Success....
Here's something I did not expect : I have reliable lock with good sound with a 22.5792Mhz clock and 96khz 24 bit files.
Foobar is playing 96/24 wav files with cm6631a asio driver set to 24bits, feeding the wm8805 isolated spdif, and the Wolfson is set to output 24 bit.
Only one "weak" link in the chain is the quad differential line driver - DS26C32AM. Anyone know an upgrade part off the top of their heads ? I can't believe 19nS prop'n delay can't be improved....
cheers
Here's something I did not expect : I have reliable lock with good sound with a 22.5792Mhz clock and 96khz 24 bit files.
Foobar is playing 96/24 wav files with cm6631a asio driver set to 24bits, feeding the wm8805 isolated spdif, and the Wolfson is set to output 24 bit. Only one "weak" link in the chain is the quad differential line driver - DS26C32AM. Anyone know an upgrade part off the top of their heads ? I can't believe 19nS prop'n delay can't be improved....
cheers
That's a strange question ! What's available ? The prop'n delay indicates this is not a fast chip so, as I understand it, there'll be longer rise/fall times and more skew, and very likely more noise. Plus, it's an obsolete part. So I'm not looking a four-seat car, or a five seater. I'm looking for the right-sized car, and not an obsolete one please. Oh, and in a nice colour. ;-)
But hey, let's say 6-8ns. Thanks !
Then I need to work out the best circuit for the spdif + Newava pulse trans. Since the spdif is feeding the quad diff driver, I think I don't really need to attenuate it much at all. I'm using something like this but i'm not happy it's working that well
The prop'n delay indicates this is not a fast chip so, as I understand it, there'll be longer rise/fall times and more skew, and very likely more noise. Plus, it's an obsolete part. So I'm not looking a four-seat car, or a five seater. I'm looking for the right-sized car, and not an obsolete one please. Oh, and in a nice colour. ;-) But hey, let's say 6-8ns. Thanks !
Then I need to work out the best circuit for the spdif + Newava pulse trans. Since the spdif is feeding the quad diff driver, I think I don't really need to attenuate it much at all. I'm using something like this but i'm not happy it's working that well
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
So you're using the 'line driver' as a receiver? (checks Google...) Ah, now I see that part is actually a quad line receiver not a driver.
There does seem to rather a common myth on DIYA that faster is always better. I don't swallow this - there's an appropriate speed for each application, for S/PDIF 20nS sounds about right. If you go faster you'll generate more PSU noise from the receiver as it switches. You could though look at LVDS receivers (like DS90LV032 for example) - they typically run much faster than the one you mentioned which is really designed for RS422.
There does seem to rather a common myth on DIYA that faster is always better. I don't swallow this - there's an appropriate speed for each application, for S/PDIF 20nS sounds about right. If you go faster you'll generate more PSU noise from the receiver as it switches. You could though look at LVDS receivers (like DS90LV032 for example) - they typically run much faster than the one you mentioned which is really designed for RS422.
GLT - I like your extra ground return bridges. They say imitation is the sincerest form of flattery... I will be copying your idea !
http://hifiduino.wordpress.com/2013/01/14/building-the-99-es9018-dac-board/
http://hifiduino.wordpress.com/2013/01/14/building-the-99-es9018-dac-board/
By all means. I had a piece of Nordost flat cable...
Been looking at the SAW clocks. The closest to 100 MHz carried by the local distributors are 106 MHz and 75 MHz. At < $4 each, definitely worth a try.
According to Epson, these SAW clocks perform better than 3rd overtone clocks (like the CCHD-950) because they operate on fundamental frequency.
But they don't measure as good as the Crysteks
Been looking at the SAW clocks. The closest to 100 MHz carried by the local distributors are 106 MHz and 75 MHz. At < $4 each, definitely worth a try.
According to Epson, these SAW clocks perform better than 3rd overtone clocks (like the CCHD-950) because they operate on fundamental frequency.
But they don't measure as good as the Crysteks
Cool - I'll be very interested to hear what you think, but I regret I'm not going to have time to try them myself for some time yet. The new semester starts next week so I'm trying to get this DAC mostly done before I have to return to teaching, and I've also got a dissertation to work on so ... time to hang up the irons for 14 weeks.
BTW, your Arduino pages make that look really easy - is it ? What do people usually struggle with ?
Did you decide on op amps yet ? I've settled on LME49990 for the buffers but still trying op amps for the I/V.
I've put links to it in the first post - thank you for your blog. You should be mightily pleased with that - a quality blog.
Cheers
Last edited:
do remember there is a new CCHD-575 from crystek in 100MHz which is a good performer and somewhat cheaper than the cchd-957
epsons claims of better performance from fundamental are meaningless. jitter can be measured reliably and they perform worse in the only ways that matter. at that price not much is lost of course, but honestly, you will measure the success by listening to it?
epsons claims of better performance from fundamental are meaningless. jitter can be measured reliably and they perform worse in the only ways that matter. at that price not much is lost of course, but honestly, you will measure the success by listening to it?
Last edited:
The journey IS the destination...
For 3.3V clocks, only the 75MHz and the 106 MHz are available.
The only reason I bring these up is because over at the other thread, side by side listening comparison, listeners preferred the BII with the saw clock to the one with the cchd clock.
I think the hardest part is familiarizing with the Arduino environment (which is actually pretty easy). I don't get many questions. Most implementations "just show up" in some forum or blog
I was thinking of first using the DAC in voltage mode single ended by flipping the phase of the second set of DACs and summing it with the first set (thus having 16 DACs per channel in SE mode). And then also try the passive filter from abraxalito.
Have a set of OPA627 somewhere in the closet that I will try.
umm I have a result for CMOS 3v3 SAW 100MHz right in front of me, the one I linked you to. here they have 878 of them at 3.72 each. maybe yopu mean a different model?
but like I said go for your life its cheap.
I can see it (the 100MHz or 98.304MHz) as a viable alternative if you dont want to blow the money on a crystek or better, I really can, they seem reasonable for the money. We are talking about a clock here, if it had other harmonics, they would be part of the jtest, so unless people prefer more error like they can prefer distortion, then it hurts my head to think of why you could prefer something worse.....and it is worse.
it seems to measure about the same, or a bit worse than the generic place-holder clock in the fifo kit
sorry I cant say I place much stead in this audible comparison, a controlled test I suppose? yet another magical unknown specification that science forgot; yet brings audible levels of improvement when it comes to music listeners. Dark Jitter....people have been saying they prefer clocks that are clearly borderline browning out their dacs as well.
but like I said go for your life its cheap.
I can see it (the 100MHz or 98.304MHz) as a viable alternative if you dont want to blow the money on a crystek or better, I really can, they seem reasonable for the money. We are talking about a clock here, if it had other harmonics, they would be part of the jtest, so unless people prefer more error like they can prefer distortion, then it hurts my head to think of why you could prefer something worse.....and it is worse.
it seems to measure about the same, or a bit worse than the generic place-holder clock in the fifo kit
Last edited:
They're making the same noises over on the Oppo 105 thread too - SAW subjectively trumps the more expensive Crystek, despite measuring worse. I'm curious to understand why
yes, the same few (2?) people are making the same noises, one of whom uses it in his clock module upgrade....
- Status
- Not open for further replies.