DIYHiFi.org

For the sake of audio
It is currently Fri Jul 25, 2014 8:03 am

All times are UTC




Post new topic Reply to topic  [ 17 posts ]  Go to page 1, 2  Next
Author Message
PostPosted: Fri Feb 27, 2009 7:12 pm 
Offline

Joined: Tue Nov 11, 2008 1:46 pm
Posts: 97
Location: Toronto
Okay pretty much a n00b here with a n00b type question.

The PCM2707, at least in the datasheet implementation is setup to use a 12Mhz crystal from which it derives it's clock requirements. Now reading the datasheet, it says that you can also use an external clock.

Given the PLL that is in the PCM2707, can you extract any better performance out of it using a better clock?

I'm tempted to use a better clock if for no other reason then "just cuz", but the only thing that I not clear on from the datasheet is this comment:
Quote:
Because no clock disabling pin is provided, it is not recommended to use the
external clock supply.

Are they try to say not to use an external clock because you can't disable it?

For background reference, this is for a simple DAC that I'm working on as a hobby. Signal path is USB->PCM2707->I2S->PCM1794. I won't litter this forum with the work of amateurs, but you can look here if you want to see it.


Top
 Profile  
 
PostPosted: Sat Feb 28, 2009 12:23 pm 
Offline
User avatar

Joined: Tue Feb 03, 2009 8:30 pm
Posts: 45
Hi erik,
I'm working on a project similar to your. The mine is PCM2707->SRC4192->2xPCM1794-IV converter. It maight be an idaa sharing experience.
Concerning the PCM2707 I read a lot about it. It seems to have 500ps of jitter that it's too much for a hi-end system. The man measured it used the data sheet schematic.
People say that the jitter is due to the USB asichronous transfer mode only, but the PCM2707 has 500ps againt the 2500ps of the oldest PCM2704, so I suppose that it could be inprove a bit using good componets, good power supply and layout.
I'm going to use an XO instead a crystal but I'm sure that it couldn't make miracles. This the way I use ad ASRC in my project.

Aiace


Top
 Profile  
 
PostPosted: Sat Feb 28, 2009 1:14 pm 
Offline
User avatar

Joined: Mon Dec 13, 2004 2:32 pm
Posts: 1323
@ erik_s:

You can safely use an external clock with PCM270x, that’s how I use it for years. Quoted text is a bit ambiguous. FWIW, I’m feeding that external clock from the same supply I use for PCM2706. Whether this qualifies as “external supply” or not... you bet. Yes, it does sounds better with good external clock, even if general jitter performance is actually determined by PCM’s PLL, and its jitter is way higher than jitter of any barely decent XO. There is apparently more on jitter than a simple number of picoseconds.


@ aiacel:

Sorry if this sounds a bit rude, but essentially, claiming something like 500ps is too much for high end system has no more sense than claiming something like 0.05% THD is too much for high end system. We all know it can go down to 0.00005%, don't we? Actually, all the fuss about the USB has been started exactly with these BB chips. If you look a couple of years back, you will even find the claims of achieving 70ps with them. I've never managed to make anything even close to it, and I have no idea how did they do that, but I am sure that the chip can work reasonably well, even if it is not any definitive solution.

BTW, where does that information about different jitter performance of PCM2704 and PCM2707 come from?


Top
 Profile  
 
PostPosted: Sat Feb 28, 2009 6:05 pm 
Offline
User avatar

Joined: Tue Feb 03, 2009 8:30 pm
Posts: 45
Hi Pedja
Don't warry you haven't been rude at all. I'm going to try and explain what I mean when I wrote Hi-end.
Erik is going to use the PCM1704 that is a 24bit DAC. If he is going to use 192kHz, the jitter limit is 0,3 ps, too far away from the 500ps of the PCM2707. But 500ps colud be right for a 16bit 44kHz DAC (that has 350ps limit). In other words he might not have any advantage to use a 24 bit instead of a less expensive DAC. Obviusly these numbers are just theory, the realty is worse.
I'm trying to use an ASRC to overecome the probem, but I'm not sure it will work.
Certanly I mistake to use "hi end" but it's was more short to write... :beg:

I don't remeber where I read about the 500ps, it may be wrong; but here there is an intersting discussion http://www.diyhifi.org/forums/viewtopic.php?f=2&t=432&st=0&sk=t&sd=a

Where did you find the you can achive 70ps with PCM2707 ? My PCB has been just shiped to me, but maybe with some wires I might improve it ;)

Hi
Aiacel


Top
 Profile  
 
PostPosted: Sat Feb 28, 2009 7:58 pm 
Offline
User avatar

Joined: Tue Jan 31, 2006 4:16 pm
Posts: 567
Location: Zinzinnati, Ohio
Gang,

In regard to jitter and the PCM270x series. Allot of this is due to the fact that the part is changing the internal clocking every 1ms. We used the Wavecrest DTS system with the following results running the Faber MAC based Oscillator at 1KHz sine wave to various dacs and measuring the WCLK jitter at the output of the I2S USB receiver @ 44.1KHz Fs:

PCM2706 3433ps
TAS1020B Adaptive mode changed every 4ms standard code: 2838ps
TAS1020B Adaptive my slow mode PLL code: 632ps
TAS1020B Async USB mode internal PLL: 482ps
TAS1020B Async USB mode OSC to MCLKi port generating I2S output: 73.2ps

FYI for reference the TAS1020B Adaptive mode standard code 4ms also had SPDIF inputs and the WCLK jitter was like 671ps for 44.1KHz being feed from the Prism dScope III SPDIF generator.

The PCM2706 and the TAS1020B Adaptive standard code were both other people's products, but the PCM2706 did have a 12MHZ crystek oscillator.

I feel the adaptive mode itself is much to be blamed for the high jitter. As you can see if you re-write what is called the SoftPLL code for Adaptive mode you can regain allot of jitter but then again you need the Keil compiler and a whole lot of patients as the "Reference Code" that TI provides is too huge to understand in one pass.

Thanks
Gordon

_________________
Gordon
Chief Scientist
Wavelength Audio


Top
 Profile  
 
PostPosted: Sat Feb 28, 2009 8:18 pm 
Offline
User avatar

Joined: Mon Dec 13, 2004 2:32 pm
Posts: 1323
aiacel wrote:
+Erik is going to use the PCM1704 that is a 24bit DAC. If he is going to use 192kHz, the jitter limit is 0,3 ps, too far away from the 500ps of the PCM2707. But 500ps colud be right for a 16bit 44kHz DAC (that has 350ps limit). In other words he might not have any advantage to use a 24 bit instead of a less expensive DAC. Obviusly these numbers are just theory, the realty is worse.
I'm trying to use an ASRC to overecome the probem, but I'm not sure it will work.

It is not actually that wrong to associate jitter sensitivity to the resolution, but it is not that simple either, as simple numbers don’t really tell much. In fact, if you assume 44.1kHz / 16 bits R2R conversion versus 176.4kHz / 24 bits R2R, 3 picoseconds of jitter will produce entirely the same artifacts. Whether the distribution of intrinsic artifacts of converter covers them or not is another story, and again not short story.

Quote:
I don't remeber where I read about the 500ps, it may be wrong; but here there is an intersting discussion http://www.diyhifi.org/forums/viewtopic.php?f=2&t=432&st=0&sk=t&sd=a

Been there... :) As far as I know, the PCM2704/5/6/7 use the same PLL. The PCM2702 and PCM290x may be inferior though.

Quote:
Where did you find the you can achive 70ps with PCM2707 ? My PCB has been just shiped to me, but maybe with some wires I might improve it ;)

I found the claims like that. Must be some forum, possibly AudioAsylum. The number looked to me rather like an outcome of biding.


Top
 Profile  
 
PostPosted: Sat Feb 28, 2009 11:22 pm 
Offline

Joined: Tue Nov 11, 2008 1:46 pm
Posts: 97
Location: Toronto
Pedja,

Thank you for the vote that it will work.

Aiacel,

I decided to not use the SRC4192 based upon what I've read. While the SRC can do something to reduce jitter it is not the be all and end all device. And the other problem that I looked at is that is that all my music is 44.1/16. I could send the stream to the PCM1794 at 192/24, but I'm not convinced that the SRC would make it sound better, even if the PCM1794 may sound better at the higher bit rate.

Gordon,

Having lurked here for a while, I expected for you to come out with exactly what you said. You response is "perfect" for me in that it put hard facts to the various solutions. I'd love to use the TAS1020B. Slight problem with that is the last real programming I've done was about 20 years ago when I was trying to be a Computer Science major. Give me a year or two and I might be there.

In the meantime, I still end up with a sort of question. Crystals and Oscillators are rated in PPM. This will be their "accuracy". The crystal that I'm using now is rated at +-30PPM. Quick look at Digikey shows a stock of Crystek Oscillators in the +-25-100PPM range (for 12Mhz). This implies to me that you can end up choosing an oscillator that is no better than the crystal that your are replacing. Would you agree that this could be a problem?

Based upon using Digikey as a supplier, the oscillator that I'm looking at right now is the FOX924B (+-2.5ppm). Anyone have any comments about this part?


Top
 Profile  
 
PostPosted: Sat Feb 28, 2009 11:29 pm 
Offline
User avatar

Joined: Mon Dec 13, 2004 2:32 pm
Posts: 1323
For how long should we repeat that the ppm tolerance has nothing to do with jitter? :banghead:


Top
 Profile  
 
PostPosted: Sat Feb 28, 2009 11:55 pm 
Offline

Joined: Tue Nov 11, 2008 1:46 pm
Posts: 97
Location: Toronto
Pedja wrote:
For how long should we repeat that the ppm tolerance has nothing to do with jitter? :banghead:


Read line one of my first post. Sorry.

Can you point me to something I can read to make me less dumb?


Top
 Profile  
 
PostPosted: Sun Mar 01, 2009 12:11 am 
Offline
User avatar

Joined: Mon Dec 13, 2004 2:32 pm
Posts: 1323
OK, sorry.

Absolute frequency tolerance, expressed in ppm (parts per million), is one issue, often very important in many application fields, but it is however of no use in the audio. This because it really doesn't matter if your system plays 0.01% (this is 100ppm, which is easily attainable) faster or slower than is supposed to do. It also really doesn't matter for the XO driving the PLL (that's what for it is in this case), so long as it is within the PLL pull-up range.

The jitter, as opposed to the absolute frequency tolerance, is a short term variation in frequency. These short term variations, again, have their own frequencies. SONET proposes 12kHz-20MHz as a range that jitter should be specified. This is probably fine in telecommunications, but in the audio the lower frequency variations, down to 1Hz or less, are of interest too.

Some call above mentioned absolute tolerance a "DC jitter component", but it is really no more than poetry.

HTH.


Top
 Profile  
 
PostPosted: Sun Mar 01, 2009 1:05 am 
Offline

Joined: Tue Nov 11, 2008 1:46 pm
Posts: 97
Location: Toronto
No worries. After I posted, I went and read a bit and then my brain kicked in.

So if I've got this right:

It is like matching resistors. For a lot of cases, we don't care about the absolute value of the resistor, we just want two that have the same value so we can place them in matching spots in each channel.

In this case, we don't care that over 100,000 cycles that the clock averages 12MHz. What we care about is that, over X cycles, where X is small, that we have a stable frequency (whether it be exactly 12MHz or not).

What we want is this (showing just high and low):
Code:
--   --  --  --
  --  --  --  --


not this
Code:
---  --   --  ---
   --  ---  --


But wouldn't this also be similar to the resistor scenario - as you approach higher absolute tolerance, you will also get more accurate local tolerance? That is to say super low PPM (0.01ppm) will also equate low jitter by necessity to achieve the long measurement accuracy. Now then again, $265CDN at Digikey for 0.01ppm is a little beyond my budget.

Now to go an find an appropriate clock....


Top
 Profile  
 
PostPosted: Sun Mar 01, 2009 1:12 am 
Offline
User avatar

Joined: Mon Dec 13, 2004 2:32 pm
Posts: 1323
As I said, you simply don't need a long term accuracy, as it has nothing to do with short term "accuracy" (a.k.a. jitter), and no, they are not related. So you have no reason to spend even one cent for it.


Top
 Profile  
 
PostPosted: Sun Mar 01, 2009 7:37 pm 
Offline
User avatar

Joined: Tue Jan 31, 2006 4:16 pm
Posts: 567
Location: Zinzinnati, Ohio
Gang,

Jocko did some measurements on my Crystek custom units. These look real good for jitter if they are fed with a really good low noise supply. He measured -100dB phase noise on my 22.5792MHZ units. Mouser has the C33xx SMD 5mmX7mm 12Mhz units that would be your best bet for the PCM270x series.

Pedja,

Yes you are right the PCM29xx and 2702 units do have much more inferior of a setup as these were what TI calls the series one usb audio devices. They learned allot from these when they went to the PCM270x series. I have some proto board around here from like 2000 with the PMC29xx part on it with a guitar and mic input thingy I made with tubes and stuff. Probably not worth testing it for jitter.

Thanks
Gordon

_________________
Gordon
Chief Scientist
Wavelength Audio


Top
 Profile  
 
PostPosted: Sun Mar 01, 2009 8:34 pm 
Offline
User avatar

Joined: Mon Dec 13, 2004 2:32 pm
Posts: 1323
Well, yes... I have recommended Crystek to the DIY audio world back in 2004. :cool:

Yet... that 22.5792MHZ of yours is Crystek? I’ve never come across any Crystek for any CD frequency.


Top
 Profile  
 
PostPosted: Sun Mar 01, 2009 8:41 pm 
Offline
User avatar

Joined: Tue Jan 31, 2006 4:16 pm
Posts: 567
Location: Zinzinnati, Ohio
Pedja,

I ordered these custom and have several of the required frequencies now. Mouser has some of the 24.576 but no one had any 11.2896 or 22.5792 so I bought reels of them.

Thanks
Gordon

_________________
Gordon
Chief Scientist
Wavelength Audio


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 17 posts ]  Go to page 1, 2  Next

All times are UTC


Who is online

Users browsing this forum: Google [Bot], vzs and 4 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB® Forum Software © phpBB Group