[dip16amp]

Quote:

Originally Posted by thomaspf Benchmark is just using a AD1896 chip so you can read up everything about the performance of this resampler in the data sheet. To my knowledge the jitter removal with this chip works really well if you implement a stable local clock which the DAC1 has. However my point was not about jitter. In the process of removing jitter these chips actually modify the data that is being send to them. The more clock skew and jitter the more modifications. The algorithm turns jitter into digital broadband noise. Basically every time you play a song a different set of samples is actually being converted by the actual DAC chips. I am not making any statements whether you can hear any of that I am just explaining how I understand the digital section of these DACs work. Cheers Thomas

I don't think a data error caused by removing jitter would be any different from a jitter error changing good data. Either way, an error would show up on an Audio Precision jitter measurement test, but their tests show no increased output noise.

 

[Mikenet]

I posted another thread here asking how the AD1896 actually removes jitter. I didn't want to take this thread too far off topic, and the post got rather long. thomaspf and others, could you take a look at it?

 

[thomaspf]

Quote:

I don't think a data error caused by removing jitter would be any different from a jitter error changing good data.

Good point, however there is a difference. The changes in the analog waveforms caused by jitter (these are not errors in the bit stream but occur with the same bits) are frequency modulations that can reach pretty high levels. The broadband noise by the resampler is smeared out over the whole band and contributes very littel to the overall spectrum.

That is why this works so well. Looking at the data sheet of the ad1896 rather than the marketing material on the benchmark site clarifies this quite well. There are a couple of other DACS using this chip and they seem to make less noise about the uniqeness of theor dejitter circuit.

The other factor to watch for is clock skew. Even with a very low jitter source you will get a lot of recalculation simply by having the crystal of the source and the crystal driving the resampler on the the DAC side beind different. Getting good results with one specific source does not means this applies to all sources.

Cheers

Thomas

 

[dip16amp]

Quote:

Originally Posted by thomaspf Good point, however there is a difference. The changes in the analog waveforms caused by jitter (these are not errors in the bit stream but occur with the same bits) are frequency modulations that can reach pretty high levels. The broadband noise by the resampler is smeared out over the whole band and contributes very littel to the overall spectrum. That is why this works so well. Looking at the data sheet of the ad1896 rather than the marketing material on the benchmark site clarifies this quite well. There are a couple of other DACS using this chip and they seem to make less noise about the uniqeness of theor dejitter circuit. The other factor to watch for is clock skew. Even with a very low jitter source you will get a lot of recalculation simply by having the crystal of the source and the crystal driving the resampler on the the DAC side beind different. Getting good results with one specific source does not means this applies to all sources. Cheers Thomas

Looking at a couple of other claims show some differences. I don't expect other companies to trademark their jitter features but showing performance is important.

The Apogee Mini-Dac claims it's dual-stage clock ensures almost total immunity to jitter at the input stage. No jitter specs or jitter tests are provided. This seems to indicate that it is not quite as good and may have slow lock problems of two-stage PLL circuits.

The Bel Canto DAC2 does claim to eliminate jitter but does not provide jitter specs or tests. It would be nice to see actual performance results compared to Benchmark DAC1.

Since this thread is looking for a digital out to the DAC1, a bit perfect output would seem to be the main concern. Trying to reduce the jitter output of them or using coax instead of toslink for lower jitter into the DAC1 would not matter.
A jitter immune DAC will not have a problem with jitter.

 

[shplorgh]

Quote:

Originally Posted by dip16amp A jitter immune DAC will not have a problem with jitter.

There are no "jitter immune" DACs. It's not jitter immune just because the company that makes it says so.

 

[dip16amp]

Quote:

Originally Posted by hungrych There are no "jitter immune" DACs. It's not jitter immune just because the company that makes it says so.

Thats why I rely on their DAC1 jitter performance tests with the Audio Precision tester and it's results. I would do my own tests if I had the proper equipment but I have no reason to think that their results are fake or incorrect. Other companies that do make claims to be jitter immune and do not provide test results may not true.
If you download the DAC1 manual and look at their tests and test results, that does more than just "says so". If you don't believe what they provide, can someone show how much jitter it takes for jitter artifacts appear?

 

[Mikenet]

Page 20 of the AD1896 datasheet shows the jitter filtering curves of its internal PLL. Any signal it would pass would end up as variations in the ASRC source rate, and the ASRC would embed the jitter artifacts into the output data.

I posted about torture testing the DAC1 last night.

 

[thomaspf]

So what exactly does the audio precision actually show and how is that related to the noise spectrum that these resamplers create with real music?

To see the effect of the resampling and the increase in broadband noise for the best possible case just have a look at the data sheet. Since the noise is very low the analog noise in the Dac1 masks the increase but these graphs are for a single frequency. I would not expect to have this picture look much different with a Mini-Dac.

Don't get me wrong, the AS1896 chip seems to work really well but I would be more careful with taking measurement made for marketing purposes with real world performance. The BelCanto DAC2 has exactly the same chip btw. so does the Headroom MAX-Dac.


Cheers

Thomas

 

[dip16amp]

Quote:

Originally Posted by Mikenet Page 20 of the AD1896 datasheet shows the jitter filtering curves of its internal PLL. Any signal it would pass would end up as variations in the ASRC source rate, and the ASRC would embed the jitter artifacts into the output data. I posted about torture testing the DAC1 last night.

Not sure if the AD1896 datasheet jitter performance is the same as what the DAC1 does. If the DAC1 performance is just from the standard use of the AD1896, then the Bel Canto, Apogee, and any other DAC using it should have the same performance that the DAC1 measures. We don't have other DAC jitter test results to compare to but if they are the same, then they also would be jitter immune, at least up to the maximum jitter amount that the Audio Precision tester can input.

The digital inputs to the DAC1 are PCM in S/PDIF or AES format. Your test proposal mentions a signal generator and clock but does not have detail on getting that into the DAC1 that it will understand. Controlling and measuring the amount of jitter being input to the DAC is not clear either. Measuring the analog output for jitter artifacts would have to be as good or better than the Audio Precision tester to show them.

I don't think jitter is a problem with the DAC1 and probably isn't an issue with other DACs that use the AD1896 but it would be nice to compare performance tests on them all. I don't think there is a cheap, easy, and accurate way to do jitter performance tests. Are there ways that will stand up to professional testers?

 

[dip16amp]

Quote:

Originally Posted by thomaspf So what exactly does the audio precision actually show and how is that related to the noise spectrum that these resamplers create with real music? To see the effect of the resampling and the increase in broadband noise for the best possible case just have a look at the data sheet. Since the noise is very low the analog noise in the Dac1 masks the increase but these graphs are for a single frequency. I would not expect to have this picture look much different with a Mini-Dac. Don't get me wrong, the AS1896 chip seems to work really well but I would be more careful with taking measurement made for marketing purposes with real world performance. The BelCanto DAC2 has exactly the same chip btw. so does the Headroom MAX-Dac. Cheers Thomas

http://www.benchmarkmedia.com/appnot...ultralock.html

It would be nice to see how other DACs perform in these tests.

 

[thomaspf]

I guess that is what I am trying to tell you. There is no such thing as a proprietary Ultralock technology!


This is just a chip which is also being used in a bunch of other DACs as well. These tests are not very relevant to begin with and the fact that one manufacturer has managed to convince you it is important gives credit to their marketing department but not necessarily to their enigneering.

Please observe that they compare the curves to some unspecified DAC not to any of the designs with a secondary PLL like the ones mentioned in this thread.

We are all in violent agreement that low jitter is important, the question is whether these graphs are anything special.

Cheers

Thomas

 

[dip16amp]

Quote:

Originally Posted by thomaspf I guess that is what I am trying to tell you. There is no such thing as a proprietary Ultralock technology! This is just a chip which is also being used in a bunch of other DACs as well. These tests are not very relevant to begin with and the fact that one manufacturer has managed to convince you it is important gives credit to their marketing department but not necessarily to their enigneering. Please observe that they compare the curves to some unspecified DAC not to any of the designs with a secondary PLL like the ones mentioned in this thread. We are all in violent agreement that low jitter is important, the question is whether these graphs are anything special. Cheers Thomas

It doesn't really matter what they call it. If the DACs that use the AD1896 are immune from jitter, then there is no need to worry about feeding it lower jitter. That is what I am saying. It doesn't matter if toslink or coax feeds them lower jitter.

 

[thomaspf]

Well immune is a bit strong and the question remains as to what type of jitter.

Then there is the issue of clock skew. A DAC with a synchronous PLL follows the clock of the sender and basically adjusts the pitch of the playback to the frequency of the crystal in the sender.

A DAC with an asynchronous sample rate converter will start changing the samples even if there is no jitter but the inevitable difference in the frequency of the sender vs. the receiver. Effectively you tune to the pitch of the receiver by recomputing new samples. In fact, you are likely to end up with different samples each time you play a track. How much of this is audible? Who kows.

The DAC1 is definitely a great choice if you like the sound. I only have listened to the old model which I did not like but some people on the amp section have reported that Benchmark made some changes on to improve the sound.

Cheers

Thomas

 

[dip16amp]

Quote:

Originally Posted by thomaspf Well immune is a bit strong and the question remains as to what type of jitter. Then there is the issue of clock skew. A DAC with a synchronous PLL follows the clock of the sender and basically adjusts the pitch of the playback to the frequency of the crystal in the sender. A DAC with an asynchronous sample rate converter will start changing the samples even if there is no jitter but the inevitable difference in the frequency of the sender vs. the receiver. Effectively you tune to the pitch of the receiver by recomputing new samples. In fact, you are likely to end up with different samples each time you play a track. How much of this is audible? Who kows. The DAC1 is definitely a great choice if you like the sound. I only have listened to the old model which I did not like but some people on the amp section have reported that Benchmark made some changes on to improve the sound. Cheers Thomas

Clock skew is not an issue. From the DAC1 page:

The DAC1 totally isolated the conversion clock from the digital interface clock.
Jitter on a digital audio input can never have any effect on the DAC1 conversion clock.

The upsampling conversion clock is never phase-locked to the reference clock. Instead the converter oversampling-ratio is varied with extremely high precision to achieve phase-lock to the reference clock. Jitter cannot effect the audio conversion, and test bench performance is repeatable in any installation!

 

[thomaspf]

Sorry, you need to read the data sheet of the AD1896 not the Benchmark marketing material. There is a section on theory of operation.

Clock skew is the difference in the frequency between the input and the output side of the sample rate converter. It has nothing to do with jitter.

This has been a great discussion but I think we have reached the end of the useful part.

Have fun and enjoy this hobby

Thomas