If these numbers
Were the only difference, then no one could possibly hear the difference, as these are all below the threshold of audibility in any real world system. This just confirms my suspicion that dual mono, while maybe theoretically better, may offer no audible benefit.
Were the only difference, then no one could possibly hear the difference, as these are all below the threshold of audibility in any real world system. This just confirms my suspicion that dual mono, while maybe theoretically better, may offer no audible benefit.
Leon...
I have no argument with what you are hearing. And I have no intent on building a dual mono B-II, as this approach is just philosophically not the way I like to do things.
On the Tridents though, I have a set of these for my new Legato II/B-II build, and I certainly expect them to offer a sonic improvement-no reason why a lower noise/impedance power supply, and the elimination of (due to no need) of electrolytic caps should not result in sonic benefits. In fact, my experience is that power supply changes often offer the biggest improvements in performance for any audio components.
I have no argument with what you are hearing. And I have no intent on building a dual mono B-II, as this approach is just philosophically not the way I like to do things.
On the Tridents though, I have a set of these for my new Legato II/B-II build, and I certainly expect them to offer a sonic improvement-no reason why a lower noise/impedance power supply, and the elimination of (due to no need) of electrolytic caps should not result in sonic benefits. In fact, my experience is that power supply changes often offer the biggest improvements in performance for any audio components.
I won't go any further than this.
Numbers don't always show a difference. Sometimes they even show a difference that seems contradictory to your subjective experience. In both cases it could be that the question or assumption behind the numbers is what is off, not the numbers themselves. One thing I hope we can agree upon is if the numbers show a difference there is a difference. How audible and then how meaningful that difference are questions everyone must answer for themselves.
If you haven't heard you don't know, so try it and listen. If you don't hear a difference, or don't value the difference you hear, great.
For me audio is a little like swimming in a mountain lake. You have to decide you are going to do it and then jump in with both feet. You can't swim without getting wet. If you just dip your toes in, it is far to easy to talk yourself out of a possibly very rewarding experience.
Numbers don't always show a difference. Sometimes they even show a difference that seems contradictory to your subjective experience. In both cases it could be that the question or assumption behind the numbers is what is off, not the numbers themselves. One thing I hope we can agree upon is if the numbers show a difference there is a difference. How audible and then how meaningful that difference are questions everyone must answer for themselves.
If you haven't heard you don't know, so try it and listen. If you don't hear a difference, or don't value the difference you hear, great.
For me audio is a little like swimming in a mountain lake. You have to decide you are going to do it and then jump in with both feet. You can't swim without getting wet. If you just dip your toes in, it is far to easy to talk yourself out of a possibly very rewarding experience.
Stephen...
I never suggested there was not an audible difference-I would never make a presumption like that with something I have not heard.
What I pointed out is that the difference shown in the above measurements would never be audible in a real system-this does not not mean there is no audible difference-it just means that any audible difference is not the result of the distortion/noise reduction. There is always the possibility of things we do not know how to measure affecting the sonic performance though.
I run balanced only, so no need to go dual mono to get good CMR. I suspect if there is an advantage to dual mono, it might show up in stereo separation measurements versus frequency, as I could see, perhaps, a single 9018 DAC chip suffering from some capacitive coupling.
Philosophically, what bothers me about running two complete DACs is that this approach doubles the production of RF onboard, so there are going to be some tradeoffs in going this route.
I never suggested there was not an audible difference-I would never make a presumption like that with something I have not heard.
What I pointed out is that the difference shown in the above measurements would never be audible in a real system-this does not not mean there is no audible difference-it just means that any audible difference is not the result of the distortion/noise reduction. There is always the possibility of things we do not know how to measure affecting the sonic performance though.
I run balanced only, so no need to go dual mono to get good CMR. I suspect if there is an advantage to dual mono, it might show up in stereo separation measurements versus frequency, as I could see, perhaps, a single 9018 DAC chip suffering from some capacitive coupling.
Philosophically, what bothers me about running two complete DACs is that this approach doubles the production of RF onboard, so there are going to be some tradeoffs in going this route.
There seems to be a fundamental misunderstanding about the way the ES9018 handles SPDIF.
Where SPDIF has its issues is when SPDIF receivers have to obtain their clock from the SPDIF signal. The ES9018 does not do this at all. The SPDIF data is input completely asynchronously, it is not merely re-clocked, it is never generating a clock in the first place.
The truth is both I2S and SPDIF input measure exactly the same. That is because in the end they are treated exactly same. The only difference is what the DAC does to load up the data queue.
Because there is no clock recovery involved, there is really no penalty to using SPDIF except that you are limited to 24bit input where I2S can do 32bit. Just like anything else though, if the signal input is junk, the output will be too. So you still have to have good source.
Where SPDIF has its issues is when SPDIF receivers have to obtain their clock from the SPDIF signal. The ES9018 does not do this at all. The SPDIF data is input completely asynchronously, it is not merely re-clocked, it is never generating a clock in the first place.
The truth is both I2S and SPDIF input measure exactly the same. That is because in the end they are treated exactly same. The only difference is what the DAC does to load up the data queue.
Because there is no clock recovery involved, there is really no penalty to using SPDIF except that you are limited to 24bit input where I2S can do 32bit. Just like anything else though, if the signal input is junk, the output will be too. So you still have to have good source.
Thinking...
Speculatively: this will likely depend. Are you questioning if it would be better to use a separate SPDIF receiver and then feed the B-II I2S from that? The SPDIF receiver onboard the ESS 9018 is a pretty sophisticated and unique design (at least from what I can read in the public domain documentation)-I would not be surprised to find that the ESS 9018 SPDIF receiver actually outperforms other options (like the standard Crystal receiver).
Now if the I2S is generated not from SPDIF, but from, say, Russ' forthcoming asynchronous USB receiver, that will be another matter. The async USB to I2S should have an advantage, because the data is never subjected to the clock embedded format of SPDIF, and the necessary clock recovery circuitry (which adds jitter). Additionally, these points may be somewaht moot, due to the ESS' onboard "jitter rejection" via the ASRC, but in reality I doubt it-My B-II certainly sounds different when fed from different sources, so, as usual, less jitter in results in better sound out. Interestingly, one B-II owner has reported on the TPA forums, that the B-II does sound better when fed I2S from a Hiface Evo asynchronous USB interface versus SPDIF from the same source.
Could these differences be measured? Sure, but proper measurements of jitter are very complex, and require very specialized gear. Very few people have the gear to really measure jitter differences.
Speculatively: this will likely depend. Are you questioning if it would be better to use a separate SPDIF receiver and then feed the B-II I2S from that? The SPDIF receiver onboard the ESS 9018 is a pretty sophisticated and unique design (at least from what I can read in the public domain documentation)-I would not be surprised to find that the ESS 9018 SPDIF receiver actually outperforms other options (like the standard Crystal receiver).
Now if the I2S is generated not from SPDIF, but from, say, Russ' forthcoming asynchronous USB receiver, that will be another matter. The async USB to I2S should have an advantage, because the data is never subjected to the clock embedded format of SPDIF, and the necessary clock recovery circuitry (which adds jitter). Additionally, these points may be somewaht moot, due to the ESS' onboard "jitter rejection" via the ASRC, but in reality I doubt it-My B-II certainly sounds different when fed from different sources, so, as usual, less jitter in results in better sound out. Interestingly, one B-II owner has reported on the TPA forums, that the B-II does sound better when fed I2S from a Hiface Evo asynchronous USB interface versus SPDIF from the same source.
Could these differences be measured? Sure, but proper measurements of jitter are very complex, and require very specialized gear. Very few people have the gear to really measure jitter differences.
Last edited:
There seems to be a fundamental misunderstanding about the way the ES9018 handles SPDIF.
Where SPDIF has its issues is when SPDIF receivers have to obtain their clock from the SPDIF signal. The ES9018 does not do this at all. The SPDIF data is input completely asynchronously, it is not merely re-clocked, it is never generating a clock in the first place.
The truth is both I2S and SPDIF input measure exactly the same. That is because in the end they are treated exactly same. The only difference is what the DAC does to load up the data queue.
Because there is no clock recovery involved, there is really no penalty to using SPDIF except that you are limited to 24bit input where I2S can do 32bit. Just like anything else though, if the signal input is junk, the output will be too. So you still have to have good source.
The S/PDIF still has a clock even if they strip it off. And the data keeps flowing so they need to somehow delay this and reclock it into the DAC. The only way would seem to be with a PLL or whatever they want to call it now. Since they can't stop or change data flow. S/PDIF doesn't work that way.
Thus, to me sounds like ASRC in a roundabout way. Albeit with a different approach.
It has the ability to generate a clock yes, but it is *only* that generated clock that is ill and in the case of the ESS DAC it is not used at all. In any way.
SPDIF does not even really "have a clock", but you can derive a clock from it. That is the heart of the issue where standard receivers are used.
No clock extraction (which usually requires some type of PLL) is done at all in the case of ESS method.
The data is just data which is queued, and gets stuffed into a buffer. Much the same way that asynchronous USB works.
SPDIF once you are not running through a PLL to obtain a clock is just like any other 1 wire protocol.
SPDIF does not even really "have a clock", but you can derive a clock from it. That is the heart of the issue where standard receivers are used.
No clock extraction (which usually requires some type of PLL) is done at all in the case of ESS method.
The data is just data which is queued, and gets stuffed into a buffer. Much the same way that asynchronous USB works.
SPDIF once you are not running through a PLL to obtain a clock is just like any other 1 wire protocol.
Last edited:
I just rigged my Musiland to output I2S last night. So far, both SPDIF and I2S sounds the same.
In my crude setup, so far spdif seems to be a "more solid" signal compared to I2S. With SPDIF I can use lowest dpll bandwidth setting in BuffaloII and with I2S I get drop-outs unless I increase the bandwidth of the DPLL. Russ suspects is noise in the I2S lines. The wires I am using are only 3 inches long.
In my crude setup, so far spdif seems to be a "more solid" signal compared to I2S. With SPDIF I can use lowest dpll bandwidth setting in BuffaloII and with I2S I get drop-outs unless I increase the bandwidth of the DPLL. Russ suspects is noise in the I2S lines. The wires I am using are only 3 inches long.
Well I can certainly relate to what you state there. I've doubted for about 2 years about building the (then Buffalo) DAC. Going dual mono is also something I had great doubts about, not for philosophical reasons but because it requires one to simply buy about twice the hardware for a tiny bit of extra performance. And I only started building the DAC as I wanted a DIY companion to my balanced power amp, with good volume control. It has become a slightly bigger project than anticipated...
For me however the end result is all that matters. And I'm more than satisfied with the result, however it's achieved. The modular nature of the new DAC board allows for an almost infinite freedom to mix and match as one sees fit. And I would certainly recommend going for the Tridents well before going dual mono. It's simply a much more afffordable improvement, and one I'd like to see as an integral part of the BIII board if one ever is going to be designed.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.