Distortion due to intersample overshoots, for example.
That is just digital clipping, isn’t it.
Agree on a difference in graceful handling of signal faults like the above and dropouts etc.
Like the diskman that was made to handle them📀
Agree on a difference in graceful handling of signal faults like the above and dropouts etc.
Like the diskman that was made to handle them📀
ASR has their dac expert. He and I have had some discussion. He is not sure about some things that might be audible with some dacs. At least he is willing to say to say so. He's still thinking about it as far as I know.
There are also other people over there who are labeled as technical experts, some pf whom aren't really expert on dacs. Doesn't keep them from thinking they know more than they do.
BTW, a new study came out related to overconfidence. Attached for your reading pleasure.
There are also other people over there who are labeled as technical experts, some pf whom aren't really expert on dacs. Doesn't keep them from thinking they know more than they do.
BTW, a new study came out related to overconfidence. Attached for your reading pleasure.
Feynman's complete talk itself is fascinating. Here's a transcript:
https://calteches.library.caltech.edu/51/2/CargoCult.htm
Another "audiophile" belief I've seen on ASR, and also promoted by manufacturers, is one I discovered when wondering if audio recording interfaces could be used to record ultrasonic sound (in a biological/mouse experiment done by an associate), as all current ones have sample rates of 96k and 192k as well as CD standard 44.1k. Audio interface manufactures don't spec the frequency response above 20kHz regardless of sample rate, and posters on ASR don't care about anything above 20kHz, but they sure want more and more sample points to "more accurately reproduce that 20kHz sine wave" that the higher sample rates give them. They just don't believe the Nyquist theorem.
The reliable solution to recording ultrasonics is to design the electronics from scratch or use industrial equipment such as National Instruments.
It's digital (or analogue, but usually digital) clipping due to an input signal that isn't clipped. A bad but common mastering practice is to normalize the largest samples to full scale, but the interpolation (or reconstruction) filter output may then have to exceed full scale in between the samples. If there is no headroom for that, the DAC clips.
That does reduce intersample overshoots. (Of course a bit of digital attenuation before any filtering takes place is more effective.)
There is an ongoing discussion on test procedure for intersample overs at ASR. https://www.audiosciencereview.com/...le-test-procedure-for-dacs.49050/post-1756967
Who knows it may some day end up as part of the measurement suite at ASR.
Yes inter-sample overs are interesting but they're not going to produce the huge variation in DAC sounds that some people talk about. It's more of a niche thing.
In my experience, two exactly the same DAC circuits (AD1865N-K NOS DAC) do not sound the same depending on the passive elements used in the construction. Mainly the electrolytic capacitors, resistors and wiring were changed from sample to sample, while the active elements were identical. I wouldn't know about two different DAC devices measuring well. I don't have the proper equipment to measure THD and such.
One niche thing that can be measured. Do you think it's responsible for people thinking a Chord Dave sounds better than a Topping D90?
But when a DAC is measured it's measured at the analogue outputs so the signal has already been through all those caps etc. The measured distortion is well below audible range so why would changing caps to more expensive ones help? It makes no sense.
I didn't use coupling capacitors on those DACs, only DC coupling. All those (small) differences are from electrolytic capacitors in the power supply. Some were made with Nichicon FW, Elna Silmic 2 and similar, and some with Samwha WL, Panasonic FC/FR and similar. There is quite a big difference in price.
Some things really don't make sense, but they are heard. I don't bother with it much.
Some things really don't make sense, but they are heard. I don't bother with it much.
I know a lot of people don't like Amir at ASR but he's done a YouTube video where he shows a really dirty AC mains signal and feeds it into a Topping DAC and the DC signal that comes out it's power supply is ruler flat, and the distortion measurement is the same with the normal AC power signal or the deliberately distorted one. Changing caps like that does not affect sound. What it tells you really is that the placebo effect is strong and anyone can be fooled by it so be very careful.
Perhaps other aspects are more important in human hearing, such as lateral direction hearing being time based in microseconds?
No. That show's that the psu filters in the Topping DAC are well engineered. If he measured a stock Topping DAC, then took out the psu and decoupling and coupling caps, measured their parasitics and then changed them for other caps with the same capacitance but different parasitics and then measured the DAC again and the measurement results were exactly the same, then you could assume that the capacitors do not affect sound.
Since this thread is about well measuring dacs which property of dacs influence lateral direction hearing and which well measuring dacs have issues with that property?
I have not yet seen a test that focuses on time differences between the channels, perhaps ypu know of one?
What would cause this presumed time difference between the channels?
Besides a time difference in microseconds would mean a difference of 1 or more samples at 768kHz. Would be very easy to see in measurements.
Besides a time difference in microseconds would mean a difference of 1 or more samples at 768kHz. Would be very easy to see in measurements.
Last edited:
- Status
- Not open for further replies.