+7+...
- the solid state answer for this are the offers with abyssal THD+N
My point is, -100dBu noise floor or perceived harmonics or IMDs (10*THD+N) should be enough for typical ears.
If this can be done also with zero-phase and micro-precision performance (such that all harmonics still recompose the signal correctly) then it is my definition of analytically perfect.
Is this 'perfect' an equivalent for 'good' or 'best' sounding: I do not know. Probably not.
From personal experience:
- the 40 years of memories of music + room or venues are ALL either live or tape or valve based. They are "young, alive and multidimensional".
- in contrast, the memories of music from reproductions with solid-state amps : neither one is remembered with the room or venue, all of them are 'dead', all are 'old' immediately after finish of the song. They simply don't penetrate or stick to my hypothalamus.
I am sure that I do not have the sufficient knowledge or I have not searched enough for solid states which resemble the feelings of valves.
Still, my visits to HighEnd München gave me so far no new results: solid states remain solid states, and valves remain valves.
The mystery is still there, for me.
Back to my point:
- if the abyssal THD+N on solid states did not solve the mystery, then why not make a step back (to normality): 100dB 10*THD+N is my proposal.
Else, please give me examples of solid states which measures AND sounds exactly like good valves.
P.S. in my office and in the lab I still have small old valve radios from Loewe!
But actually in most cases apples are compared to oranges or formulated more exactly amps with small idle power in the output stage (Class AB - only solid state amps) vs. hard running amps (Class A - both tube and solid state amps).
As I know, there are no tube amps with idle power values comparable to those of typical Class AB solid state amps.
If one only choose solid state Class-A amplifiers (of which there aren't many on the marked) when comparing valve vs. transistor,
one wouldn't get the impression "solid states remain solid states, and valves remain valves"
Examples of solid states which measures AND sounds at least similar like good valves are the mentioned models in attachment No 2, 3, 4 and 10 from post #2 in this thread. The associated amp reviews are to find in "Stereophile".
P.S.: for new cost effective diy projects I recommend Buscher Audio for class A amp units with sonic character like good sounded tube amps (from Germany):
André Buscher, Paulinenweg 3a, 51149 Köln.
It would be interesting for me to know whether there are also integrated operational amplifiers in the TO220 outline (like outline of e.g. TDA2030) that differ from the known versions in that they run with a higher quiescent current in all stages - e. g. in a topology like AD817 or AD825
- input stage 10-13mA,
- cascode stage of folded cascode 30-40mA and
- output stage 200mA.
Last edited:
about me, I do have a bit of decent design&build experience
- up to 2x500W hybrid (10xLM3886/ch), balanced, bridged, parallel, global feedback as voltage or current output variants. I could not obtain more than 130KHz BW on it with 3mH loads (subwoofers). However, it was a stable performing 15Kg block (before PS, case and accessories).
- a 2x30W hybrid circlotron - class A, MOSFET based. While this one was much better performing, reaching 400-600KHz BW on 3mH loads, it was long-term stable only up to 20W. Reaching 30W meant killing many MOSFETs or power supplies.
- studying a 2x80W circlotron I realized that power supply requirements&issues are very serious. I did not build it, as I was afraid of killing the set of very special power supplies (>1000€/each) which I bought specially for this.
Keeping the 2x20W circlotron in constant current source configuration with a decent Rout >2.5KOhm (admittedly closest to valves, also by Pass among others), I modified & tried the few speakers I have (87-97dB efficiency range): but no combination was sounding 100% as good as a valve-based system. It means, a very good sound overall, but no crystal clear memories of audio + room (which any good valve will easily do) were detected. A bit of them, but not as pregnant and clear as with valves.
Same strategy I have for HighEnd Audio show in München: I walk in all performing rooms, noting just the room number (not/never the name or looking for the type of hardware, be that a 5Kg or 50Kg unit!). Then, if in the next show day (or better in 2 days) some memories will be detected as 'audio+rooms', I'll go back in there and ask for details. So far, these notes sent me back to rooms where hardware were valve based.
The other very good sounds, indeed, were the rooms with powerfull&heavy pure class A amps. But I won't buy those units (or permanently build them as DIY projects), as for me the important function would be to give memories music+room while at around 95-100dB SPL. Maybe I'll finalize a couple of the 'lightweight category' 2x20W MOSFET circlotrons in CCS configuration with case, dedicated PCBs and everything, as it really sounded excellent on good classical drivers. Maybe the rest of the money and attention I'll invest in better&exotic drivers and room acoustic serious upgrades.
Additionally I have another test, actually better and harder than the 'memories+room' one, but which cannot be experienced at HEA: rock vocal+guitar at loud volume (beyond 110dB in free space). Admittedly there are always very few units at HEA which give me the impression they can give a perfect natural sound for this test. You clearly know it when you meet it: the strong feeling of "wow, this rock voice+guitar thing is natural". Much like the crystal stereo image, which many don't know what it is until they experience it. How hard is to make it, how easy is to destroy it...
Maybe I should quit this solid state challenge and start doing circuits with "proper&simple" components, the valves, at least for preamp stages like PS Audio and others do.
Last edited:
It would be interesting for me to know whether there are also integrated operational amplifiers in the TO220 outline (like outline of e.g. TDA2030) that differ from the known versions in that they run with a higher quiescent current in all stages - e. g. in a topology like AD817 or AD825
do you mean this:?
composite amp 1875
finished products:
kaltecs versions
i do not have this amps build or heard
- input stage 10-13mA,
- cascode stage of folded cascode 30-40mA and
- output stage 200mA.
do you mean this:?
composite amp 1875
finished products:
kaltecs versions
i do not have this amps build or heard
@tiefbassuebertr
(i.e. Roender's FC100) in a TO220 case?
did you mean this https://www.diyaudio.com/community/...ge-audio-power-amplifier.111756/#post-1349109
(i.e. Roender's FC100) in a TO220 case?
Upgrading a bit my language:
Nevertheless, my experience is not as great as others. I'm basically here still to learn from the better ones.
Of course, my new quality-metric proposal (10*TD+N)rms >100dBu in 20KHz still holds.
I changed THD with TD, to include the IMDs, and of course we need the rms version.
- hybrid amp : I meant a composite amp
- global feedback : I meant error correction.
Nevertheless, my experience is not as great as others. I'm basically here still to learn from the better ones.
Of course, my new quality-metric proposal (10*TD+N)rms >100dBu in 20KHz still holds.
I changed THD with TD, to include the IMDs, and of course we need the rms version.