Thank you very much chris.
There seems to exist a concise version of that document.
Only 9 pages, to get the high level of it:
http://www.ti.com/lit/ug/slyu032/slyu032.pdf
There seems to exist a concise version of that document.
Only 9 pages, to get the high level of it:
http://www.ti.com/lit/ug/slyu032/slyu032.pdf
I am curious to ask something the 2 persons who said the AD1862 is the top (miro1360 and diyiggy)
Have you head the AD1955?
And If yes, how is it compared to the AD1862?
Have you head the AD1955?
And If yes, how is it compared to the AD1862?
AD1955 is delta-sigma DAC on my to do list. I have not tried it yet, but I expect a good quality.
Wrong threads, or focusing on the wrong opinions? Something was obviously wrong.
There are many people here who believe this. I don’t buy it either, but I don’t see what makes their subjective opinions any more or less wrong than any others.
It seems to me that we all hear rather differently. For example @Markw4 said he heard distortion on my (multibit) PhiDAC - something I don't hear, except on particular types of music which I doubt are Mark's choice for listening.
As you might expect, I have a rather dim view of subjective impressions. I guess you could say I find them all equally unreliable 🙂. Your point on hearing differently is interesting. It’s not something people bring up around here very often. If that includes cognitive influences then I’d have to agree. I’m not sure about biologically or neurologically, but won’t pretend to be an expert there. It would make for a more peaceful forum in any case. I like this thought.
Yeah subjective impressions are unreliable - but I don't find them to be so unreliable they're to be automatically dismissed. Perhaps we hear differently because of our 'training histories' - if we're bugged by something in what we hear and figure out what it is, its generally not possible to 'unhear' it, i.e. learn not to notice it again. Is that what you mean by 'cognitive influences' ?
Perhaps, I was thinking of preconceived notions. For example, many say silver cable sounds bright. It’s possible people can’t focus on anything but treble once they know it’s silver.
It’s possible that people could be “defect hunting” as you describe. I’d buy that for gross defects and distortion. Not sure about the arguments concerning stuff 100+ dB down etc.
I think subjective impressions are a moving target, too. For example, when I first got my HD800s I had been using AKG K701 for a while. I almost sent them back for the sibilance - but after a few days you adjust. Now I find the HD650 and K701 a bit dull.
Last edited:
I recall a talk by Bob Carver describing how he couldn't 'shake off' the sound of silver wire once he knew it was silver. Can't find any link to that though.
One man's 'gross' presumably is another man's 'inconsequential'. I listen differently now that I've heard (fairly subtle) differences in output coupling 'lytics. The difference is in how the lower frequency ambience seems to cling to the speakers - the better caps (typically Panasonic ones but I've not experimented much) don't leave a 'hole in the middle' of the soundstage i.e. the LF is subjectively more free of the transducers.
Yep, concur. I used to find my AKG240s a bit too dull at the top (back when the DAC I was working on had just 1VRMS output), preferring Superluxes. But now when give them a bit more drive (2V) they sound just fine and the Superluxes are a tad harsh.
Late last year a hobby DAC I made was used at a meeting of Dutch audio DIY enthousiasts. It is a massively oversampled single-bit sigma-delta DAC with to today's standards rather limited dynamic range, and it was used with its steep phase-linear digital oversampling filter. They seemed to like it a lot, and I even got compliments from someone who normally only listens to non-oversampled multibit DACs because he doesn't like digital filters or sigma-delta modulators.
Was this due to the circuit design (little passband ripple, headroom for intersample overshoots, properly dithered quasi-multibit sigma-delta modulation) or because it looked impressive in its large wooden box with ten valves, an FPGA board and a reconstruction filter with large potcores? I guess I'll never know.
Was this due to the circuit design (little passband ripple, headroom for intersample overshoots, properly dithered quasi-multibit sigma-delta modulation) or because it looked impressive in its large wooden box with ten valves, an FPGA board and a reconstruction filter with large potcores? I guess I'll never know.
Last edited:
In actual fact, there has been a deplorable decline in quality from the beginning in favor of manufacturing simplicity, covered by promotional ploy. For instance, DACs should have a large chip area.
Totally wrong. Do you know how many transistors are in some old crap DAC from 1990 with a huge die? I heard that we may have made progress in CMOS process technology since then. I have a cat toy with a CPU fabricated on a 28nm process.
What kind of crazy theories drive your false assertion that die size has anything to do with sound quality?
You should call Analog Devices and TI and let them know all of their high end SAR ADCs and DAC counterparts would be better off in the early 90s.
Last edited:
I can't go into details, but I can assure you that unless you need very high speeds, designing analogue circuits is far easier in old BiCMOS processes than in 28 nm CMOS. For one thing, the voltage handling is much better in old processes. Digital, on the other hand, just gets faster and cheaper with smaller feature sizes.
I have the impression that most of the high-performance CMOS sigma-delta audio DACs are made in fairly old-fashioned CMOS processes. The digital core supply is often 1.8 V rather than the 0.9 V to 1 V you would expect for 28 nm and the analogue supply is often as high as 5 V. I guess the manufacturers look for a compromise between analogue signal handling and digital density and end up using fairly old-fashioned processes.
I have the impression that most of the high-performance CMOS sigma-delta audio DACs are made in fairly old-fashioned CMOS processes. The digital core supply is often 1.8 V rather than the 0.9 V to 1 V you would expect for 28 nm and the analogue supply is often as high as 5 V. I guess the manufacturers look for a compromise between analogue signal handling and digital density and end up using fairly old-fashioned processes.
If you look at a number of the high voltage foundry processes, they're hanging around the 350nm node, so yeah, the ultra tiny transistors are not needed or beneficial for analog/mixed mode necessarily.
Doesn't mean the processes are ancient though, just less need to push the lithography.
Doesn't mean the processes are ancient though, just less need to push the lithography.
Sure, I didn't say they were using 28nm CMOS for audio converters, or that it was easier and necessarily advantageous. The numbers do show that the modern converters outperform anything from the 90s, despite being lower power and smaller.
- Status
- Not open for further replies.