I don't have any problem running my BIII at 192KHz with dual xo clock board at 45.xxx/49.xxx MHz XO, or with Si570 clock board. What is your MCLK frequency for now? For ESS DAC running by an external MCLK for sync mode, MCLK frequency has to be greater then 256*Fs at any time, that's the real limitation.
Regards,
Ian
... hmmm strange... in my case (FIFO+buffalo iiise_IVYIII) in synchro mode with
22.x ,24.x crystek`s i hear thin , cold (but spacious with very good stereo but flat dynamic) sound ... in asynchronous mode (100MHz BIIISE) sound is very impressive - rich and full. better than my previous system - dual mono Opus dac with FIFO .... i think 90MHz is minimum/optimum for Buffalo (Sabre) Dac`s...
Thanks Ian!
I have skim read it just now and noticed two things only:
- Page 5 diagram shows 'isolator board' but the page is dedicated to describing the 'no isolator' connection. I think that diagram should change to 'FIFO board' instead of 'isolator board'.
- Copyright statement at the end of your doc shows 2011, that's soooo two years ago man
Otherwise looks excellent as usual and I'll probably read in more detail tonight some time.
Cheers,
Chris
I have skim read it just now and noticed two things only:
- Page 5 diagram shows 'isolator board' but the page is dedicated to describing the 'no isolator' connection. I think that diagram should change to 'FIFO board' instead of 'isolator board'.
- Copyright statement at the end of your doc shows 2011, that's soooo two years ago man
Otherwise looks excellent as usual and I'll probably read in more detail tonight some time.
Cheers,
Chris
Thanks Ian!
I have skim read it just now and noticed two things only:
- Page 5 diagram shows 'isolator board' but the page is dedicated to describing the 'no isolator' connection. I think that diagram should change to 'FIFO board' instead of 'isolator board'.
- Copyright statement at the end of your doc shows 2011, that's soooo two years ago man
Otherwise looks excellent as usual and I'll probably read in more detail tonight some time.
Cheers,
Chris
Thanks Chris, good eye
. Corrected. Please wait my V1.0 for the wiki. I'm busy at production now. So far so good. Hopefully I can post GBIV details by this weekend.Regards,
Ian
oh haha so it is
I think its quite funny that soundcheck mentioned in another thread that we like many buyers of DIY products were paying beta testers. I argued this in your case, because
1. he doesnt have a clue what hes talking about as has no first hand experience of the fifo
2. I think I can remember a single intermittent solder joint that caused an intermittent error, the rest since the first fifo, have always turned out to be user error to my knowledge. if there have been other issues people have kept it super-secret
3. for a DIY module I dont think anyone could expect better after sales support and warranty (hell its better than the majority of companies)
soundcheck only pops his head in every now and again to illustrate his incomplete understanding of how the product works, he thus has no basis for opinion. whats he scared of do you think? I know he traditionally trades on his ability to vanquish the evil PC jitter demons, maybe thats it? its not the first time ive seen him drop a little hint that there is issues with fifo without knowing what hes talking about, hes done it here in the thread too. very odd.
I think its quite funny that soundcheck mentioned in another thread that we like many buyers of DIY products were paying beta testers. I argued this in your case, because
1. he doesnt have a clue what hes talking about as has no first hand experience of the fifo
2. I think I can remember a single intermittent solder joint that caused an intermittent error, the rest since the first fifo, have always turned out to be user error to my knowledge. if there have been other issues people have kept it super-secret
3. for a DIY module I dont think anyone could expect better after sales support and warranty (hell its better than the majority of companies)
soundcheck only pops his head in every now and again to illustrate his incomplete understanding of how the product works, he thus has no basis for opinion. whats he scared of do you think? I know he traditionally trades on his ability to vanquish the evil PC jitter demons, maybe thats it? its not the first time ive seen him drop a little hint that there is issues with fifo without knowing what hes talking about, hes done it here in the thread too. very odd.
Last edited:
I think you'll find that Ian's doc version is just reflecting the PCB version so that we know what doc goes with what pcb revision.
Ahhh yep, I get ya now.
I've tended to ignore that type of argument/discussion in the past, sort of let it fade into the background as white noise. The level of quality of a DIY implementation is, most of the time, down to the DIYer that builds/integrates it. If they don't have the skills to select high quality PCBs and circuit designs that will integrate well together, the DIYer ends up with a rubbish/half baked end product, where is the surprise in that?
Throwing a bunch of money and 'flavor of the day' forum projects at an aluminium enclosure is not really going to get a great result, ever. It might get something above average, by chance, it might turn out to be below average. It's all in the implementation at every level, at the implementation of the PCB design, at the selection of the combination of boards, at the arrangement and routing of the wiring inside the enclosure etc etc. Also, contrary to popular believe, buying pre-assembled boards does not alleviate the need for the DIYer to be consciously making design decisions to reach whatever target they set out trying to achieve.
EDIT: It is also worth noting that version numbers are notoriously missleading in terms of a level of polish on a product. Ian could easily have gone through revisions incrementing v0.1, v0.2 etc and released version v0.4 or v0.5. In internet browser wars, people thought that Chrome was lagging behind Firefox just because Chrome was at v1.5 and Firefox was up to 5 or 6. Then both Mozilla and Google started consciously incrementing the first digit and now we've got Chrome 24
Last edited:
Moving this discussion to the FIFO thread, seemed out of place in another vendor's thread regarding a totally different device.
From memory there were replies to that effect. The source should not matter, providing it is not producing radiated or conducted noise that are able to get into the DAC side of the circuit.
The remaining jitter is dominated by clock selection for most users. No one has had measurement equipment good enough to demonstrate this categorically. Ian's focus on his iterations are on improving the mclk and providing faster mclk to suit ESS high speed DACs better, because commonly available (and cost effective) crystals are not available in 95/98MHz for sync clocking.
There are some who have mentioned that subjectively they feel there might still be some differences, I have seen no reason to believe them yet. It is more likely that they have hamstrung one and there is increased radiated noise associated with one than the other, that is just a poor design of the source, Ian's design with isolator even allows for elimination of that problem in most cases.
If you followed the thread, or at least flicked to the important posts by Ian (which I've even collated into a wiki page to make that easier) you would see that Ian has measured his complete set of devices as bit perfect. He has then fed it signals into the FIFO+reclocker that have VERY significant jitter on his measurement equipment and in those conditions, at the output of the FIFO he has measured so little jitter that it is in the noise floor of the measurement.
I cannot understand what is wrong with continuing to optimise the ultimate. Continuing to improve the implementation and having the best solution architecture are not mutually exclusive.
From memory there were replies to that effect. The source should not matter, providing it is not producing radiated or conducted noise that are able to get into the DAC side of the circuit.
The remaining jitter is dominated by clock selection for most users. No one has had measurement equipment good enough to demonstrate this categorically. Ian's focus on his iterations are on improving the mclk and providing faster mclk to suit ESS high speed DACs better, because commonly available (and cost effective) crystals are not available in 95/98MHz for sync clocking.
There are some who have mentioned that subjectively they feel there might still be some differences, I have seen no reason to believe them yet. It is more likely that they have hamstrung one and there is increased radiated noise associated with one than the other, that is just a poor design of the source, Ian's design with isolator even allows for elimination of that problem in most cases.
If you followed the thread, or at least flicked to the important posts by Ian (which I've even collated into a wiki page to make that easier) you would see that Ian has measured his complete set of devices as bit perfect. He has then fed it signals into the FIFO+reclocker that have VERY significant jitter on his measurement equipment and in those conditions, at the output of the FIFO he has measured so little jitter that it is in the noise floor of the measurement.
I cannot understand what is wrong with continuing to optimise the ultimate. Continuing to improve the implementation and having the best solution architecture are not mutually exclusive.
I do not intend to highjack this thread. Though it seems that some of the followers take my remarks about respectively reference to this project personal and attack me over here on presumably rather save ground.
Please, don't put things out of context!!!
1.
Nobody is saying that products shouldn't get improved. The discussion over there was about buying a board to find out a month later there's a new and better one. People were not happy with that situation. I brought up some other well known DIY projects. Ian's as well as Twisted Pear's as two of many, where they'd face the same situation. I wanted them to stop the bashing against HifiMeDIY ( we're talking about a 100$ full digital amp board).
I wasn't happy by myself since I bought my latest stuff not long ago. I finally told them to be happy since those guys are listening and improving their stuff.
2.
You again bring up the IMO more then annyoing "should" not make any difference under conditions XYZ.
Since 2006 I'm optimizing transports since pretty much all audio interfaces I tried suck regarding jitter and/or noise and/or other distortions supression. I'm not the only one there are thousands out there. Even interfaces worth many thousands $ exhibit differences if something gets changed upstream. None of them I know is immune on incoming jitter and/or noise.
There are thousands of users using respectively have been using my stuff on my latest free and non-commercial project. Meanwhile there are tons of so called audiophile apps out there, which does the same thing what I've been doing since 2006. (The well kown OSX app Amarra has been sold at 1500$ in the begining mainly just to get the noise and/or jitter down and to get out bit-transparent data)
Meanwhile HW companies make a fortune in selling Sata filters, 400$ USB interface cards, fiber attached PCI-e cards, linear PC power supplies, USB filters, audiophile PCs, audiophile USB-SPDIF converters and so forth and so forth. People buying all that and PC after PC just to get the noise and jitter under control because most of the audio interface manufacturers fail to do so -- since years.
That's why I won't accept a "should-not-matter". I've got a pile of "should-not-matters" in the workshop. On the paper a plain Sabre claims "should-not-matter". People over here and elsewhere (including me) experienced what that means.
Either somebody who's is running Ian's reclocker has tested thouroughly the reclocker against all kind of transport sided optimsations resp. variations or not. Measurements alone won't suffice.
If Ian managed to get all this under control, for sure he can make a fortune with his stuff. If upstream improvements would still make a difference, even with the reclocker in place, it would just be another card... ...and he might be challenged to get it done even better on the next revision.
Cheers
Please, don't put things out of context!!!
1.
Nobody is saying that products shouldn't get improved. The discussion over there was about buying a board to find out a month later there's a new and better one. People were not happy with that situation. I brought up some other well known DIY projects. Ian's as well as Twisted Pear's as two of many, where they'd face the same situation. I wanted them to stop the bashing against HifiMeDIY ( we're talking about a 100$ full digital amp board).
I wasn't happy by myself since I bought my latest stuff not long ago. I finally told them to be happy since those guys are listening and improving their stuff.
2.
You again bring up the IMO more then annyoing "should" not make any difference under conditions XYZ.
Since 2006 I'm optimizing transports since pretty much all audio interfaces I tried suck regarding jitter and/or noise and/or other distortions supression. I'm not the only one there are thousands out there. Even interfaces worth many thousands $ exhibit differences if something gets changed upstream. None of them I know is immune on incoming jitter and/or noise.
There are thousands of users using respectively have been using my stuff on my latest free and non-commercial project. Meanwhile there are tons of so called audiophile apps out there, which does the same thing what I've been doing since 2006. (The well kown OSX app Amarra has been sold at 1500$ in the begining mainly just to get the noise and/or jitter down and to get out bit-transparent data)
Meanwhile HW companies make a fortune in selling Sata filters, 400$ USB interface cards, fiber attached PCI-e cards, linear PC power supplies, USB filters, audiophile PCs, audiophile USB-SPDIF converters and so forth and so forth. People buying all that and PC after PC just to get the noise and jitter under control because most of the audio interface manufacturers fail to do so -- since years.
That's why I won't accept a "should-not-matter". I've got a pile of "should-not-matters" in the workshop. On the paper a plain Sabre claims "should-not-matter". People over here and elsewhere (including me) experienced what that means.
Either somebody who's is running Ian's reclocker has tested thouroughly the reclocker against all kind of transport sided optimsations resp. variations or not. Measurements alone won't suffice.
If Ian managed to get all this under control, for sure he can make a fortune with his stuff. If upstream improvements would still make a difference, even with the reclocker in place, it would just be another card... ...and he might be challenged to get it done even better on the next revision.
Cheers
Last edited:
I did. they sound stiff and mechanical compared to the crystek 957. I prefer the crysteks.
Best Regards
Not much more expensive than the Crystek, EUR 24.37 for the Tentlabs XO versus EUR 19.78 for the CCHD-957.
Also the datasheet specs are very similar.
Keep in mind that any sound impressione is SUBJECTIVE, as every impressions posted in this thread and in any other thread also.
Simply from the technical specs you can expect similar results, they are not more than standard XO at reasonable price IMHO.
Now I'm testing the sound result of my XO using the Laptech crystal and I'll publish phase noise performance as soon as it will be measured, I won't tell nothing about "sound result" since it's totally subjective. BTW, the crystal costs CDN 36 + VAT + shipping, so, if you compare with this oscillator, the Crystek and the Tentlabs are cheap devices.
If you think technical specs are not suitable to make a choise, IMHO, I believe you have to get both to comparing directly in YOUR SYSTEM.