I've been using Rhopoint 120ohm for a while and been very happy, but just got some 133ohm tx2725 to go in. Texas Components will make any value you like for no extra cost if you're prepared to wait a couple of weeks
In terms of developing better SQ this is, I believe the best way to develop a completed DAC as we are then the consistent comparator. But we only find out how we have gone when we can directly compare DACs. This is all we are really saying otherwise we are in the dark. Doede has already done most of this and enlightened us. And who knows yet, what he will do next. We can look forward to that too.
Hmmm, I'm not doubting that you hear improvements, btw., I think your build is impressing!
Nevertheless - let's assume an unlikely but possible scenario: Let's assume, you first built a 4-decker and 1 board was somehow broken which distorted the signal but you never knew that. Then you built your stack up to 11 decks, whereas the 10 good boards compensate the broken board, so the audio gets better - but not due to any parallelizing effect but due to overriding a distorted signal.
You may think that such a scenario is unlikely, but quite on the contrary I suspect that such things may be quite common. For instance, it may well be that my S03 recklocker I built has a broken buffer (as I incidentally swapped 3 and 5V, it's still unclear if the buffer is broken, btw.) and adds some jitter to my design instead of removing it. I'd never had got that if I would not have cross-checked the circuit with my DSO.
For that reason I'm suggesting some measurement setup so that anyone can (with a limited amount of investment) check if his build complies to the standards of a reference design (e.g. of Doedes build).
Unfortunately I'm not very experienced in choosing the right equipment for that, but I assume that something like this hardware could be a start:
UR22:|http://www.steinberg.net/
or perhaps that (has SPDIF in/out which could be nice for testing, however, is maybe not bitcorrect as it has some digital mixer implemented):
Product: US-366 | TASCAM
I think it's a pity that it seems not yet to be possible to get the jitter of an RPi/BBB below that one of a common CD-Player as the following graph shows (taken from https://hifiduino.wordpress.com/2014/11/13/raspberry-pi-b-digital-audio/ ):
Perhaps this would work with Ians reclocker circuit? Together with the BBB, which can do 24bit (the RPi can only do 16bit) this could maybe lower the jitter and thus improve the sound even more.
Best Regards,
Hermann
I think you've raised some interesting questions regarding jitter and I'm looking forward to your further findings. There are obvious merits in empirical measurements, but at the end of the day it's personal perception that counts. I think it's an utopian thought to have a bench mark test setup. I had the privilege to listen/compare my DAC, and if at all possible I would highly recommend trying to organize a "local" get together.
Over the last 1.5year of following this thread, I've seen the DDDAC take a lot of abuse by myself and others, with very few casualties. These are seriously though boards. It's unlikely Chanh has a faulty board (both left and right channels at the same time?). Especially since he build his very meticulously.
The way the modifications have come about might seem slightly subjective. But there are a few here who know who's who and what's what.
Last edited:
try bridging from pin3 of the buffer to pin6 to bypass it and see if you see any difference. I have mine set like that just for simplicity
An externally hosted image should be here but it was not working when we last tested it.
I think you've raised some interesting questions regarding jitter and I'm looking forward to your further findings. There are obvious merits in empirical measurements, but at the end of the day it's personal perception that counts. I think it's an utopian thought to have a bench mark test setup. I had the privilege to listen/compare my DAC, and if at all possible I would highly recommend trying to organize a "local" get together.
Over the last 1.5year of following this thread, I've seen the DDDAC take a lot of abuse by myself and others, with very few casualties. These are seriously though boards. It's unlikely Chanh has a faulty board (both left and right channels at the same time?). Especially since he build his very meticulously.
The way the modifications have come about might seem slightly subjective. But there are a few here who know who's who and what's what.
Good to know that the DDDAC is quite sturdy (that's fortunate!).
Regarding personal perception - yes in the end it's what really counts. But what exactly is "personal perception"? Is this really something we can rely on? I give you some examples/thoughts:
- When finishing another DDDAC for a friend some weeks ago we did some listening tests with 5 friends, all of us are not that experienced but have done some listening in our lifes. We checked the DDDAC against an older Sony CD-player and found that the sound of the DDDAC was superior, finer detail, good soundstage etc. Later on we found out that due to a wrong driver in the RPi, all FLAC-files we played were mono(!). So - we all 5 were overwhelmed with the richness of the sound and the broad soundstage of the DDDAC and overlooked that it was mono (later on the sound was even better in stereo ). So - what happened? Were we all half-deaf at that day or simply carried away in some way over our "cool build"?
- Setting up good listening tests seems to me quite complex. For instance at least for me it is very difficult to remember the perception over time, especially if I e.g. change the I2S signal from the DIYINKH isolator to the S03 reclocker - I simply "forgot" how it sounded before. So the only way to me seems to switch the audio directly between two sources, however, this is also not that simple: Often, I have no second test unit I can compare it to. Moreover I need to sit in the sweet spot of my speakers, where I don't have any means to manually switch the audio signal, moreover, the audio volume has to be the very same so that I'm not tricked by it. All that may improve with listening experience, but I doubt that we won't be tricked by our minds from time to time. At least (and that's most important) it's fun doing listening tests!
- I know quite some people who put some sticker on their mobile phone to reduce the radiation and report that they feel a lot less stressed when phoning. Although this sticker is - as most of us know - nonsensical, their personal perception is altered, but does that mean that in case someone finds out that there is a "darker background and better soundstage" after drawing mandalas on his caps, should we take this as objective and do the same?
- I wonder if and how the personal perception of listeners alters as we know now that the S03 reclocker in asynchronous operation with the RPi very likely increases jitter from 2 to 10ns? Will people deduce from this that higher jitter sounds better with the DDDAC? Or perhaps change to the theory that the sound improvement was not jitter-related at all but due to e.g. the isolator ICs on the S03 board?
- Listening tests
- Cross checking with measurements
- Close collaboration, sharing of setups, measurement setups, listening test setups etc.
Best Regards,
Hermann
Hmmm, I'm not doubting that you hear improvements, btw., I think your build is impressing!
Nevertheless - let's assume an unlikely but possible scenario: Let's assume, you first built a 4-decker and 1 board was somehow broken which distorted the signal but you never knew that. Then you built your stack up to 11 decks, whereas the 10 good boards compensate the broken board, so the audio gets better - but not due to any parallelizing effect but due to overriding a distorted signal.
You may think that such a scenario is unlikely, but quite on the contrary I suspect that such things may be quite common. For instance, it may well be that my S03 recklocker I built has a broken buffer (as I incidentally swapped 3 and 5V, it's still unclear if the buffer is broken, btw.) and adds some jitter to my design instead of removing it. I'd never had got that if I would not have cross-checked the circuit with my DSO.
For that reason I'm suggesting some measurement setup so that anyone can (with a limited amount of investment) check if his build complies to the standards of a reference design (e.g. of Doedes build).
Unfortunately I'm not very experienced in choosing the right equipment for that, but I assume that something like this hardware could be a start:
UR22:|http://www.steinberg.net/
or perhaps that (has SPDIF in/out which could be nice for testing, however, is maybe not bitcorrect as it has some digital mixer implemented):
Product: US-366 | TASCAM
I think it's a pity that it seems not yet to be possible to get the jitter of an RPi/BBB below that one of a common CD-Player as the following graph shows (taken from https://hifiduino.wordpress.com/2014/11/13/raspberry-pi-b-digital-audio/ ):
Perhaps this would work with Ians reclocker circuit? Together with the BBB, which can do 24bit (the RPi can only do 16bit) this could maybe lower the jitter and thus improve the sound even more.
Best Regards,
Hermann
Very revealing and on face value shows ES9023 USB 12S1 is hugely the best if it can be taken on face value. Doede on his round up evaluation meeting if I remember found the USB input not the best. So it is not just the jitter which is the spectre at the top of the list. The phase noise is just where you dont want it and vinyl of course has none of it. So vinyl versus CD SACD FLAC versions of genre, carefully selected would make good direct comparisons at a meet.
I have to keep reminding myself that a lot of music is enjoyed as direct live broadcasts with simpler electronic chain from actual music event to reproduction equipment to ear. BTW I listen to some BBC R3 FM live stuff that can be hair raisingly real.
Again has anyone any idea of where a valve DAC buffer sits compared to the solid state i/v for SQ. This question cannot go away for anyone one who has not considered it.
Hi All,
A sonically satisfying 2015 for all!
@rhlauranna,
I belief that moving to more dac boards will bring better sound quality. I only tried to point out that there are more ways to improve and an order can be made for costs and efforts.
If you want cost no object than Chans build is a nice example!
@ Hermann,
It is my experience in 25 years audio engineering/building that measurements do not always corrolate to sound quality. Compare for instance a single ended triode amp with a transistor amp. The latter will always measure better; but my ears tell me what sounds better
As making good measurements and quantifying these is not easy, I second Stijn that this is a utopian task.
Regarding the jitter of the acko reclocker: I have some experience in listening to jitter levels and I think I can tell if jitter is reduced or not. To my ears the jitter is lower when running the rpi through the acko.
The sound becomes more clear, better focus, more dynamic with the acko. If the jitter would be higher this will be easily audible.
So it might be that this is an example that measuring and understanding the measurements is not so easy!
Your ears are the best measuring instruments, train your ears, trust them, learn to correlate what you hear, and most of all, enjoy the music!
Regards,
A sonically satisfying 2015 for all!
@rhlauranna,
I belief that moving to more dac boards will bring better sound quality. I only tried to point out that there are more ways to improve and an order can be made for costs and efforts.
If you want cost no object than Chans build is a nice example!
@ Hermann,
It is my experience in 25 years audio engineering/building that measurements do not always corrolate to sound quality. Compare for instance a single ended triode amp with a transistor amp. The latter will always measure better; but my ears tell me what sounds better
As making good measurements and quantifying these is not easy, I second Stijn that this is a utopian task.
Regarding the jitter of the acko reclocker: I have some experience in listening to jitter levels and I think I can tell if jitter is reduced or not. To my ears the jitter is lower when running the rpi through the acko.
The sound becomes more clear, better focus, more dynamic with the acko. If the jitter would be higher this will be easily audible.
So it might be that this is an example that measuring and understanding the measurements is not so easy!
Your ears are the best measuring instruments, train your ears, trust them, learn to correlate what you hear, and most of all, enjoy the music!
Regards,
I have used a valve I/V converter in a dac that I used before the dddac. The tube was a 6922 and I tweaked the design and power supply to the max.
I have found the direct balanced output of the dddac to be more open, uncoloured and transparant in comparison; and that was the unmodified standard dac design.
I think this is an example of: "less is more".
You must take into account that I use fully directly heated heated triode amps with an audio consulting silver tvc. So my findings must be seen in this relation.
Regards,
Last edited:
Both your quotes are well received. It fits in with my expectation. Less is often more, or occasionally different. Even some op amps produce a great sound, when carefully selcted for buffering the i/v out. The problem manufacturers have is they often keep adding more (perceived added value) to justify selling price. When you think how simple a dac can be constructed it gives <less for more> but clearly cannot easily be sold for big money.
So this is where the DIYers are the winners. We can go to the heart of the matter and keep it to the least no of parts. I am still impressed with the multistacking nirvana.
Thanks again for your helpful remarks
As less complexity can be better, then I still feel that a well modified single deck can give very good results. If you make a single deck very accurate, then where does the main benefit of adding extra decks come from? Just the fact that more voices singing together give a stronger output? If so, this can be achieved simply and cheaply with a buffer stage. It certainly works well for me I'm in no rush to add extra decks to mine.... How does this solution compare to a well modified multi deck stack? Hopefully we're getting a few dddac guys together soon for a session, so we'll know more after that
I'm in no rush to add extra decks to mine.... How does this solution compare to a well modified multi deck stack? Hopefully we're getting a few dddac guys together soon for a session, so we'll know more after that It is my experience in 25 years audio engineering/building that measurements do not always corrolate to sound quality. Compare for instance a single ended triode amp with a transistor amp. The latter will always measure better; but my ears tell me what sounds better
As making good measurements and quantifying these is not easy, I second Stijn that this is a utopian task.
Regarding the jitter of the acko reclocker: I have some experience in listening to jitter levels and I think I can tell if jitter is reduced or not. To my ears the jitter is lower when running the rpi through the acko.
The sound becomes more clear, better focus, more dynamic with the acko. If the jitter would be higher this will be easily audible.
So it might be that this is an example that measuring and understanding the measurements is not so easy!
Your ears are the best measuring instruments, train your ears, trust them, learn to correlate what you hear, and most of all, enjoy the music!
Regards,
At first, thanks for your feedback as I'm constantly learning from every reply!
Well, no doubt that one can hear things that are very hard if not impossible to measure, especially if someone has such a long experience as you have.
Problem is - at least for me - that my ears are not (yet) that trained, so I need to look out for other means to improve things, which is for me an analytic approach combined with measurements, paired with listening.
For the S03 jitter of the reclocker: Acko has confirmed that the S03 in asynchronous mode has a fixed 10ns jitter as this is a conceptional issue:
http://www.diyaudio.com/forums/group-buys/227502-amanero-isolator-reclocker-gb-155.html#post4168613
So, asynchronous reclocking can only lock the jittered signal to the phase of the master clock, thus introducing so to say "fixed jitter" instead of "random jitter".
The other is the fact that the RPi can only do 16 bit which can be deduced from the measurements of the I2S clock speed:
So this new knowledge - which I could only gain through simple DSO measurements - lead me to other, new ideas as I'd like to have 24bit resolution along with a jitter of < 100ps as I personally think, lower jitter is better. So the best way to do so would be (I think):
- Use the BBB with synchronous reclocking, which has the disadvantage that I currently have to manually switch clocks for different sampling rates.
- Use some other embedded computer that can generate an I2S signal with 16/24 bit depth in all sampling rates - regardless if jitter is added or not (unclear which device this would be - RPi can't do 24bit, BBB can't do 44.1kHz signals). Then combine it with Ian's reclocker, which buffers the signal and thus eliminates any kind of jitter. (However, it seems that it's hard to get).
Hermann
Attachments
Dusty,
Look up the botic driver by miero for the BBB in the TPA forum to solve this problem,
Use some other embedded computer that can generate an I2S signal with 16/24 bit depth in all sampling rates - regardless if jitter is added or not (unclear which device this would be - RPi can't do 24bit, BBB can't do 44.1kHz signals).
You then need two clocks on the s03 and abracadabra peanut butter sandwiches , synchronous clocking at the correct format rate......
Easy and cheap.....well not exactly cheap, but considering the cost of some of our gear......
Laters,
Drew.
Look up the botic driver by miero for the BBB in the TPA forum to solve this problem,
Use some other embedded computer that can generate an I2S signal with 16/24 bit depth in all sampling rates - regardless if jitter is added or not (unclear which device this would be - RPi can't do 24bit, BBB can't do 44.1kHz signals).
You then need two clocks on the s03 and abracadabra peanut butter sandwiches , synchronous clocking at the correct format rate......
Easy and cheap.....well not exactly cheap, but considering the cost of some of our gear......
Laters,
Drew.
As less complexity can be better, then I still feel that a well modified single deck can give very good results. If you make a single deck very accurate, then where does the main benefit of adding extra decks come from? Just the fact that more voices singing together give a stronger output? If so, this can be achieved simply and cheaply with a buffer stage. It certainly works well for me
This has got to be a good way forward and could mirror at least in the UK what Doede did in his country.
Last edited:
Lots of people are using these for powering bbb / rPi / WaveIO / s03 reclocker
http://www.diyinhk.com/shop/audio-k...e-dac-power-supply-regulator-33v55v-1ax4.html
They give you 2 supplies which you can set for either 3.3v or 5v. You also need a Transformer with 2 x 6v to 9v with about 15va
Thinking about multiple DAC chips it may be better that with stacked chips running on their own clock will actual smooth out the jitter with slightly diffferent timings giving a slighlty softened wave form that still may represent a better audible treble? Using a fast external clock may be partly counterproductive producing a jitter summing and accurately overlay each DACs jitter reinforcing it.
Any chip expert got a view on this
Last edited:
I think the reason it sounds "better" is that the signal is being averaged over a number of chips that will all be ever so slightly different internally. I figure that the resulting waveform running through a single IV resistor/stage has a summing effect and is then more accurate to the original source. If each dac ran it's own IV resistor there would probably not be the same result in performance.
That's about what I figure anyway, may be right, may be wrong.
Laters,
Drew.
That's about what I figure anyway, may be right, may be wrong.
Laters,
Drew.