agreed,
I guess a stronger distinction needs to be made between measuring the output of these DAC chips, with minimal I/V or data sheet app circuit filtering vs a quality Audio DAC circuit
My definition of a quality Audio DAC circuit implementation would use these chips and the required higher order low pass filtering to give good measurements even into the older AP or other analyzer with HF IM problems - because a quality Audio DAC circuit won't have the HF hash on the properly filtered output
Know the DAC chip’s output spectrum, design the output filter to reduce all of the out of band switching glitches, images, shaped noise, clock feedthru…
I would aim for 40-50 kHz filter corner to accommodate 96k source, give some margin
Audio CD limits while not obviously limiting with music signal are too close to the edge for a conservative engineer looking at uncertain results on audibility when the added margin is so cheap today
good practice now that DSP is so cheap is to upsample 44.1 and do the sharp transition band anti-imaging filter in digital, so the same high quality higher fc analog filters can be used for all formats
and there are some tough rquirements for op amps in the filters, I/V to avoiding gnerating IM in the op amp diff pair - and techniques which can help as I mentioned in my previous few posts
Last edited:
PMA, your SACD 10KHz square wave is very similar to my 10KHz square wave made with a 30 ips analog tape recorder with phase compensation. I think this is the high frequency response where we have the potential for true fidelity.
Last edited:
I also "believe" in wide frequency response, without sudden sharp cut-off at 20 kHz. To me, SACD sounds very natural, and I prefer it. Unfortunately, the number of new SACD editions decreases fast. Hopefully I mostly listen to classical music, and many SACD classical titles have already been issued.
I should mention that sharp DIGITAL filter with cut-off frequency below 22.05 kHz is an INEVITABLE part of CD 44.1kHz playback, even if there is an oversampling! Oversampling only allows for lower order ANALOG filter. The digital filter is inevitable for the reason to suppress aliases in audio band.
Last edited:
Hmmmm, 22.6 uS between samples for a std CD spec, that implies that the fastest signal it should be giving is 22.6uS from full -ve to full +ve, if it is faster than that then there must be phase or frequency errors...
Wrinkle
Hi,
I can take some when I get to clip a 'scope back onto the output.
Perhaps I need to explain...
The DAC operates without oversampling and the squarewave is created with a software called "Dactester" and hence is technically speaking "illegal" (does not follow sample theorem).
There is no filter (digital or analog) for this measurement, if I apply the "Sinc rolloff correction" the resuts look a lot more than what Pavel posted for SACD, but that is just BTB.
Ciao T
I can take some when I get to clip a 'scope back onto the output.
Perhaps I need to explain...
The DAC operates without oversampling and the squarewave is created with a software called "Dactester" and hence is technically speaking "illegal" (does not follow sample theorem).
There is no filter (digital or analog) for this measurement, if I apply the "Sinc rolloff correction" the resuts look a lot more than what Pavel posted for SACD, but that is just BTB.
Ciao T
Wow, you can quote Voltaire, but Otto was crowned Emperor by the Pope! The theory as to Roman was that it followed in the empire traditions even though not the territory of Rome. As Otto did combine a few kingdoms it technically also made empire. So once the advertising guys got around to it, "The Holy Roman Empire" was born.
Much more accurate than many claims made in modern audio!
Sorry I saw someone else already answered your question. If you know John V. and his little buddy Stanley L. I would expect you had run across Bob Adams of DBX and ADI, he has written many articles in the JAES. No malicious intent.
Thanks, Scott. I only recently joined the AES, so I've only just gotten started purchasing articles from their library. I attended the local AES chapter for years before joining, and they referenced Vanderkooy many times, thus the bulk of my personal library includes his contributions.
It was also popular in the local AES chapter to test the "new" SACD players as they came out, using very high-end test equipment, in an attempt to measure how close the actual audio came to the marketing claims. Thus, I've been exposed to tests which rather humble SACD versus its claims. My apologies if SACD has been testing very well in circles around diyAudio, but my exposure has been the opposite - that 96 kHz and 192 kHz PCM test better.
By the way, these same engineers showed a very nice, real-time demonstration of the effects of jitter in a digital audio stream. They had a setup with very low jitter where they could introduce jitter on demand, and then the attached real-time frequency spectrum display would show the damage to the audio as jitter increased. I've also seen the effects of jitter on the noise floor in my own designs, but that's another topic.
I will look out for Bob Adams in the JAES library. I'm sure I will enjoy the reading.
Sorry if I've come in late and missed the equivalent 10 kHz tests from 192 kHz PCM, but my first inclination is to compare SACD v. PCM192 v. 30 ips tape, all with the same test. Again, I'm not asking for brand new tests, but it would have been nice to see the graphs side-by-side.
SACD's 2.8 Mhz bit rate is closer to 24/96 on Shannon-Hartley information rate
you could save some on the bit depth in PCM with noise shaped dither and 16 bits which works especially well with high sample rates as a final consumer delivery format - you definitely want 24 bits in the studio
Sony currently recommends a 50 kHz analog low pass filter - for the noise shaping used it really needs to be 6th order - this is the SACD step response limit
96k PCM can come close at ~40 kHz
analog tape has a AC bias of up to a few 100 kHz which also has to be removed with a filter - but tape head gap and tape speed are limits too - there is simply no cmparison possible if you want modern ADC levels of S/N, distortion and frequency response flatness - tape is objectively worse in all of these
you could save some on the bit depth in PCM with noise shaped dither and 16 bits which works especially well with high sample rates as a final consumer delivery format - you definitely want 24 bits in the studio
Sony currently recommends a 50 kHz analog low pass filter - for the noise shaping used it really needs to be 6th order - this is the SACD step response limit
96k PCM can come close at ~40 kHz
analog tape has a AC bias of up to a few 100 kHz which also has to be removed with a filter - but tape head gap and tape speed are limits too - there is simply no cmparison possible if you want modern ADC levels of S/N, distortion and frequency response flatness - tape is objectively worse in all of these
Considering the noise levels in the best listening environments, you could easily get by with 18 or 19 bits, with 20-bit being the closest standard. Not quite as convenient as 16-bit or 24-bit, but it's at least smaller.
When Sony recommends 50 kHz low pass, are they talking about the -3 dB point of the filter, or does the system need to stop 50 kHz? (I've always been baffled by the challenge of a Nyquist filter, because it seems like you need to completely stop all frequencies above Nyquist and yet most filters barely reach -100 dB at Nyquist, so there is always a little aliasing).
Seems like one could work with 192 kHz PCM, using a gentler low pass slope, and approach the step response of tape without any of the shortcomings of tape.
"analog tape has a AC bias of up to a few 100 kHz which also has to be removed with a filter"
Always something new.
Always something new.
- Status
- Not open for further replies.