But......
Traditional singing/music........and finally, what comes out of a loudspeaker's cone, traveling through the air to your ears..... is still Analog.
THAT, cannot and will never change.
See above.
I don't consider a ~80dB max SNR signal as clean nor do I believe that a digital step with 100dB+ end-to-end SNR adds anything worth worrying about.
Sorry, this part I don't understand.
Bad wording on my part. Wouldn't consider myself low income. Low disposable income is more like it. Nowadays my preferred luxuries are a kid, a house and a car. Quite unfortunate that having kids has become a luxury, but here we are.
My 2017 budget system was a PC and a set of Adam Audio F5 speakers. Room EQ'd and time aligned with drc-fir. Quite high rez for the money. I was born in 86 and I like the time we live in for all the possibilities we have, tech wise.
Not that simple with a sigma-delta DAC. SNR is not a constant. FFT measurements are generally PSS (periodic stead state). DACs also produce non-PSS noise (and or various distortion artifacts that may look like noise on an FFT). Anyway, SNR changes with the instantaneous DC level of the analog outputs.
Below is some of what ESS said in a marketing presentation where they talking about problems with dacs where they thought they had advantages over their competitors. They didn't talk about other types of problems they know about but don't see any advantage in talking about. There are other things too though, and hiding the real noise floor when no audio signal is present is one of the well known tricks in the sigma-delta dac business. There is considerably more that we could talk about too, say, idle tones, noise skirts, substrate coupled noise, hump distortion, etc. Point is that sigma-delta modulators used in dacs do weird stuff that either doesn't happen in audio amplifiers or else tends to rare in well-designed units.
Anyway, here are some quotes from ESS for those things they were willing to admit exist (a couple of points about noise and SNR have been bolded):
From the ESS Hyperstream Modulator presentation:
There is a slide which initiates discussion about audiophiles with the words, "Understanding what audiophiles are hearing."
"The surprising reality is that sigma-delta DACs can be audibly distinguished from a conventional DAC despite measuring very much better than that DAC."
"...an important point: The human ear detects signals well below the noise level of the DAC."
"The ear is exquisitely sensitive to "unusual" noise sources. Your ancestors camped out by a waterfall (white noise) and yet their 'ears pricked up" when they heard a hint of a predator moving in the undergrowth. (The equivalent visual phenomenon is "seeing something out of the corner of your eye). Noise, to a large degree, can be accommodated by the ear and is not troubling, but the tiniest "anomalous" noise is raised to the conscious level."
"Sigma-delta modulators create non-periodic steady state noise (non-PSS) artifacts..."
"Periodic Steady State analysis is common in RF circuits. It means that the system is forced to repeat a pattern of behavior over and over again with a certain time period. Any artifact is presumed to also repeat in this time period."
"Audio measurements such as THD and DNR are done in the Periodic Steady State. Therefore, they will not activate non-PSS noise. You will not find non-PSS noise by looking at THD, DNR, and SNR."
"As the audio signal moves, the noise does not remain the same."
"Non-PSS noise is the biggest issue, but experiments suggest there are more problems. For example: Audiophiles rate as inferior systems that have variable excess phase noise."
"We find that an unconditionally stable loop sounds better in listening tests."
Last edited:
True, and let's just hope it stays that way. There are several AI companies who will keep trying to fool us. How long will it be before Mr. Roboto is singing to us......OH, wait a minute, how about a tone deaf human and a big bunch of auto tune (digital by the way). We are there already. I have learned that my attempts at singing can't be fixed with any autotune algorithm that I can afford.
The US Supreme Court just ruled that AI produced art cannot be granted a copyright. I imagine that it's just a small step before that's applied to music. At least they got this one right.
Digital, or Analog, both technologies impart imperfections. Both have chased them down to nearly infinitesimal levels when properly implemented. Everyone is entitled to their own personal preferences, but much of what we hear today goes through a bit of both except for direct human to human interaction.
For those that continuously chase the state of the art there are now some 32 bit floating point audio interfaces that sample at up to 192 KHz. Some claim so much dynamic range that there isn't even a level control. Does any of the popular DAW recording software even support this yet?
Zoom quotes "The technology built into the UAC series allows the user to capture the highest audio quality possible, with an ultra-wide dynamic range and an incredibly low noise floor. 32-bit Float recording, along with sample rates up to 192 kHz, processes your audio at an increased bit-depth and at full bandwidth. All of this is done without the need to set gain." And "32-bit Float recording gives musicians peace-of-mind, ensuring that their audio will never clip. No matter how loud or soft the signal, the audio capture is crisp and clean.
It's the other way around: DAWs have been using 32 bit floating point precision for quite some time for their internal processing. Only now are consumer grade ADCs starting to catch up to where the entire pipeline can be 32 bit float from start to finish.
Probably all of them. 32-bit float is a storage format, but real data converters aren't good for even 24-bits so 24-bit fixed saves some disk space. Besides IIRC 32-bit float is good for 25 significant bits since you get a free bit in the mantissa with floating format. The other 8-bits are the exponent, which is not meaningful to a fixed point dac (which they all are).
Also, some DAWs switched to 64-bit processing years ago.
Might also mention converting 24/192 hi-res PCM to DSD256 IME sounds better with some dacs.
Last edited:
That marketing material is over 12 years old. Don't know about ESS but their competitors have made some advances since. How long do you plan to quote it?
Most DAWs have indeed used 32 and 64 bits for their internal signal processing and generation. The question is whether they have I/O paths already coded, or at least hooks for this in place to be handled by the audio interface drivers. Maybe I only see 16 and 24 bit choices because there's no 32 bit hardware hanging on the other end of my USB cable, or drivers installed for 32 bit hardware. I'll stick with the shiny red 24 bit boxes that I already have. They are already far better than my hearing.
Nothing attached to a typical PC and powered from the same noisy UPS will ever make the 110 dB dynamic range quoted by some interface makers. I have even seen PC motherboards claiming 108 or more from their own on board audio. Translation......the codec chip we use can do 108 dB on a good day going downhill with a strong tailwind, so that's our spec.
For the last 12 years of my career at Motorola I worked in a research group that morphed into an IC design group. One of the chips in our custom chipsets contained three sigma delta DAC's for driving speakers. For the next generation chipset, we needed some DACs for setting DC bias levels in the transmitter control loop, so one of the brainiacs just copied the "speaker DAC" despite me explaining that it won't work. Standard Sigma Delta DACs work best on a constantly moving signal like audio, otherwise they "hunt" generating noise and DC offsets. Some newer audio interfaces have "LF response to DC" as an output option so that they can be used to drive the control voltage input on modular synthesizer equipment, so this is possible, but not without consideration in designing the DAC.
Regarding sigma-delta dacs that can be used down to DC, ESS PRO series dacs measured at various DC levels are shown to produce very low noise in the steady-state as be seen from a graphic in their presentation from 'over 12 years' ago:
The graphs are made by sending the dacs a digital signal that represents a fixed DC level, then measuring noise, probably with an FFT. The DC level is stepped and another noise measurement taken.
Also, it is common for modern dac outputs to be DC coupled. They do seem remain quite stable and quiet so long as more or less random HF/RF modulator noise is sufficiently filtered out (along with some low level clock noise). However, dynamic noise variations have also been observed following some transients.
In any case, I have not heard a dac that sounds as 'real' as an unusually high performance vinyl playback system. Maybe not surprising that there are still new patents being made that continue to improve sigma-delta audio dacs. Some of the improvements may not show up very well in typical FFT measurements, but some of them do help SQ. IIUC, Bruno Putzeys used a PWM Sigma Delta modulator in his Mola Mola dac. The architecture is still under patent protection so it is not widely used, at least not that I know of. An FPGA IP version looks like this:
https://patents.google.com/patent/US6972704/un
As can be seen from the patent link, the idea is well over 20 years old. Must be obsolete then? Some people might think so.
FPGA IP: https://www.teledynedalsa.com/en/pr...gn/design-products/pwm-sigma-delta-modulator/
The graphs are made by sending the dacs a digital signal that represents a fixed DC level, then measuring noise, probably with an FFT. The DC level is stepped and another noise measurement taken.
Also, it is common for modern dac outputs to be DC coupled. They do seem remain quite stable and quiet so long as more or less random HF/RF modulator noise is sufficiently filtered out (along with some low level clock noise). However, dynamic noise variations have also been observed following some transients.
In any case, I have not heard a dac that sounds as 'real' as an unusually high performance vinyl playback system. Maybe not surprising that there are still new patents being made that continue to improve sigma-delta audio dacs. Some of the improvements may not show up very well in typical FFT measurements, but some of them do help SQ. IIUC, Bruno Putzeys used a PWM Sigma Delta modulator in his Mola Mola dac. The architecture is still under patent protection so it is not widely used, at least not that I know of. An FPGA IP version looks like this:
https://patents.google.com/patent/US6972704/un
As can be seen from the patent link, the idea is well over 20 years old. Must be obsolete then? Some people might think so.
FPGA IP: https://www.teledynedalsa.com/en/pr...gn/design-products/pwm-sigma-delta-modulator/
Last edited:
No, but over 12 years old unsubstantiated marketing claims are still nothing but unsubstantiated marketing claims.
I'd expect them to publish the odd AES paper if there is actually some new insight in all of this beyond some unproven technobabble. Why wouldn't they if it helps them sell more products?
Claims to the effect that factual information can be disregarded or dismissed as 'technobable' is respectful?
But no, I was not trying to insult you. I do seriously question how much you actually understand of what ESS said. It was not technobable to me. It was science understood by the PhDs at ESS.
EDIT: If you have a question about the meaning of some specific technical part of it I will try to explain it and or try find some supporting literature.
But no, I was not trying to insult you. I do seriously question how much you actually understand of what ESS said. It was not technobable to me. It was science understood by the PhDs at ESS.
EDIT: If you have a question about the meaning of some specific technical part of it I will try to explain it and or try find some supporting literature.
Last edited:
Still sounds like marketing info. Unnamed "competitors" could be anything. And a phd is not evidence of quality science either, and without context a meaningless attempt to make something sound smarter. Especially if it's a marketing phd....
Keyword: Processes.
And like anything 'processed', it's not naturally the way it started out.
There's processed meat, cheese, potato chips, etc etc....
And processed lovemaking thanks to Viagra. 😳
Agreed, on that point. However, Martin Mallinson is a respected scientist with multiple patents, and an old friend of Scott Wurcer. IIRC Scott helped them by doing some dac measurements, and they sent him some of their measurements for his comments. Scott once posted a distortion residual image of non-PSS distortion in an ESS dac. IMHO they are serious guys.
- Status
- Not open for further replies.