Hi,
The "performance" measured by AP2 is down to the added fuzzy distortion (except it is called "noise shaping" to hide the fact and make it "something good").
Read the datasheets for the parts, most tell you exactly what their modulator core handles.
You can use the formulas in Lipshitz/Vanderkoy paper to calculate how much fuzzy distortion, ooops, noiseshaping, is needed to get a given performance and you can infer easily how many samples need to be averaged (how large the FFT window needs to be) to show this level or accuracy and you can reverse infer the accuracy to which each sample can be reflected in the analogue output, for a non-steady state signal...
Other than that, given the TDA1541 matches 16 Bit Performance closely and it can operate at 8 * OS it is IN PRINCIPLE possible to get 19 Bit equivalence from the TDA1541, in practice this is limited by the -110dB (unweighted) noisefloor. If we then employ noiseshaping and suitable averaging we could get even lower noisefloor and more dynamic range.
I am sure with enough inventiveness it is possible to make a "20 Bit performance" DAC using a TDA1541. But does that make the TDA1541 a "20 Bit DAC"?
Ciao T
The "performance" measured by AP2 is down to the added fuzzy distortion (except it is called "noise shaping" to hide the fact and make it "something good").
Read the datasheets for the parts, most tell you exactly what their modulator core handles.
You can use the formulas in Lipshitz/Vanderkoy paper to calculate how much fuzzy distortion, ooops, noiseshaping, is needed to get a given performance and you can infer easily how many samples need to be averaged (how large the FFT window needs to be) to show this level or accuracy and you can reverse infer the accuracy to which each sample can be reflected in the analogue output, for a non-steady state signal...
Other than that, given the TDA1541 matches 16 Bit Performance closely and it can operate at 8 * OS it is IN PRINCIPLE possible to get 19 Bit equivalence from the TDA1541, in practice this is limited by the -110dB (unweighted) noisefloor. If we then employ noiseshaping and suitable averaging we could get even lower noisefloor and more dynamic range.
I am sure with enough inventiveness it is possible to make a "20 Bit performance" DAC using a TDA1541. But does that make the TDA1541 a "20 Bit DAC"?
Ciao T
This discussion reminds me of a remark Ed Dell made in his interview:
"When the recordings Columbia did in the 70’s came out on CD, it revealed the extent of embarrassing engineer manipulation. They made a lot of changes in the final mix, with multi miking. I remember an AES presentation by a former Columbia engineer admitting that the constant moving of control sliders was making the music pretty much unlistenable. This came evident after we got the resolving power of the CD; we found that our precious music was heavily doctored".
So I believe it is important to make a distinction between the technical capabilities of the medium and the actual content. Even with a technically superior medium you can have awfull sounding music, and vice versa.
Technically, CD and BluRay are superior to vinyl and tape, yet much of the best music around is on vinyl and tape. OTOH, I am convinced that many great sounding BluRay music is simply the result of superior recording and mastering and that it would sound as good on CD.
My 2 eurocents worth.
jan
AD1853 | Stereo, 24 Bit, 192 kHz, Multibit Sigma-Delta DAC | Audio D/A Converters | Digital to Analog Converters | Analog Devices
They claim 24bit resolution. Maybe the modulator is only 9.5-14 bits, but with oversamling and decimation, you still end up with >110db dynamic range.
I think you need to look at this as a system. You cannot pull just the modulator part out and claim its worse than a 1980's device.
They claim 24bit resolution. Maybe the modulator is only 9.5-14 bits, but with oversamling and decimation, you still end up with >110db dynamic range.
I think you need to look at this as a system. You cannot pull just the modulator part out and claim its worse than a 1980's device.
'He' will try to answer to your arguments. (I suppose i have to read "why they don't use an equalizer" ?).
Because it was the same in the mastering rooms, so everything was sounding OK before production. An other reason, that kind of non linearity in monitors affect not only the general response curve, but all the mixing process, like the level of instruments and dynamic treatments the sound engineer bring to his work ( if he use creative multi-track recording techniques). And i insist on the dynamic treble limitation of the magnetic tape at high level witch helps to create warm big sounds and reduce distortion perceptions in a nice, but non transparent way.
About signal/noise, well, and as far as i knows, nobody, till near, had the desire to use Dolbys during digital recording process. Reason why this company was obliged to turn his efforts to movie industry. The hisses of the tape is something witch walk on your side during all analog recording sessions (with some advantages on the coherency of your work). About dynamic, yes, they react in an absolute different way. Digital is dead silent, and if it limit the definition of very low signals, it has more room than you can use, unless you are interested to record silences and reverberation details or crowd ambiances in live recordings, more than the instruments themselves, that are naturally hidden by the hiss of the analog tapes, anyway, so you don"t notice anything wrong on this aspect, that helps...
Of course you have to take care to work very near the clipping level, and never reach-it with digital equipment witch is kind of affraying.
Digital implies to work in an absolute different way. On my opinion, the main limitation is the slope at the high end of the frequency response curve. But, using 96kz....
Last, we were all very used to the analog sound (culture). I remember i was shocked listening to (may-be ?) the first full digital record: "Bop till you Drop" from Ry Cooder. The feeling that something was missing, sounding both dynamic and precise, and thin. Now, with the habit, i find this very good album just nice .
I don't wanted to begin a war digital VS analog. Both have their advantages and inconveniences. And i love both of them for what they offer, as i said.
May-be i'm a stupid person, but, yes, sometimes, if feel an analog copy to sound better, in some ways than the original, like some tube amps, witch brings agreeable pair harmonics. but i'm never shocked by a great difference brought with digital processes on what i hear.
Anyway, industry had decided for us, and analog is nearly dead. Some can be nostalgic, Are-you ?
John, i both love old Ampex mutiltracks and Studers ones. Each ones with their advantages. Studer had a better response curve because they used different optimized heads gaps (entrefers ?) for the recording and reading heads, so you get a different sound during rerecording session (reading via recording head) than during mixing sessions. And, because all your work was done during re-re, i used to feel more coherency with Ampex and often prefer it, despite less treble. Did it reserve some aggressive answers from fans of the one or the other ?
I'm not attacking-you in any way and deserve respect for your work. Can't you consider people can have slight different feelings, tastes and opinions than yours, and still be respectable enough to not use authority arguments against us ?
All i have said is the product of my experience, measurements and reflexions. i believe the same from you ? We just had different lifes, so different experiences, both respectable.
Hi,
Of course they do. And even then they do not even deliver 20 Bit analogue performance... So I guess it is best described as a DAC that accepts 24 Bit Data, but does not offer 24 Bit analogue performance (like almost all DAC's out there).
With oversampling and noise shaping you can do a lot of things, as you can with averaging, but to equate these with the kind of behaviour shown by classic analogue systems and to imply the results are equivalent is, how shall we say, adventurous. If it had not by commercial pressures become the standard approach to digital audio and hence backed by massive lobbies it may have been considered a most extraordinary claim...
Now instead any dissent from such designs is declared extraordinary. The irony nearly kills me.
The modulator IS WORSE. As for dither/noise shaping etc. they remain somewhat contentious, not in the mainstream, to be sure...
It is of course very simple for anyone who cares to make their own experiments and compare classic multibit architecture DAC's to more modern "fuzzy distortion" generators sold as DAC's...
Quite a few have found such experiments rather illuminising and quite possibly learned much more from them than from any textbook.
Ciao T
Of course they do. And even then they do not even deliver 20 Bit analogue performance... So I guess it is best described as a DAC that accepts 24 Bit Data, but does not offer 24 Bit analogue performance (like almost all DAC's out there).
With oversampling and noise shaping you can do a lot of things, as you can with averaging, but to equate these with the kind of behaviour shown by classic analogue systems and to imply the results are equivalent is, how shall we say, adventurous. If it had not by commercial pressures become the standard approach to digital audio and hence backed by massive lobbies it may have been considered a most extraordinary claim...
Now instead any dissent from such designs is declared extraordinary. The irony nearly kills me.
The modulator IS WORSE. As for dither/noise shaping etc. they remain somewhat contentious, not in the mainstream, to be sure...
It is of course very simple for anyone who cares to make their own experiments and compare classic multibit architecture DAC's to more modern "fuzzy distortion" generators sold as DAC's...
Quite a few have found such experiments rather illuminising and quite possibly learned much more from them than from any textbook.
Ciao T
Hi,
Well, given that have worked on commercial products that have either one, I find your comment most curious. The AD1853 is marginally better than the TDA1541 in measured terms, but nothing earth shaking. Yet the AD1853 must use aggressive noiseshaping to attain this small improvement in measured performance.
Sonically the comparison is a no-comparison. Even the OEM Customer who's DAC this is (and who insisted on using this AD Chip, as he had a large stock of them) agreed it was much worse, subjectively, than a TDA1541 based one. But he was happy enough with the designs results anyway, as it reportedly did to a Mark Levinson top of the range DAC what is normally described as "blowing into the weeds".
The AD1853 has one thing going for it though, it is cheap.
Ciao T
Well, given that have worked on commercial products that have either one, I find your comment most curious. The AD1853 is marginally better than the TDA1541 in measured terms, but nothing earth shaking. Yet the AD1853 must use aggressive noiseshaping to attain this small improvement in measured performance.
Sonically the comparison is a no-comparison. Even the OEM Customer who's DAC this is (and who insisted on using this AD Chip, as he had a large stock of them) agreed it was much worse, subjectively, than a TDA1541 based one. But he was happy enough with the designs results anyway, as it reportedly did to a Mark Levinson top of the range DAC what is normally described as "blowing into the weeds".
The AD1853 has one thing going for it though, it is cheap.
Ciao T
When you work in a recording studio, you can compare hundred times a day, on all the instruments, and real time if any difference before and after quantization. With good equipments, nobody is able to point any difference in a coherent way. With analog tapes, you here instantly the difference and nobody makes an error in the before after tape position of your hidden switch. Difference is obvious. I made this demonstration several time with a lot of people arguing against Digital. They remained voiceless, no more argues.
checked: Oh yes, I remember those.
I also remember a funny review of the 1st Bill Haley and the Comets on CD. The reviewer was not sure that the better resolution was kind to the music. It just wasn't made to listen to on clean equipment.
Oh, yes. That is all what i try to point out: the difference between good sound and transparent process. Hard days nights for the "Wall of sound" of Berry Gordy, in a modern digital studio ;-)checked: Oh yes, I remember those.
I also remember a funny review of the 1st Bill Haley and the Comets on CD. The reviewer was not sure that the better resolution was kind to the music. It just wasn't made to listen to on clean equipment.
Sy,
Give you two more.
First, I goofed in the Sim, need to redo it, so I am now unsure on the whole subject. Which is why I hate to post Sim results, especially hasty ones.
Secondly, I found one of my reference to earlier notes of noise in caps, especially showing higher noise in Silver Mica's:
PM noise generated by noisy components (presented at Frequency Control Symposium, 1998. Proceedings of the 1998 IEEE International)
Ascarrunz, H.D. - SpectraDynamics, Lafayette, CO;
Aimin Zhang - National Institute of Meteorology, Beijing, PRC;
Ferre-Pikal, E.S. - NIST, Boulder CO;
Walls, F.L. - NIST, Boulder CO;
Ciao T
Give you two more.
First, I goofed in the Sim, need to redo it, so I am now unsure on the whole subject. Which is why I hate to post Sim results, especially hasty ones.
Secondly, I found one of my reference to earlier notes of noise in caps, especially showing higher noise in Silver Mica's:
PM noise generated by noisy components (presented at Frequency Control Symposium, 1998. Proceedings of the 1998 IEEE International)
Ascarrunz, H.D. - SpectraDynamics, Lafayette, CO;
Aimin Zhang - National Institute of Meteorology, Beijing, PRC;
Ferre-Pikal, E.S. - NIST, Boulder CO;
Walls, F.L. - NIST, Boulder CO;
Ciao T
Hey Jan,
There are obviously many reasons for why a CD may sound bad. But they make many modern Hybrid SACDs with both CD and SACD content, with classical music. So it's easy to do a comparison. And in the comparisons I've made, the SACD format sounds better.
Now, an objectivist may not hear a difference. Probably just say, sounds the same to me. And that's fine. High-end audio is a subjective sport ...
The Sony PCM 1650 digital recorder used for almost all of the first CD's used a NINE bit A/D converter followed by a Seven bit A/D on the residual. That followed Stockham's design.
The first audio A/Ds (CA 1968) were built using carbon comp resistors. It was found the Allen Bradley hot molded versions worked the best. That limited the resolution. As folks learned more and wanted better resolution they went to mil spec resistors.
Scott,
Try baking the mica capacitor at a low temperature for a few days to see if the losses are due to moisture. For fun you could try it on the poly cap. But my crystal ball says it will be worse even after a short bake, if you can scrape the remnants up to test it.
The first audio A/Ds (CA 1968) were built using carbon comp resistors. It was found the Allen Bradley hot molded versions worked the best. That limited the resolution. As folks learned more and wanted better resolution they went to mil spec resistors.
Scott,
Try baking the mica capacitor at a low temperature for a few days to see if the losses are due to moisture. For fun you could try it on the poly cap. But my crystal ball says it will be worse even after a short bake, if you can scrape the remnants up to test it.
Right.
... If i remember well, the first machine which had changed my mind about digital used a delta sigma (1bit). But even at this period, most of the annoyances where dues to the analog filterings and output stages of the AD/DACs.
I remember my 12 bits cd player from philips with modified analog stages, not so bad for the times.
The first digital mutitracks where very unstables with temp, a nightmare, can't even remember their names...
ouch, don't we are far away from the topic ?
Last edited:
Hi,
Actually, most poly cap's bake fine. The Mica's may have drawn moisture due to long storage...
Baking tubes, carbon resistors and other stuff can also have interesting results...
Ciao T
Try baking the mica capacitor at a low temperature for a few days to see if the losses are due to moisture. For fun you could try it on the poly cap. But my crystal ball says it will be worse even after a short bake, if you can scrape the remnants up to test it.
Actually, most poly cap's bake fine. The Mica's may have drawn moisture due to long storage...
Baking tubes, carbon resistors and other stuff can also have interesting results...
Ciao T
- Status
- Not open for further replies.