The silliness of the SACD vs. CD argument of course is confusing the messenger with the message. The CD is an information carrier, as is the SACD, and DSD. And that information is totally sufficient in each of those cases to provide as detailed and "correct" a picture of the encoded event as anyone could want. Where the faltering occurs is that the mechanism that does the real time looking up of the information is not too brilliantly implemented much of the time, and "roughens" up the message at the time of delivery.
I've done enough experiments to be thoroughly convinced that I could take the highest resolution, "audiophile" recorded digital files, DSD128 or PCM 352/24, drop them down to boring ol' CD information density level, and in fact go yet another step, turn them into high quality MP3 files. And then reverse the operation, translate from that MP3 right back up to the original high resolution format. Now have the original, and a "severely corrupted", multiple generational degraded version thereof - have people play the two on the "best", most discriminating setup, AB style, and would happily put money on "golden ears" not being able to differentiate them ...
A couple of days ago I downloaded a demo track from a new, "audiophile" website doing double speed DSD downloads -- ultimate resolution versions from master tapes. And what were the highest frequencies that had any meaning in that track? About 23KHz - the striking of a metallic percussion instrument. I separated out all content above 20kHz, and replayed only that material at maximum volume, with my ear pressed hard against the driver. And what did I hear? The hissing and slight gurgling of the electronics, precisely the same as when I pressed the stop button ...
Frank
I've done enough experiments to be thoroughly convinced that I could take the highest resolution, "audiophile" recorded digital files, DSD128 or PCM 352/24, drop them down to boring ol' CD information density level, and in fact go yet another step, turn them into high quality MP3 files. And then reverse the operation, translate from that MP3 right back up to the original high resolution format. Now have the original, and a "severely corrupted", multiple generational degraded version thereof - have people play the two on the "best", most discriminating setup, AB style, and would happily put money on "golden ears" not being able to differentiate them ...
A couple of days ago I downloaded a demo track from a new, "audiophile" website doing double speed DSD downloads -- ultimate resolution versions from master tapes. And what were the highest frequencies that had any meaning in that track? About 23KHz - the striking of a metallic percussion instrument. I separated out all content above 20kHz, and replayed only that material at maximum volume, with my ear pressed hard against the driver. And what did I hear? The hissing and slight gurgling of the electronics, precisely the same as when I pressed the stop button ...
Frank
JC, got some measurements to 'prove' this MYTH?
You seem fixated on 'revealed knowledge' as expounded by Yourself and scorn measurements of any sort .. both those you have promoted in the past ... as well as other peoples.
If you can't measure distortion, Sam will be happy to oblige to test Blowtorch for you.
There are many factors that are vital to low distortion and good sound ... such as good decoupling and earthing practice that are MORE important than using Golden Pinnae parts hand carved from solid Unobtainium by virgins.
Theoretical calculations mean nothng if basic principles are ignored.
_________________
Hornman, I can confirm that many (all?) Golden Pinnae amps, (including some designed by JC which he now disowns) have worse distortion into real speakers.
Often they will show signs of oscillation on parts of the waveform and this is dependent on the thermal and signal history too.
Last edited:
Can you explain what we are seeing? Test conditions? Load? What are you measuring? What are the 2 curves?
Regulation was not under discussion, but:
TI's is around 55 db at 3 kHz, ST's is under 45, A good opamp such as the AD797 adds another 95 dB. Now ST lists 90 uV/V broadband noise.
But when you use a regulator IC you often have ripple much higher than 5%. (3 volts is typical before a regulator.) Add Fletcher Munson weighting to the energy from a real rectifier and what do you think the final S/N ratio is? (try it around 3K)
Scott, 3V -55 dB -95 dB -40 dB (harmonic level at 3K re 120 Hz as measured on a DC supply) + ? dB (preamplifier gain) + ?dB (amplifier gain) / 2.83 V (1W 8ohms) = -? dBw. That's my math before you ask. So what is the loudspeakers sensitivity. Headphones would show it, if used as a microphone or phono preamp under I think most all conditions! (90 dB/W loudspeakers at 10 cm. also.)
Oh don't forget the noise goes into all the preamp stages but adds mostly to the input stage S/N.
Mainly silicon, i believe
<snip>
Topology, if I'm not mistaken, is similar to Vendetta's first stage.
Best,
Attachments
Last edited:
Are you talking noise or ripple here? Or none of the above?
Where are you pulling this 5% number from? I'm looking in the 7815 datasheet and it says ripple rejection better than -60dB.
I think he's talking about raw supply (before the regulator) noise. By bringing 3kHz in, he sorta forgets the weighting of the terms of that order in the Fourier series representing 120Hz ripple...
I used the same concept for my protection circuit and as a dynamic distortion magnifier to look and listen to them.
I think this idea could help to design an OPA, with a reduced open loop gain and a high open loop bandwidth, in order to reduce further the distortion.
But i don't know if it is possible to fine tune enough resistances values on the wafer.
There is several known tracks for error corrections, i wonder why they are underused.
http://www.tubecad.com/2010/04/04/Tringlotron.png
http://www.diyaudio.com/forums/head...ased-tringlotron-amplifier-2.html#post2154368
John asked how it sound.
My thoughts:
First, you can only remove around 20db of distortion this way (but it is not so bad). you are limited by stability.
Second, the good one, the OPA i used (if you use some) does not add its signature.
Third, as i tested it on a yet very good amp (while mid-fi, i am not sure of an obvious listening change, while measurements where better. It would be interesting to test this on a amp with more previous THD than mine.
Just AFEC it!
Yo ! I was happy to know that we can make recordings (and save them for the future) in a overkilling format.
From there to claim that CDs are just garbage seems (a little ?) excessive.
I have wonderful CDs witch sound absolutely wonderful. And i can even enjoy MP3s in my phone, listened across my Koss Porta pro (i love deep basses
May-be i'm more involved by the music than the ultimate perfection of my (midle-fi) system ?
This said, if i understand well, and in order to be hype, i have to pretend that there is no life below 24/96, with the exception of vinyls which are the audio nirvana despite the surface noise, the clics, the distortion and this incredible distance with the master copy i own on tape ?
Fas42,
I follow you on the down sampling of a high res file down to the red book cd level, that would seem plausible, but to then turn that into an MP3 and say that you can get back to where you where I don't see how that is possible. The MP3 format is a lossy format, how do you not lose information if you have losses in the basic format? Something has to give to in effect zip the music down to MP3 or am I missing something here? Now if you said you used a flak file or Wave file with no loss I can accept what you are saying, but MP3?
I follow you on the down sampling of a high res file down to the red book cd level, that would seem plausible, but to then turn that into an MP3 and say that you can get back to where you where I don't see how that is possible. The MP3 format is a lossy format, how do you not lose information if you have losses in the basic format? Something has to give to in effect zip the music down to MP3 or am I missing something here? Now if you said you used a flak file or Wave file with no loss I can accept what you are saying, but MP3?
Ummmm... so how did you manage to get 5% (0.75V) 120Hz output ripple out of a 7815? That's really something that TI and ST would be interested in.
Yes, lossy but very carefully lossy, that's the key thing! If you only hear MP3 done with sloppy encoding, at a low bit rate, and replayed on barely tolerable gear of course it will sound miserable ... but it doesn't have to be that way. If you process using a high quality software tool, like LAME, at the highest allowable rate, with all the right settings then it will be essentially indistinguishable from the original. Yes, people can train themselves to hear the differences but casual, even "golden ear" hearing will struggle to pick it up. Where the MP3 does fail, and I can hear this, is that the CPU is working flat out doing the decoding in real time for playback, and this "pollutes" the playback via the usual interference suspects. So, if you do the decoding off-line, by resampling right back up to high resolution status for later replay you eliminate this problem area.
Christophe's MP3 sample is a good test here: playing that back as a DSD file the quality of treble, and unscrambling of the musical elements is far superior.
EDIT: An analogy is digital TV: this is all encoded using lossy compression, and for 99.9% of the time, on a good broadcast, no-one will notice anything. But drop the quality of that encoding, and then the picture will degrade in a highly objectionable way ...
Frank
Last edited:
I don't have a clue as to what you are talking about here. Maybe some day we will have a Simon Codex.
Not sure that this strong a statement is appropriate. The data sheet discloses the circuit, another spin on the JE990 inductor trick, in any case there is hardly a fair comparison based on cost.
Fair enough. I was speaking figuratively, as in that a different order of processing was taking place, the electronic interference pattern, dance, within the circuits differs -- I haven't actually watched the CPU meter doing this sort of replay.
Though, it is interesting to watch: on our desktop, playing something like a 176 WAV from disk used essentially nill CPU; yet, a DSD was something like 20%. The level of usage is all over the place, depending on everything, as usual - you always have to compare oranges with oranges ...
Frank
I don't know what would satisfy anyone here. We have many 'retirees' and 'amateurs' doing our 'inventory' when it comes to measurement and distortion estimates.
Samuel Groner has done us all a great service in making such extensive measurements of most common IC op amps and even a few discrete designs.
Do any of you out there do anything useful in this area? Do you have the test equipment to at least go down to 1 part in a million? If not then maybe you are exceeding your qualifications in criticizing us here.
Now what about my 'estimate' of the Blowtorch's harmonic distortion?
First of all: ANY open loop system that is also Class A has a predictable distortion curve, ONCE you get an accurate set of measurements at some typical level. You can go both up or down in level and PREDICT the distortion with a great degree of accuracy.
Come on everybody, look it up in a textbook on non-linear circuit analysis.
Let me give an example of how I would (and do) estimate distortion in an open loop, Class A system:
First, I get a set of exact measurements of each harmonic at some part of the operating level within the design parameters of the design, i.e. NO pre-clipping levels, as this will throw off the estimate.
Then: For 2'nd harmonic, estimate that its value drops directly with voltage level.
For 3'rd harmonic, estimate that its value drops directly with the SQUARE of the voltage level.
For 4'th harmonic, estimate that it drops to the CUBE of the voltage level.
Etc, etc.
Now what about 7th harmonic? It should drop as the 6th power, which implies that if you drop the voltage level in 1/2, then the 7th harmonic should drop by a factor of 64.
so successive iterations should give reasonably accurate results, sometimes even better than some Spice programs, which appear to 'dither' at very low levels, or else I could show you a practical example from one of my textbooks on the subject.
If you can show me where I am in error, I would appreciate it.
Samuel Groner has done us all a great service in making such extensive measurements of most common IC op amps and even a few discrete designs.
Do any of you out there do anything useful in this area? Do you have the test equipment to at least go down to 1 part in a million? If not then maybe you are exceeding your qualifications in criticizing us here.
Now what about my 'estimate' of the Blowtorch's harmonic distortion?
First of all: ANY open loop system that is also Class A has a predictable distortion curve, ONCE you get an accurate set of measurements at some typical level. You can go both up or down in level and PREDICT the distortion with a great degree of accuracy.
Come on everybody, look it up in a textbook on non-linear circuit analysis.
Let me give an example of how I would (and do) estimate distortion in an open loop, Class A system:
First, I get a set of exact measurements of each harmonic at some part of the operating level within the design parameters of the design, i.e. NO pre-clipping levels, as this will throw off the estimate.
Then: For 2'nd harmonic, estimate that its value drops directly with voltage level.
For 3'rd harmonic, estimate that its value drops directly with the SQUARE of the voltage level.
For 4'th harmonic, estimate that it drops to the CUBE of the voltage level.
Etc, etc.
Now what about 7th harmonic? It should drop as the 6th power, which implies that if you drop the voltage level in 1/2, then the 7th harmonic should drop by a factor of 64.
so successive iterations should give reasonably accurate results, sometimes even better than some Spice programs, which appear to 'dither' at very low levels, or else I could show you a practical example from one of my textbooks on the subject.
If you can show me where I am in error, I would appreciate it.
Last edited:
Since you presented no data, just assertions and "creative" comparisons, you're unlikely to have anyone be able to point out an error.
- Status
- Not open for further replies.