You mean, problems you *think* exist with the digital filter because someone else told you so. Other people, including me, think they are full of it.
It's amusing that people think that ESS is competent enough to design the best audio D/A converter, including a high-order sigma delta modulator, and yet not competent enough to get a basic interpolation filter right.
Even if it were a problem, it can be easily bypassed and replaced.
I have a PhD in electronic engineering to guide me. I have known him for 45 years and he used to design D-A converters at Zeltex, an early commercial company. He is going for a patent that might improve things. Unfortunately, his input to me is still confidential, until the patent is made. Apparently, Benchmark already knows about this problem.
Last edited:
Patent on an interpolation filter, I'm sure that's going to be a good one. Should stop wasting the USPTO's time.
JC, what is it that you hear is wrong with the interpolation, and what is the nature of the improvement (if you have heard it) ?.
Dan.
Dan.
Maybe he does. We'll see if it's ever granted and worth the paper it's printed on.
It's a little difficult to imagine that there can be anything truly novel in this area, considering that it is a staple in data acquisition systems and image / video processing. People inside the little audio bubble sometimes like to think they've invented things that have existed and been used elsewhere for a very long time.
Nevermind the fact that it is trivial to get close to an ideal filter with any realization in an external DSP or FPGA.
Last edited:
What's an ideal filter, how close is the stock filtering, and how close is this "realization in external DSP" ?.
Dan.
Reconstruction filter - Wikipedia
http://www.ti.com/lit/ds/symlink/pcm1792a.pdf
Page 8 for a typical audio implementation.
I happen to think what the PCM1792 implements, for example, is more than good enough. Could it be improved? Sure, there could be even less passband ripple, and it could be even sharper so that it's in the stopband by Fs/2.
Based on the datasheet group delay number, it looks like this is done in 111 taps. You can do thousands to millions of taps in real-time depending on how much processing power you are willing to throw at it and how much delay you are willing to tolerate.
Last edited:
Most here are not qualified to tell me about audio design. They just lack the audio experience, or the pertinent education in audio related subjects to teach me much. I do trust my ears to tell me when audio is 'right' or 'compromised significantly'.
Looking at your Constellation circuit John, I can see why there’s a world wide shortage of JFETs.
Get with the program man - use some BF862’s and go single ended: quieter, more elegant, more compact
More later
😉
Get with the program man - use some BF862’s and go single ended: quieter, more elegant, more compact
More later
😉
Don't Misunderstand, Thelma Houston, 1975, Sheffield Lab. 🙂
I confess I do a lot of listening with digital; they stop being offensive to me from 24/96 on, and let me listen though them to the music. And they are so convenient.
But do I enjoy the sound? Not really; for that, I need the good old vinyl, whenever I want to test some new gear or try out some mods to old gear, or for the shear pleasure alone. I know, LP's are not practical, and, moreover, they don't last, so each spin brings them closer to their death. I can't really describe their sound, my literary capabilities are too limited in any of the languages I know, but it's kind of like when I was standing in a slot canyon for the 1st time, "wow" is all I could say.
And this audio nirvana doesn't have to cost an arm and a leg: a humble Denon DL-103 will get you there for $200.
You can rightly point out your commercial success and experience with analog electronics.
On the flip side, we could be skeptical that you have the requisite experience with digital to blame some fault with ES9038 on its digital filter specifically.
I have some fantastic sounding LP’s - I seem to get more listening pleasure out of them than CD’s and as someone remarked a while back, I am attracted to expense and inconvenience of the format.
0.4nV/rt Hz is hardly earth shatteringly low though. Certainly adequate for all but the most pathological cartridges though.
- Status
- Not open for further replies.