










Please try the options below:
- Try to access the feature directly from the eBay Home Page, instead of using a bookmark.
- Wait a few minutes and try to access the feature again.
- Check our Announcement Board to see if the feature is currently unavailable.

About eBay | Announcements | Buy Hub | Security Center | Resolution Center | Buyer Tools | Policies | Stores | Site Map | Help | eBay official time
I'm not the only one either not answering email????
Buy It Now
Hello XXXXXXXX (Not you? Sign in)
Error
Seller not accepting your bid
<li type="square"> Unfortunately, the seller cannot accept your bid on this item. This can happen for a couple of reasons. For example, the seller may be limiting how many items you can buy from him or her in a specific time period. Learn more.



















So when a 192Khz stream is passed to the the upsampler it gets into troubles, ans I wonder why? I thought that the upsampler should have operated as bypass when a 192Khz stream was played, but instead something wrong is happening.
Having just started playing with the CS8421 myself, perhaps I can help. Look at page 22 of the datasheet, para 4.3.7 - here it explains how fast a master clock is needed for various input/output sample rates. I am assuming your board is working in master mode on the output, slave on the input.
I have no clear idea what the schematic is of the 8421 board that you're using but my best guess is its not using a fast enough master clock to work correctly at 192kHz incoming rate. The equations they give there mean the master clock must be at least 24.96MHz to work correctly - if they are using 24.576MHz (a very common choice, its 512 * 48k) then 192kHz incoming won't work.
Your solution is simple - fit a 25MHz oscillator - of course the output will be running slightly faster than 192kHz but that's normally not a problem.
Hi Abraxalito,
That makes since if the clock isn't fast enought it can't see the signal can it?
Now to find a good clock circuit!!!
Thanks
That makes since if the clock isn't fast enought it can't see the signal can it?
Now to find a good clock circuit!!!
Thanks
Hello
I think I understood the problem, even though I don't have an electronic backgorund (I am a poor mathematician). I guess abraxalito is right.
Tracing the upsampling board I could check that it is consistent with the schematic (see attached png file where I report only the upsampling part). So the MS_SEL pin is connected to a 1.96KOhm and this to ground, which according to the datasheet it means that Input is Slave and Output is master.
Now the quartz on my board is XTI=24.576MHz. Going back to the datasheet in section "6.1 Clocking" it says
"If the output is set to master, Fso<= XTI/128 and Fsi<=XTI/130."
Then if I am feeding a 192KHz=0.192MHz I have
Fso=0.192MHz while XTI/128=0.1934063
Fsi=0.192Mhz while XTI/130=0.1904308
so it seems that the problem is that my quartz doesn't accomplish the second condition. To get the second conditions I need XTI=130*192KHz = 24.96MHz.
Now since I am not an expert I wonder why these guys did the board this way.
Cheers
Pietro
I think I understood the problem, even though I don't have an electronic backgorund (I am a poor mathematician). I guess abraxalito is right.
Tracing the upsampling board I could check that it is consistent with the schematic (see attached png file where I report only the upsampling part). So the MS_SEL pin is connected to a 1.96KOhm and this to ground, which according to the datasheet it means that Input is Slave and Output is master.
Now the quartz on my board is XTI=24.576MHz. Going back to the datasheet in section "6.1 Clocking" it says
"If the output is set to master, Fso<= XTI/128 and Fsi<=XTI/130."
Then if I am feeding a 192KHz=0.192MHz I have
Fso=0.192MHz while XTI/128=0.1934063
Fsi=0.192Mhz while XTI/130=0.1904308
so it seems that the problem is that my quartz doesn't accomplish the second condition. To get the second conditions I need XTI=130*192KHz = 24.96MHz.
Now since I am not an expert I wonder why these guys did the board this way.
Cheers
Pietro
Attachments
Hi Abraxalito,
That makes since if the clock isn't fast enought it can't see the signal can it?
Now to find a good clock circuit!!!
Thanks
That's a big question. Expecially in a resampler clocking is the key ingredient. I heard that these
Audiocom International
and these
NewClassD Audio Clocks
are high performance clocks, a friend of mine installed the first one into his Marantz and had impressive results.
But here we have an upsampler and the CS datasheet says that the clock must be as near as possible to the chip.
Anyway, the real question is if the upsampler improves the sound for this board. In my high resolution system I can definitively say YES. For low resolution files (16/44.1) the improvement is huge especially in micro-dynamic. I guess this is because the upsampler being near the dac has the role to reduce jitter and it does it very well. For high resolution files, ie 24/96KHz I guess that the upsampler still does something better but the difference is not that big.
I want the upsampler in and, well let's hope that somebody with experience will help us to fix the problem.
Best Wishes
Pietro
....
Your solution is simple - fit a 25MHz oscillator - of course the output will be running slightly faster than 192kHz but that's normally not a problem
I though the quartz should have a frequency which is a multiple of the output frequency, 25MHz/192Khz=130.208333. Is ti safe to mount a 48MHz crystal? I have two of them that I can rip from an unused radio board that could fit.
Cheers
Pietro
Last edited by a moderator:
Yes, that's why I said the output frequency would change slightly - in fact to 195.3kHz. But is this a problem? I'm running my CS8421 at 'strange' output frequencies without anything going wrong. Sounds awesome.
Great info, nice to hear from someone with some digital chops.
For a direct replacement, TentLabs sells their excellent XO modules for around $35US that should drop right in. A dedicated supply for it would be an option.
For a direct replacement, TentLabs sells their excellent XO modules for around $35US that should drop right in. A dedicated supply for it would be an option.
[QUOTEFor a direct replacement, TentLabs sells their excellent XO modules for around $35US that should drop right in. A dedicated supply for it would be an optionQUOTE]
Are there any documentation that the Tentlabs XO modules are better.
Are there any documentation that the Tentlabs XO modules are better.
Is ti safe to mount a 48MHz crystal?
No. The master clock of the CS8421 tops out at 27MHz.
What would the result be if I mounted a 27Mhz instead?No. The master clock of the CS8421 tops out at 27MHz.
It would work fine I think - it depends on your output DAC's ability to be overclocked. I have a CS8421 driving into an AD1955, with a 27MHz master clock, this DAC is quite heavily overclocked (datasheet maximum is 20MHz) but still works and sounds great.
Yes, Guido has full documentation for them. The phase noise is by far the lowest of any realistically priced units. They put the units from the ebay shops to shame.
It would work fine I think - it depends on your output DAC's ability to be overclocked. I have a CS8421 driving into an AD1955, with a 27MHz master clock, this DAC is quite heavily overclocked (datasheet maximum is 20MHz) but still works and sounds great.
Have you a fully DIY dac or rather it's a kit? I have never seen the CS8421 paired with the well respected AD1955. I think upsampling with good clocking is the best strategy to achieve low jitter at low cost.... it would be very nice to get some project out of the forum, I see that CS8421 is sold on 8-DIP adapters...
Pietro
Its the one I'm showing pictures of on my blog - I bought it on Taobao (Chinese eBay). The CS8421 is supplied separately as a 10pin DIL plug-in module. Certainly the CS8421 option for this DAC is low cost - around 60rmb which translates to just under $10.
Hi, I just purchased this DAC and went for the OPA627 opamps. I have not seen OPA627s labelled in this way before - are they in fact genuine? 2 X Dual to Mono Op amp AD827 LM358 NE5532 to OPA627 - eBay (item 120477065449 end time Sep-30-10 01:00:01 PDT)
Hi, I just purchased this DAC and went for the OPA627 opamps. I have not seen OPA627s labelled in this way before - are they in fact genuine? 2 X Dual to Mono Op amp AD827 LM358 NE5532 to OPA627 - eBay (item 120477065449 end time Sep-30-10 01:00:01 PDT)
If you mean the extra op-amp names they are there to get extra hits on searches.
Giga's stuff should be genuine, but it's always a gamble with china bought items. It will take a day or two before they settle properly in the DAC. By rights I'm sure you should be able to return them if you think they sound crap?
To be honest there might be better and cheaper op-amps to try (depending on your system / tastes) One of the mods is to remove the outside buffer op-amp entirely and tap the signal directly from the outputs of the first (filter) op-amp. Try a pair of LT1028's in the first op-amp position on an adapter - better than OPA627.
Lot of fun to be had with this DAC.
Tryed to find info on the NTK xtal osc. on internet but no succes.But anyway I have traded me a Tentlabs 24.576 I will use instead.Are there any documentation that the Tentlabs XO modules are better.
- Home
- Source & Line
- Digital Line Level
- Experience with this DIY DAC ?