Share some phase noise plots of the BCLK taken from the DAC chip input from another popular DAC, then we can compare with my Purestream. Also I'd like to see propagation skew between master and BCLK at the DAC chip inputs as well.
These are specs that are normally missing from most DAC manufacturer spec sheets.
Yes I'd say. For some reason everyone is more interested in my DAC than the Hypex Fusion Plate amps.
I'm not going to open a new thread on the same topic. This thread can just get back onto the topic of plate amps. Or at least general discussion of plate amp related technology until we have some Fusion amps in our hands.
Logic functions are universal, there are no magic digital filters implemented with magic logic gates. Digital audio codec developers probably do 1st concept implementation with a FPGA or high power DSP. Then the design goes through a rational cost optimization process, and the resulting (universal) logic is implemented in hard-coded gate arrays in the CODEC chip. I think best approach in this field is to select fields/applications that have not been rationalized/universalized yet. Good full range amplifiers are 110~120dB SNR in practice, which is enough even in close listening of horn tweeters. Matching codec ICs are available too, with less THD than the amplifiers. More is overkill, like selling hand painted/decorated soundcards, mistaking entertainment/services sector by engineering sector.
I will be acquiring one of these machines shortly:
https://www.microsemi.com/products/timing-synchronization-systems/test-measurement/test-sets/5125a
I'm going to be measuring dozens of clocks by several well known manufacturers. For example I'll measure 20 of each the Crystek 957, 575 in 22/24/45/49Mhz. Same with NDK 2520SD and SDA. Fox Xpresso as low end example. Then also the Pulsar, along with some high end custom Rakon units.
Following that I'll take phase noise and propagation skew measurements from several well measuring DAC's right at the DAC chip clock inputs. Should be a great eye opener. Also will clearly demonstrate that you're playing Russian Roulette buying a DAC without knowing the exact phase noise of the particular unit you purchase. Because clock to clock consistency is so poor.
https://www.microsemi.com/products/timing-synchronization-systems/test-measurement/test-sets/5125a
I'm going to be measuring dozens of clocks by several well known manufacturers. For example I'll measure 20 of each the Crystek 957, 575 in 22/24/45/49Mhz. Same with NDK 2520SD and SDA. Fox Xpresso as low end example. Then also the Pulsar, along with some high end custom Rakon units.
Following that I'll take phase noise and propagation skew measurements from several well measuring DAC's right at the DAC chip clock inputs. Should be a great eye opener. Also will clearly demonstrate that you're playing Russian Roulette buying a DAC without knowing the exact phase noise of the particular unit you purchase. Because clock to clock consistency is so poor.
So you start by knowing the conclusions, and planning to do the measurements that lead to the conclusions?
Well I already heard the story from the fellow that measures and sorts each one of our clocks now. Time to do my own experiments, and publish them in a clearly laid out fashion. I'll be taking video of the process as well to show how the machine works, and people can see the actual phase noise of their favourite clock populate on the screen first hand. Although it takes minimum 2 hours for the clocks to warm and settle to get an accurate reading.
Along with the Microsemi 5125a, I'll be getting one of these:
U8903B Performance Audio Analyzer | Keysight (formerly Agilent?s Electronic Measurement)
I like it because it can accurately measure up to 1.5Mhz. For all of our prototyping so far we have been using the Picoscope 4262. It lets you know if you have any flaws, but not accurate enough to publish the results as far as I'm concerned:
https://www.picotech.com/oscilloscope/4262/picoscope-4262-overview
U8903B Performance Audio Analyzer | Keysight (formerly Agilent?s Electronic Measurement)
I like it because it can accurately measure up to 1.5Mhz. For all of our prototyping so far we have been using the Picoscope 4262. It lets you know if you have any flaws, but not accurate enough to publish the results as far as I'm concerned:
https://www.picotech.com/oscilloscope/4262/picoscope-4262-overview
So this must be the B&W subwoofer series with the Hypex MP series amps:
DB Subwoofers | Bowers & Wilkins
DB Subwoofers | Bowers & Wilkins
So jitter doesn't exist in the real world? Clock propagation errors don't exist? If people ever want to know the actual reason why some DAC's sound better than others, you need to look at the real reasons. It's not because of fairies and unicorns. The data DAC manufacturers commonly share doesn't correlate much to real world sound quality. If it did, Oppo's should be among the finest DAC's available.
You could easily modify a DAC to sound terrible, yet still have it pass all of the industry standard tests with flying colors.
I think TNT's point is what matters is what comes out of the DAC. Yes, jitter exists, but the important thing to measure is the effect of the jitter on the output.
What aspects would you modify to accomplish that? Excessive jitter shows up on noise and intermodulation measurements.
Even the APx-555 isn't good enough to measure the full effects of jitter from the analog outputs. The very best way to measure jitter is in the digital domain right at the cusp of conversion to analog. And in order to do that you need very expensive machines. Such as the Microsemi 5125.
You could easily mess up the SRC/SDM algorithms. You could skew the timing between BCLK and MCLK. All sorts of things. When I look at some layouts, it's a plain as day they made serious design faults. Yet some of these
DAC's still measure very well.
Last edited:
"Normal" levels of jitter are easily measured with a decent 24-bit sound card / interface using the J-test signal developed by Julian Denn back in the 90's.
If the jitter is so small that you need stuff like the microsemi analyzer to measure it, it is probably way too small to be audible.
Well there's many many DAC manufacturers in the industry, including me that know how audible jitter is. The J-test simply isn't good enough. But all depends on the level of quality you're after. If all you want to do is match an Oppo Sonica, yes the J test is probably fine. But if you want to surpass the MSB Select 2 with Femto 33 clock, good luck doing it with the J test alone.