Bear,
AFAIK the questions you are asking have little/nothing to do with our implementation. We haven't done anything unusual.
I would be as eager as you to know the answers Dustin/ESS would provide. You are simply barking up the wrong tree if you expect me to be able to answer them.
I didn't mind your questions, and I don't discourage them. Just please direct them to the right folks.
Your might want to ping this thread as Dustin monitors it regularly:
http://www.diyaudio.com/forums/showthread.php?threadid=117238&goto=newpost
I was going to offer to lend you a Buffalo and an ESS eval board to test. But your lack of respect has killed that idea.
Cheers!
Russ
AFAIK the questions you are asking have little/nothing to do with our implementation. We haven't done anything unusual.
I would be as eager as you to know the answers Dustin/ESS would provide. You are simply barking up the wrong tree if you expect me to be able to answer them.
I didn't mind your questions, and I don't discourage them. Just please direct them to the right folks.
Your might want to ping this thread as Dustin monitors it regularly:
http://www.diyaudio.com/forums/showthread.php?threadid=117238&goto=newpost
I was going to offer to lend you a Buffalo and an ESS eval board to test. But your lack of respect has killed that idea.
Cheers!
Russ
Russ,
If there is a lack of respect, I really don't quite see it. There are numerous posts on specific implementation details, and someone had previously posted the squarewave response. I was merely reading along... I waited quite a while and then asked what I think is a reasonable question. I asked if anyone (especially those who posted 'scope shots already) could do a measurement.
You said " you have only seen the output from one Sabre" - which implies explicitly that the output from other Sabres are different, and that you have this information! Yes? And, mikelm followed that up by asking directly.
I never suggested that you know the "answer" to my question. I did say that you OR Brian might, or someone else might have the info. And btw, someone else said that I said that the overshoot was "a serious problem". I did not say anything of the sort.
In fact it may not ever be audible, and may not ever occur in normal operation. Also, even if it does or did, it might be less significant in terms of audibility than artifacts of other sorts that come from other DACs using other methods.
I also said that anything that does digital in the 44.1kHz. world is going to have to have compromises. That is ANYTHING. The only question is which compromises are acceptable (and why) and what degree of which compromise is better or worse.
I also said I am interested in the chipset and its potential.
I also said that I had heard a friends and although I couldn't make any determination because the system wasn't quite as good as I would have preferred.
Then you told me to "2)...Borrow, build, or buy one for yourself" if I wanted to know... next you tell me that you were formerly going to send me an eval board, etc... but now ur not??
So, why are you getting mad at me??
What did I do?
Let's not fight - I'm not interested in that. I certainly have no bone to pick with you.
_-_-bear
If there is a lack of respect, I really don't quite see it. There are numerous posts on specific implementation details, and someone had previously posted the squarewave response. I was merely reading along... I waited quite a while and then asked what I think is a reasonable question. I asked if anyone (especially those who posted 'scope shots already) could do a measurement.
You said " you have only seen the output from one Sabre" - which implies explicitly that the output from other Sabres are different, and that you have this information! Yes? And, mikelm followed that up by asking directly.
I never suggested that you know the "answer" to my question. I did say that you OR Brian might, or someone else might have the info. And btw, someone else said that I said that the overshoot was "a serious problem". I did not say anything of the sort.
In fact it may not ever be audible, and may not ever occur in normal operation. Also, even if it does or did, it might be less significant in terms of audibility than artifacts of other sorts that come from other DACs using other methods.
I also said that anything that does digital in the 44.1kHz. world is going to have to have compromises. That is ANYTHING. The only question is which compromises are acceptable (and why) and what degree of which compromise is better or worse.
I also said I am interested in the chipset and its potential.
I also said that I had heard a friends and although I couldn't make any determination because the system wasn't quite as good as I would have preferred.
Then you told me to "2)...Borrow, build, or buy one for yourself" if I wanted to know... next you tell me that you were formerly going to send me an eval board, etc... but now ur not??
So, why are you getting mad at me??
What did I do?
Let's not fight - I'm not interested in that. I certainly have no bone to pick with you.
_-_-bear
Russ White said:Hi Leo,
Any chance you could check it now that you have the IVY tuned up?
Just curious as I have not looked it it on a decent scope yet.
Cheers!
Russ
Hi Russ,
Sorry the delay, I've been a little snowed under at work
Anyway heres a latest shot of a 1k Square wave using the latest filter settings for the IVY, it looks the same but sounds better than the old filter
0db square wave, Time/DIV on scope is set to 50uS
Cheers,
Leo
Bear,
My beef with you is your reaction (knee jerk or not) to my suggestion that you obtain a Sabre based DAC for you to do your own testing to answer your own questions. I did not even suggest it had to be my implementation.
Also when I said "only one Sabre" I meant only the ES9008 and not the ES9018.
A little respect goes a long way.
Cheers!
Russ
My beef with you is your reaction (knee jerk or not) to my suggestion that you obtain a Sabre based DAC for you to do your own testing to answer your own questions. I did not even suggest it had to be my implementation.
Also when I said "only one Sabre" I meant only the ES9008 and not the ES9018.
A little respect goes a long way.
Cheers!
Russ
leo said:
Hi Russ,
Sorry the delay, I've been a little snowed under at work
Anyway heres a latest shot of a 1k Square wave using the latest filter settings for the IVY, it looks the same but sounds better than the old filter
0db square wave, Time/DIV on scope is set to 50uS
Cheers,
Leo
Very nice. Thanks Leo! I really appreciate it.
Cheers!
Russ
Russ White said:
I'm sorry, what???
Your post on that was your post, I didn't even care who measured what on what implementation.
As far as the questions being "mine" geez man, they are general questions that pertain to any implementation of ANY DAC made by anyone.
Your "beef" seems to now be shifting as to cause.
In case you don't or didn't percieve the implied or meta message in what you said publicly to me in your post it reads like this:
"2)...buzz off jerk - you are rocking my boat and raining on my parade"
I'd say that the lack of respect Russ is coming from your side, not mine. Seems to me like you have a chip on your shoulder about something or other that I have no clue about, other than someone saying something, anything no matter how minor about something you are selling here.
_-_-bear
whohhhh.....this topic is quickly spiraling out of control. stop wasting this (up till now) great topic..... take some relaxation tea or go smoke something.......
I have seen to many great designers/diy-ers leave forums because of endless discussions like this. I would hate it if this happens with Russ of Brain.
Cant wait until my buffalo32s is delivered and I hope we can discuss improvement/tweaks on these pages in the future!
Scanspeakman
I have seen to many great designers/diy-ers leave forums because of endless discussions like this. I would hate it if this happens with Russ of Brain.
Cant wait until my buffalo32s is delivered and I hope we can discuss improvement/tweaks on these pages in the future!
Scanspeakman
suggestion for Variac
To enhance the informational content of this thread, I suggest the following:
1) move the recent 1k scope plots to thread 117238 "ESS Sabre Reference DAC"
2) There, those who made the images can supply additional information for comparison, such as square wave source, square wave sampling frequency as seen by the DAC, DAC input and output circuits
3) delete this post plus all the fruitless "you said - I said" posts
4) We drop this, move forward and have a nice weekend.
Thanks
To enhance the informational content of this thread, I suggest the following:
1) move the recent 1k scope plots to thread 117238 "ESS Sabre Reference DAC"
2) There, those who made the images can supply additional information for comparison, such as square wave source, square wave sampling frequency as seen by the DAC, DAC input and output circuits
3) delete this post plus all the fruitless "you said - I said" posts
4) We drop this, move forward and have a nice weekend.
Thanks
There is a Buffalo tweaking thread here that it might sit more comfortably in: http://www.diyaudio.com/forums/showthread.php?s=&postid=1816317#post1816317
Digital Reproduction of Squarewaves
Wow, it really may be time to end this fruitless squarewave discussion. Let me put a few things into perspective:
(a) A square wave has very slowly decaying harmonics up to infinite frequency. It is by no means bandlimited.
(b) The signal you want to convert to digital MUST be rigourosly bandlimited to less than half the sampling frequency. If you don't do that, aliasing distortion results.
(c) Aliasing distortion cannot be removed later on.
(d) If you use a soundcard, you are responsible for properly bandlimiting the signal you input; the soundcard is not doing that for you. Garbage in, garbage out. If you want to burn a test CD for testing squarewave response, you have to generate the wav-file samples with the computer.
(e) Oversampling or upsampling changes none of this; it is done for other reasons.
(f) Obviously, a squarewave can only be approximated. The approximation looks better, the higher the ratio of sampling frequency to squarewave frequency. And again, it is the original CD 44.1 kHz sampling frequency which matters, not the frequency with which samples are fed to the DAC (which may of course be higher due to oversampling/upsampling).
(g) Sampling at a higher rate and then downsampling to 44.1 will only give you a squarewave as originally sampled with 44.1. In practice it may look a little better, because filtering is easier to do in this two step process.
(h) The overshoot in the reproduced squarewave, seems to bother some people. It can be avoided by rolling off the bandwith early and slow. However, nothing is gained with that. Such filters usually open the door to significant aliasing distortion. This is why it is recommended to use the fast roll-off of the Buffalo.
(i) The risetime of a bandlimited signal is obviously also limited. The maximum which can be achieved is just slightly bigger than the slope of a sine wave at half the sampling frequency.
(j) If anyone thinks that this is not enough to relistically reproduce a cymbal attack, please consider that such a signal is by no means a squarewave. What needs to be reproduced here is the envelope of a high frequency ac-signal. Of course, sampling and playing at 96 or 192 will improve things, provided microphones, speakers and your hearing go way up into ulrasonics.
(k) If you see a nearly perfect 1000 Hz squarewave at the output of a CD-DAC, something is fishy. Such a DAC might use special tricks and artifacts to create such an output signal. I would not want it for music reproduction.
(l) The output of the Buffalo looks absolutely the way it should. A properly bandlimited squarewave would look like that prior to sampling and it looks this way after restoration at the Buffalo output. No reason to change anything whatsoever.
(m) If you don't like the squarewave response you are getting from a digital system, you might want to try this on vinyl.
Kurt
Wow, it really may be time to end this fruitless squarewave discussion. Let me put a few things into perspective:
(a) A square wave has very slowly decaying harmonics up to infinite frequency. It is by no means bandlimited.
(b) The signal you want to convert to digital MUST be rigourosly bandlimited to less than half the sampling frequency. If you don't do that, aliasing distortion results.
(c) Aliasing distortion cannot be removed later on.
(d) If you use a soundcard, you are responsible for properly bandlimiting the signal you input; the soundcard is not doing that for you. Garbage in, garbage out. If you want to burn a test CD for testing squarewave response, you have to generate the wav-file samples with the computer.
(e) Oversampling or upsampling changes none of this; it is done for other reasons.
(f) Obviously, a squarewave can only be approximated. The approximation looks better, the higher the ratio of sampling frequency to squarewave frequency. And again, it is the original CD 44.1 kHz sampling frequency which matters, not the frequency with which samples are fed to the DAC (which may of course be higher due to oversampling/upsampling).
(g) Sampling at a higher rate and then downsampling to 44.1 will only give you a squarewave as originally sampled with 44.1. In practice it may look a little better, because filtering is easier to do in this two step process.
(h) The overshoot in the reproduced squarewave, seems to bother some people. It can be avoided by rolling off the bandwith early and slow. However, nothing is gained with that. Such filters usually open the door to significant aliasing distortion. This is why it is recommended to use the fast roll-off of the Buffalo.
(i) The risetime of a bandlimited signal is obviously also limited. The maximum which can be achieved is just slightly bigger than the slope of a sine wave at half the sampling frequency.
(j) If anyone thinks that this is not enough to relistically reproduce a cymbal attack, please consider that such a signal is by no means a squarewave. What needs to be reproduced here is the envelope of a high frequency ac-signal. Of course, sampling and playing at 96 or 192 will improve things, provided microphones, speakers and your hearing go way up into ulrasonics.
(k) If you see a nearly perfect 1000 Hz squarewave at the output of a CD-DAC, something is fishy. Such a DAC might use special tricks and artifacts to create such an output signal. I would not want it for music reproduction.
(l) The output of the Buffalo looks absolutely the way it should. A properly bandlimited squarewave would look like that prior to sampling and it looks this way after restoration at the Buffalo output. No reason to change anything whatsoever.
(m) If you don't like the squarewave response you are getting from a digital system, you might want to try this on vinyl.
Kurt
Kurt, I very much appreciate your input here. Thanks!
That also goes for Dustin, Leo, Ross and the rest of you. Many of you have had positive contributions to the project in general and to my growing experience in general. Thanks!
I don't consider these technical discussions off topic. I know enough to know that I still have a lot to learn.
I just want to keep things civil.
Now from some brown ale...
Cheers!
Russ
That also goes for Dustin, Leo, Ross and the rest of you. Many of you have had positive contributions to the project in general and to my growing experience in general. Thanks!
I don't consider these technical discussions off topic. I know enough to know that I still have a lot to learn.
I just want to keep things civil.
Now from some brown ale...
Cheers!
Russ
That's not what I meant. Even if you could cut a perfect squarewave on vinyl, the stylus would never be able to trace it. Thus, you would have to significantly bandlimit your squarewave prior to recording. Back at square one? No, the square wave on vinyl would probably look a lot worse than via CD.
Kurt
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.