Actually, this wouldn't "eliminate" the FIFO, the DAC just wouldn't rely on an external clock anymore. In the slaved devices the FIFO would still sync to the clock of the master DAM1021. Since the clock would originate locally it is to be expected that performance would be better. Especially compared to something like an RME sound card, which is very convenient to work with but unfortunately has poor quality clocking.
Unfortunately the statements in the post you have pointed out are totally wrong.
Regardless of the price, the MSB Galaxy Femto Clock has a phase noise around -134 dBc at 10 Hz from the carrier, that means it's a state of the art oscillator at such that frequency (22/24 MHz).
While the Si570 is a crappy oscillator just as indicated in its datasheet.
You can take a look at this link Si570 phase noise - YFL Proj
The Si570 was measured with the R&S FSWP signal source analyzer and the conclusion is "the Si570 is 58 times as noisy as the HS9001A", but also "1233 times cheaper!".
The Oscillator Phase Noise Look-Up Tool on Silabs website shows -127 dBc at 100 Hz from the carrier that probably means -80 dBc at 10 Hz from the carrier, so more than 50 dBc worse than the MSB clock.
And please note that Silabs has measured the Si570 with a integration bandwidth suitable for telecommunication, since the lower limit is 100 Hz.
I'm not sure the quality of the RME clock is much worse than the Si570, I don't believe since the quality of the Silabs clock is very poor.
You should measure the phase noise of the RME clock to compare with the Si570.
Be ready for a big surprise from the comparison.
I think it would be strategic to not be so sure about the clock impact on SQ - it might bite back if a truly scientific test of pn impact could be performed.
So far, only anecdotic tales talk of nirvana. Other more "cooled of" tests do not show such a magnificent impact. I have done some myself and I'm less convinced, not to say right-out sceptic, of the last 20-30 dBc/Hz at 1 Hz.... but I bow to the technology effort - very impressive work!
I think that an untouched Si570 would be quite sufficient and that it would be other stuff in the DAM that needs more attention (like 7th and 9th overtone levels - see measurements at the "friendly" bunch site) to make it go all the way to the top.
But, its very good as is. And the designer is happy so...
Andrea - did you give it a listen yet :-D - no, I don't think you are interested in that really - why would you?
//
So far, only anecdotic tales talk of nirvana. Other more "cooled of" tests do not show such a magnificent impact. I have done some myself and I'm less convinced, not to say right-out sceptic, of the last 20-30 dBc/Hz at 1 Hz.... but I bow to the technology effort - very impressive work!
I think that an untouched Si570 would be quite sufficient and that it would be other stuff in the DAM that needs more attention (like 7th and 9th overtone levels - see measurements at the "friendly" bunch site) to make it go all the way to the top.
But, its very good as is. And the designer is happy so...
Andrea - did you give it a listen yet :-D - no, I don't think you are interested in that really - why would you?
//
The DAM is still in the box, now I'm busy arranging the GB (apart my real job and a little of private life such as eating and sleeping a few hours).
I will give it a listen as soon as I find the time to give it a power supply, maybe at the end of January.
I will also measure the phase noise of the LRCK.
I'm not able to do a scientific test of the impact of the PN on the sound quality, neither I can explain the impact by the physical point of view.
I believe noone can explain the relation, like noone can explain why a DHT amplifier with 2% of THD sounds much better than a solid state amp with 0.00000.... % THD.
I will give it a listen as soon as I find the time to give it a power supply, maybe at the end of January.
I will also measure the phase noise of the LRCK.
I'm not able to do a scientific test of the impact of the PN on the sound quality, neither I can explain the impact by the physical point of view.
I believe noone can explain the relation, like noone can explain why a DHT amplifier with 2% of THD sounds much better than a solid state amp with 0.00000.... % THD.
I think it would be strategic to not be so sure about the clock impact on SQ - it might bite back if a truly scientific test of pn impact could be performed.
So far, only anecdotic tales talk of nirvana. Other more "cooled of" tests do not show such a magnificent impact. I have done some myself and I'm less convinced, not to say right-out sceptic, of the last 20-30 dBc/Hz at 1 Hz.... but I bow to the technology effort - very impressive work!
I think that an untouched Si570 would be quite sufficient and that it would be other stuff in the DAM that needs more attention (like 7th and 9th overtone levels - see measurements at the "friendly" bunch site) to make it go all the way to the top.
But, its very good as is. And the designer is happy so...
Andrea - did you give it a listen yet :-D - no, I don't think you are interested in that really - why would you?
//
And there is another perspective I have pointed out several times but it looks like no one has understood.
Our oscillators are only a little part of the audio system we are developing, we are looking to design all the part of the audio chain as state of the art devices.
Soeren designs his devices to be cost effective, decent quality at affordable prices.
We don't have this limit, we are looking for the best regardless of the budget.
I respect Soeren's choice, he has to sell his device, it's a pure commercial business, so his devices will be compared to the one of his competitors.
We don't care about competitors simply because we have no competitors, we are developing this project for ourselves.
Andrea, I did not mean to express an opinion, but to recall Sørens statement.
It's not that bad and likely more than good enough for audio. Let's face reality, you are a jitter fetishist.
The reality is very simple and comes directly from the datasheet and the measurement: the Si570 is a very high phase noise oscillator, much worse than the Crystek CCHD-957 that's already a bad oscillator.
Suitable for telecommunication but not for digital audio.
You are free to use whatever oscillator you want and you are free to claim it's the best device for your taste, but from the technical point of view the Si570 is a bad oscillator.
It's not "more than good enough for audio", it's "much less than good enough for audio".
-80 dBc or maybe worse at 10 Hz from the carrier for a 24.576 MHz oscillator is a very poor phase noise performance, one of the worst ever measured.
Moreover using its PLL to track the input seems to degrade the performance further as reported from many users.
In my real life I work on IT, so I well know how using software instead of hardware is simpler and cheaper, but I also well know what results I have to expect.
Suitable for telecommunication but not for digital audio.
You are free to use whatever oscillator you want and you are free to claim it's the best device for your taste, but from the technical point of view the Si570 is a bad oscillator.
It's not "more than good enough for audio", it's "much less than good enough for audio".
-80 dBc or maybe worse at 10 Hz from the carrier for a 24.576 MHz oscillator is a very poor phase noise performance, one of the worst ever measured.
Moreover using its PLL to track the input seems to degrade the performance further as reported from many users.
In my real life I work on IT, so I well know how using software instead of hardware is simpler and cheaper, but I also well know what results I have to expect.
I think that no-one can have missed your position on this Andrea ;-D
Your repetitions start to get boring.
You say you respect Sörens choice of oscillator - ok, then you don't have to repeatedly call it crap in this thread?
I think its "bad form" to spew negative critique in a "competitors" thread. To defend my own forwarded criticism - I'm at least a user and a customer. Your purchase seems to have been done more like in spite.
//
Your repetitions start to get boring.
You say you respect Sörens choice of oscillator - ok, then you don't have to repeatedly call it crap in this thread?
I think its "bad form" to spew negative critique in a "competitors" thread. To defend my own forwarded criticism - I'm at least a user and a customer. Your purchase seems to have been done more like in spite.
//
Indeed, it is. There is no data, nothing, nada showing that close-in phase noise is an audible problem, nevermind at -80 dBc @ 10 Hz offset.
Please don’t take this as an invitation to debate further in Soren’s thread. I have no interest anyway. You have your own thread to discuss solutions to imagined problems. I do admire your work, I just don’t think it’s nice to beat people up over reasonable design decisions in their own thread.
Last edited:
I respect the choice of the oscillator for a cheap DAC suitable for watching movies when listening to music, although I don't agree any of the projectual choices.
But you cannot say "spew negative critique in a competitors thread" for at least two reasons:
- what I claim about the Si570 is confirmed by the datasheet and the measurement, so my statements are based on technical documentation; you can do the measurement yourself to demonstrate the contrary if you want, othwerwise please respect the incontrovertible reality
- I'm not a competitor, I own an IT software consulting company (ERP applications) not a "network, communication and embedded computer design" company; audio devices are an hobby for me not a business, can you understand?
Last edited:
You are wrong, this is a audio community thread in a diy audio forum, so I'm free to express my opinion about the projectual choices like any other member of the forum.
Even more if my statements are supported by the manufacturer datasheet and by the measurements.
You are free to skip my posts if you are not interested on.
And I'm not the one who claimed that the DAC is source dependent, but it's not a big surprise for me, it was enough clear from the architecture of the device.
When you track the source with a PLL the output follows the input, maybe adding more phase noise.
Are you happy with the device?
Well, I'm glad for you.
Do you want me to think with your mind?
I'm sorry, I have my opinions based on technical reasons.
A little example to clarify:
The DAC is source dependent
then
the FIFO does not work
because
with a FIFO the DAC and the source operate in different time domains so nothing coming from the source can be reflected to the DAC
Søren said in his first post: "I believe that the sound quality will be the absolute best". So there is a high demand for it. I would be very happy if an oscillator specialist like Andrea would give me such hints if it was my project.
I always find it very annoying when you bring FIFOs into play, for example, that don't really buffer "decoupled" in the end. This is also the case with Ian. It is always suspicious how much changes have an audible(!) effect far before the buffer, which should actually theoretically no longer allow any effect... It takes a certain openness and perfection to acknowledge this problem, to the extent that it is verifiably present. However, there is never a need for insults...
For listening: One reads again and again how something would sound. But nobody says anything about the listening environment. Do you go (in extreme cases) with the 1.4 volt output of the DAM into a high power amplifier (e.g. without input buffer) and wonder why there is no dynamic and power to hear? What I want to say: Listening should only begin when the system is theoretically and metrologically correct. Then, for example, the input and output impedances, etc., of the entire environment should be illuminated before listening begins. So, all conditions should satisfy the claim that Søren expressed at the beginning.
This is where the headphones can be helpful. However, they are never suitable for the final acoustic assessment.
I always find it very annoying when you bring FIFOs into play, for example, that don't really buffer "decoupled" in the end. This is also the case with Ian. It is always suspicious how much changes have an audible(!) effect far before the buffer, which should actually theoretically no longer allow any effect... It takes a certain openness and perfection to acknowledge this problem, to the extent that it is verifiably present. However, there is never a need for insults...
For listening: One reads again and again how something would sound. But nobody says anything about the listening environment. Do you go (in extreme cases) with the 1.4 volt output of the DAM into a high power amplifier (e.g. without input buffer) and wonder why there is no dynamic and power to hear? What I want to say: Listening should only begin when the system is theoretically and metrologically correct. Then, for example, the input and output impedances, etc., of the entire environment should be illuminated before listening begins. So, all conditions should satisfy the claim that Søren expressed at the beginning.
This is where the headphones can be helpful. However, they are never suitable for the final acoustic assessment.
Last edited:
As Ian don't support variable oscillators, it is truly "decoupled".
I hope you realise that Sören do know all this technology. He just makes a choice not to implement it. This is based on his own evaluation that it is not needed and want to provide the product in a certain price range. He has reached his goals. Now it is up to us to judge wether we are satisfied or not.
"the system is theoretically and metrologically correct" - this does not exists so its a moot point.
Listening is subjective and probably varies per/over day - as do your body temperature, sensitivity to taste and sharpness of vision.
Which DAC do you think sound better and when did you hear it? In which system?
//
I hope you realise that Sören do know all this technology. He just makes a choice not to implement it. This is based on his own evaluation that it is not needed and want to provide the product in a certain price range. He has reached his goals. Now it is up to us to judge wether we are satisfied or not.
"the system is theoretically and metrologically correct" - this does not exists so its a moot point.
Listening is subjective and probably varies per/over day - as do your body temperature, sensitivity to taste and sharpness of vision.
Which DAC do you think sound better and when did you hear it? In which system?
//