Actually, I have plans to attach a small board, containing only 6 parts (complete DAC) to my tiny copper bar transport and connect DAC by I2S directly. The power to the DAC will come from 12V battery.
This is more or less a bare transport with a freelodaing DAC
I suspect that this may be an interesting combination. I just wanted to make one more attempt to see if reclocking is worthwile to consider.
http://brian.darg.net/cdpro2-pdaniel
This is more or less a bare transport with a freelodaing DAC
I suspect that this may be an interesting combination. I just wanted to make one more attempt to see if reclocking is worthwile to consider.http://brian.darg.net/cdpro2-pdaniel
Bernhard said:
Here is a very informative page on that:
http://www.play-hookey.com/digital/experiments/rtl_schmitt.html
Bernhard said:
I tried transformers and didn't like them. After reading the article by Pete Goudreau on modifying Sonic Frontiers DAC, it seems he didn't like them either. http://groups.google.com/groups?selm=4te25c$hcn@agate.berkeley.edu&output=gplain
I added triplets to all chips (suggested by him) and must admit that they really improve the sound. I tried SMD 0.01 only (in the beginning), but adding additional 0.1 on top opens the sonics a bit. I'm getting also very good results with BG N (16V) and NX (6.5V) caps in all bypass positions. The best I tried so far. NX HiQ sound a bit softer and N has more direct presentation. Orientation of those caps is important also.
Konnichiwa,
The input driver CAN make a serious improvement, as can be using a TTL Level "direct" output in S/P-DIF format to the CS8412/14 inputs.
Here is why. The Cirrus Logic Receivers are optimised to operate using Pro-Level interfaces. That means 3...10V P-P or a push-pull TTL output. Conventional S/P-DIF is standardised as 0.5V P-P or in other words 20db lower. No matter how good the triggering of the Inputs is, having a signal level around 20db below that apropriate to the interface as designed will mean the jitter performace is compromised.
BTW, the 47Labs transport has two interface options (AFAIK), one 5V PP via the "DC" output, the other conforming to S/p-DIF according to spec. Guess which one sounds better?
BBTW, as long as you keep S/P-DIF connections below around 10" you can tolerate quite drastic mismatches, like running TTL via loose wires to a 75Ohm terminated or even UNTERMINATED input.
Maybe something to think.
Sayonara
Peter Daniel said:
The input driver CAN make a serious improvement, as can be using a TTL Level "direct" output in S/P-DIF format to the CS8412/14 inputs.
Here is why. The Cirrus Logic Receivers are optimised to operate using Pro-Level interfaces. That means 3...10V P-P or a push-pull TTL output. Conventional S/P-DIF is standardised as 0.5V P-P or in other words 20db lower. No matter how good the triggering of the Inputs is, having a signal level around 20db below that apropriate to the interface as designed will mean the jitter performace is compromised.
BTW, the 47Labs transport has two interface options (AFAIK), one 5V PP via the "DC" output, the other conforming to S/p-DIF according to spec. Guess which one sounds better?
BBTW, as long as you keep S/P-DIF connections below around 10" you can tolerate quite drastic mismatches, like running TTL via loose wires to a 75Ohm terminated or even UNTERMINATED input.
Maybe something to think.
Sayonara
I got this question by email:
So just to answer this here:
I fed the analogue output of the DAC to a Terratec 24 bit 96Khz soundcard line in input.
Then I used a spectrum-analyzer of an Audio program (you can use whatever you like: Soundforge, CEP, Wavelab, etc).
After trimming the DAC, I recorded some 30 seconds of signal, and had it scanned by the spectrum analyzer to get the overall picture.
It is really that simple...
So just to answer this here:
I fed the analogue output of the DAC to a Terratec 24 bit 96Khz soundcard line in input.
Then I used a spectrum-analyzer of an Audio program (you can use whatever you like: Soundforge, CEP, Wavelab, etc).
After trimming the DAC, I recorded some 30 seconds of signal, and had it scanned by the spectrum analyzer to get the overall picture.
It is really that simple...
Yesterday evening I took to the Doede Dac with me to a friend of mine, who owns large ESL's.
We compared the DDDAC in its present state to his not so young SONY CDP 50 ES player. At the time it came out (some 5 years ago I think) this was a pretty could sounding SONY, and beautifully build of course.
I had thought that DDDAC would sound much better, but unfortunately it didn't. It was a little bit more detailed, but for some reason it was sounding thinner and somewhat distorted. I kept asking myself whether there were phase problems in the sound.
The funny thing is that he uses the same preamp (An AVM Evolution V3) so it cannot be impedance matching that causes a distortion not heard in my setup.
Could it be that SPDIF from the SONY does not fit with DDDAC designed digital input? With my Philips it seems to be pretty clean.
What would be a better input configuration for the CS8412? Any advice appreciated!
Lucas
We compared the DDDAC in its present state to his not so young SONY CDP 50 ES player. At the time it came out (some 5 years ago I think) this was a pretty could sounding SONY, and beautifully build of course.
I had thought that DDDAC would sound much better, but unfortunately it didn't. It was a little bit more detailed, but for some reason it was sounding thinner and somewhat distorted. I kept asking myself whether there were phase problems in the sound.
The funny thing is that he uses the same preamp (An AVM Evolution V3) so it cannot be impedance matching that causes a distortion not heard in my setup.
Could it be that SPDIF from the SONY does not fit with DDDAC designed digital input? With my Philips it seems to be pretty clean.
What would be a better input configuration for the CS8412? Any advice appreciated!
Lucas
I would suggest using SN75179 input driver as it shown in this schematic: http://www.diyaudio.com/forums/attachment.php?s=&postid=173347
PS and bypass caps are very important here and I'm using AN8005 and BG N caps. Elso is using an AD chip at the input, and also claims big gains.
I did yesterday some comparisons between BG N type caps and NX Hi Q. As much as some say that NX Hi Q are the best grade of BG, I would not really agree with that. Maybe it's a low voltage rating, or maybe something else, but to my ears (and the other set of ears I'm using), BG N sound more clean and direct, especially on vocals. NH Hi Q is rather softer sounding and it gives impression of a some veiling and overall too mellow sound. Too me it's just too smooth, but some others might prefer it. Also, the comparison wasn't 100% accurate, as I only had 22/6.5 and 4.7/50 caps. Partsconnexion were out of 33/16 value and I will be getting them after the weekend. Nevetherless, 4.7/50 N sounded better than 22/6.5 on both receivers' supply pins and on input driver. I'm using 100/16 BG N on the DAC and it works splendidly in that location. If you are using those caps, experiment with leads orientation on the caps, as it makes quite a big difference
As to the I/V resistor choice, I tried Caddocks (MK132), Riken, and parallel combination of Riken and Dale and Riken and AN tantalum. Caddock was a first one to go (sounded too edgy), tantalum and Riken wasn't bad but I could perceive some string midrange coloration with disjointed highs and after a while it became bothersome. Riken//Dale wasn't actually bad, but the sound wasn't completely clean and one could sense a bit of abrassivness that Dales are known for. In the end I settled on 0.5W Rikens, as those sounded most clean, natuaral and convincing. The highs are silky smooth and bass is very deep. I always liked Rikens in that position. The only value that comes close to 3K is 2k7, so I guess I have to readjust Rref, but even now, with 1k6 it sounds fine.
PS: As to the sound of Doede DAC not matching Sony, I have serious doubts if paralleling 8 chips ever had any chances of sounding relatively good
PS and bypass caps are very important here and I'm using AN8005 and BG N caps. Elso is using an AD chip at the input, and also claims big gains.
I did yesterday some comparisons between BG N type caps and NX Hi Q. As much as some say that NX Hi Q are the best grade of BG, I would not really agree with that. Maybe it's a low voltage rating, or maybe something else, but to my ears (and the other set of ears I'm using
), BG N sound more clean and direct, especially on vocals. NH Hi Q is rather softer sounding and it gives impression of a some veiling and overall too mellow sound. Too me it's just too smooth, but some others might prefer it. Also, the comparison wasn't 100% accurate, as I only had 22/6.5 and 4.7/50 caps. Partsconnexion were out of 33/16 value and I will be getting them after the weekend. Nevetherless, 4.7/50 N sounded better than 22/6.5 on both receivers' supply pins and on input driver. I'm using 100/16 BG N on the DAC and it works splendidly in that location. If you are using those caps, experiment with leads orientation on the caps, as it makes quite a big differenceAs to the I/V resistor choice, I tried Caddocks (MK132), Riken, and parallel combination of Riken and Dale and Riken and AN tantalum. Caddock was a first one to go (sounded too edgy), tantalum and Riken wasn't bad but I could perceive some string midrange coloration with disjointed highs and after a while it became bothersome. Riken//Dale wasn't actually bad, but the sound wasn't completely clean and one could sense a bit of abrassivness that Dales are known for. In the end I settled on 0.5W Rikens, as those sounded most clean, natuaral and convincing. The highs are silky smooth and bass is very deep. I always liked Rikens in that position. The only value that comes close to 3K is 2k7, so I guess I have to readjust Rref, but even now, with 1k6 it sounds fine.
PS: As to the sound of Doede DAC not matching Sony, I have serious doubts if paralleling 8 chips ever had any chances of sounding relatively good
Peter Daniel said:PS: As to the sound of Doede DAC not matching Sony, I have serious doubts if paralleling 8 chips ever had any chances of sounding relatively good
Hi Peter,
What's up? normally I respect your thorough judgments and pioneer work, doing tests and then give your opinion. This last statement seems to be a drop of the ball ?? Where are your doubts based on? Many tests, with many people in many different chains have given proof that the 8 chip version sounds MUCH better than the 1 chip version. By just doubting this will not sound good, you share yourself in the company of Bernhard, who only believe that some thing can sound good if the datasheet has the best specs. At least you are not so persuasive on it
Whether the dddac dac sound better in set up A with speakers B and drive C, no one knows till it is tried (pudding and eating, remember?) I cannot judge this and this remains solely to the opinion from the listening audience at that particular session. At home I have no problem beating my Sony 777ES in CD. But again, that is no reference for Lucas, as he found it less with the Sony ES50. Which of course is just fine, may be something was not optimal, I don't know, may be he finds out, may be the Sony 50es is just very good, who can tell? But I do know it has nothing to do with paralleling 8 dacs...............
kind regards
doede
dddac said:
Well, usually I never say anything I didn't try before. I tried 2 parallel chips and 4 parallel chips and was greatly disappointed with the sound.
To give you idea how it sounded, you can read my observation on asynchronous reclocking in previous posts. The difference was quite similar.
Attachments
dddac said:
Hi Peter,
What's up? normally I respect your thorough judgments and pioneer work, doing tests and then give your opinion. This last statement seems to be a drop of the ball ?? Where are your doubts based on? Many tests, with many people in many different chains have given proof that the 8 chip version sounds MUCH better than the 1 chip version. By just doubting this will not sound good, you share yourself in the company of Bernhard, who only believe that some thing can sound good if the datasheet has the best specs. At least you are not so persuasive on it
Whether the dddac dac sound better in set up A with speakers B and drive C, no one knows till it is tried (pudding and eating, remember?) I cannot judge this and this remains solely to the opinion from the listening audience at that particular session. At home I have no problem beating my Sony 777ES in CD. But again, that is no reference for Lucas, as he found it less with the Sony ES50. Which of course is just fine, may be something was not optimal, I don't know, may be he finds out, may be the Sony 50es is just very good, who can tell? But I do know it has nothing to do with paralleling 8 dacs...............
kind regards
doede
Hi Doede,
Rest assured: I will not give up so lightly on this DAC.
I have put high hopes on better decoupling caps. Still waiting on some BG-N's to arrive...
In my setup the sound is already reasonable good, but not really warm and refined.
I am listening to a speaker system with 17 cm Etons and Eton ER4 AMT tweeter. (see this thread: http://www.diyaudio.com/forums/showthread.php?s=&threadid=13476&highlight= )
These AMT tweeters are quite revealing as to the quality of treble reproduction. However, if I recall right you yourself are familiar with this speaker, and if the DAC sounds so superb with you, I will trust your judgement and continue to improve the DAC.
Regards,
Lucas
UrSv said:
Of course it could be different and I didn't test it. But what I observed with 4 chips simply discourages me from trying even more.
Please note, I never said it couldn't sound great. I only expressed my opinion that I doubt it may sound good. I hope you see the difference here.
I'd even like to build 8chip dac, to know for sure, but I don't have that much time anymore.
Lucas_G said:
Those DACs are sensitive to everything and building one doesn't mean that it will sound the same as the other one. Depending on parts choices, PS decisions, hell; even brass or copper feet, will make a difference.
For the record, I never observed a difference in implementation of copper and brass, that I'm aware of.
misunderstanding
I did not say that it could not sound good.
8 in parallel could improve the poor linearity.
Thats why I put 4 S1 in parallel to make a good thing even better.
But why use the inferior chip if a better one is available ?
Thats like building amps with 2N3055.
1541A is not that expensive...
If Philips gives linearity 1/2 LSB for one chip and 2LSB for the other, I take it as a fact.
Bernhard
dddac said:
I did not say that it could not sound good.
8 in parallel could improve the poor linearity.
Thats why I put 4 S1 in parallel to make a good thing even better.
But why use the inferior chip if a better one is available ?
Thats like building amps with 2N3055.
1541A is not that expensive...
If Philips gives linearity 1/2 LSB for one chip and 2LSB for the other, I take it as a fact.
Bernhard
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.