need some advise in clock dividing - diyAudio
Go Back   Home > Forums > Source & Line > Digital Line Level

Digital Line Level DACs, Digital Crossovers, Equalizers, etc.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 16th March 2011, 12:06 PM   #1
jazz is offline jazz  Netherlands
diyAudio Member
 
Join Date: Dec 2001
Location: leiden, physically that is...
Default need some advise in clock dividing

Hi,

I’m planning to build a dac and I’m getting a bit lost in clock division for reclocking. I will be using a cs8412 in mode 3 (fs and bck are inputs), I2S separator as per the design by Pedja Rogic to tda1541’s in balanced mode.

I intend to use a 11,289 clock which I will divide down to 2,83 mhz for bck and 44,1 khz for fs. Previously I’ve used a 4040 counter for division as it is used in the dddac but I know this might not be the best solution. Another option is using cascaded 74hc74’s like used by mr Tent. But while reading up I came across synchronous and asynchronous binary counter circuits that might also work. Would it be advisable to use such a synchronous counter circuit (example: [URL="http://www.electronics-tutorials.ws/counter/count_3.html) or does it make more sense to use cascaded 74hc74’s?

I’m looking for a compromise between simplicity and the best solution because I will have to design the pcb myself and I want to keep the complexity within reasonable limits. So if I can limit the design to 3 to 4 logic chips I’d like that.

Regards,
Joris

Btw I am planning for a clock output to slave the transport to the dac clock.

edit:link
__________________
oh, what a beautifull day
I wanna go out and play!

Last edited by jazz; 16th March 2011 at 12:10 PM.
  Reply With Quote
Old 16th March 2011, 07:30 PM   #2
poynton is offline poynton  United Kingdom
diyAudio Member
 
poynton's Avatar
 
Join Date: Mar 2005
Location: A life on the ocean waves when I'm not at home in N. Wales (but I'm not Welsh so no sheep jokes!)
I would use 74hc74s

Andy

.
__________________
If it ain't broke, break it !! Then fix it again. It's called DIY !
  Reply With Quote
Old 17th March 2011, 06:02 PM   #3
jazz is offline jazz  Netherlands
diyAudio Member
 
Join Date: Dec 2001
Location: leiden, physically that is...
Hi,

Thanks, I'll read up some more but that's what my feeling is as well. Now to find a decent way to layout the pcb...

regards
Joris
__________________
oh, what a beautifull day
I wanna go out and play!
  Reply With Quote
Old 19th March 2011, 09:07 AM   #4
diyAudio Member
 
Join Date: May 2006
Location: Holland
Quote:
I’m planning to build a dac and I’m getting a bit lost in clock division for reclocking. I will be using a cs8412 in mode 3 (fs and bck are inputs), I2S separator as per the design by Pedja Rogic to tda1541’s in balanced mode.
This setup will lead to repeated / dropped samples. This will be audible as periodic clicks. These clicks are easiest to hear with sine wave test tones, the larger the frequency difference the shorter the time between the clicks.

The clicks are caused by the fact that the PLL clock inside the CS8412 is still used to decode incoming SPDIF signal (clock #1) and the external clock (clock #2) determines the moment this data is clocked out. Since both clocks don't run in sync and it is likely frequencies differ / drift up to 200ppm, samples have to be dropped / repeated, this causes the clicks.

Only way to get rid of the clicks is perfectly synchronizing both clocks using a (D)PLL or frequency tracker circuit based on a microcontroller for example.

The dividers need to be synchronous dividers like the 74xx161 series. The 74xx4040 series are non-synchronous ripple counters that add significant jitter (typical 200ps on BCK). With synchronous counters there is a fixed propagation delay between clock and each output as all outputs change simultaneously (synchronous). The outputs of a ripple counter change sequentially (compare it with domino effect), so the last output will have largest propagation delay (and jitter). For 11.2896 MHz clock this will lead to approx. 200ps of jitter added to the BCK signal and approx. 800ps added to the WS signal.



The PLL clock from the CS8412 usually has slightly lower jitter compared to the derived BCK and WS signals. So I suggest to run CS8412 in master mode instead, using MCK signal to synchronously reclock BCK, DATA, and WS signals.

All I2S signals need to have lowest possible jitter, if only WS and BCK are reclocked, jitter still passes through DATA signal (crosstalk).

In order to minimize jitter in the synchronous reclockers (D-flip-flops), use D flip-flops with shortest propagation delay (CLK > Q). The HC(T) series will work, but will add approx. 100ps of extra jitter for each flip-flop and approx. 50ps of extra jitter for each (cascaded) gate.



Best is to integrate DAC with source (when possible), skip the flawed SPDIF interface and use direct I2S between source and DAC. Then use a common low jitter clock to drive both source and DAC.

The best external DAC applications require slaving the source through a clock signal that's fed back from DAC to source, or a servo system that will synchronize both source and DAC clock. Here it is also advised to use I2S interface instead of SPDIF like with the Zanden Transport / DAC. Typical jitter with highly optimized slaved source will be approx. 50ps.

What Slaving Does : LessLoss high end audio power cables, audiophile power cables, audiophile cables

Last edited by ecdesigns; 19th March 2011 at 09:09 AM.
  Reply With Quote
Old 19th March 2011, 02:08 PM   #5
jazz is offline jazz  Netherlands
diyAudio Member
 
Join Date: Dec 2001
Location: leiden, physically that is...
Hi Ecdesigns,

Thanks for your reply. I still have to read it a couple of times before i fully understand. I'll do some further reading as well.

Unfortunately combining the transport and dac in to one is not an option due to the chosen case format. There simply is not enough space. I do plan on having a clock output for slaving the transport (like the tentlink). That should solve the repeated/dropped samples issue I thought (?) because all clocks should then run in sync.

As i want the option to use the dac with other transports as well I'd like to keep the spdif interface (flawed as it is) but i will think about direct i2s. Maybe using a pin header in the i2s lines is an option so i can use spdif via jumpers and i2s through an add on board. I'll think about it.

As far as I understand your post now I think you say that a synchronous clock divider is to be preferred. I think I saw such a circuit by you for dem clocking (74xx161?). I haven't drawn that out in to a logic schematic yet.

The link I tried to post gives two examples of such a synchronous divider on page 3. Is that something I should examine further and compare to the dem clock schematic mentioned above? At first glance the master clock is always in control and reclocks each division.

I will think about including a switching option between using the cs8412 in master mode and mode 3 (dip switch for mode selection and pin header+jumper in the I2s lines). I'm also considering a combination: divide by 4 using 74xx74's as per mr Tents schematic for bck and then a synchrounous divider for ws.

Please note that even though I've build a few dac's, this is the first time i'm actually trying to make my own choices in the design. Sort of experimental cooking instead of following the recipe and without any proper training or education. So if I say a few dumb things, tell me but don't send the inquisition after me please.

One more question. You refer to reclocking sdata. I thought that when using mode 3 Sdata would be reclocked in the cs8412 buffer? And even without all the previous I'm going to separate i2s in to separate left and right datastreams. If reclocking sdata is that important would it be wise to reclock sdata (and bck ws) after the i2s splitter? And if so should i do it before the i2s splitter as well?

Thanks again for your response and Poynton's. In the end the choiches i make will depend both on circuit/performance considerations and my ability to design the pcb layout. This is way complex and very difficult to me.

Kind regards,
Joris
__________________
oh, what a beautifull day
I wanna go out and play!
  Reply With Quote
Old 12th April 2011, 03:35 PM   #6
noizas is offline noizas  Lithuania
diyAudio Member
 
Join Date: May 2009
Location: Vilnius, Lithuania
hi jazz,

I'm interesting how are you succesfull in your project?

I'm also doing something like you except I want to "feed" Pedja Rogic's I2S splitter from different sources: SPDI_1-DIR9001, SPDIF_2-CS8414, direct I2S from CD player and from Valab's Teralink USB to I2S receiver.

Quote:
Originally Posted by jazz View Post

One more question. You refer to reclocking sdata. I thought that when using mode 3 Sdata would be reclocked in the cs8412 buffer? And even without all the previous I'm going to separate i2s in to separate left and right datastreams. If reclocking sdata is that important would it be wise to reclock sdata (and bck ws) after the i2s splitter? And if so should i do it before the i2s splitter as well?

Thanks again for your response and Poynton's. In the end the choiches i make will depend both on circuit/performance considerations and my ability to design the pcb layout. This is way complex and very difficult to me.

Kind regards,
Joris
I'm going to make reclocking on MLCK basis AT LEAST after I2S splitting because HEF4517 and 74157 will add delay more than 200ns from front edge of BCK.
I gues also to try delay BCK for 1, 2 or 3 period of MLCK because period of BCK is about 354ns if you gonna use 64bit frame I2S.

On the other hand, recloking I2S after receiving it from transport also can be reasonable, except this will also add some jitter .

Best wishes
Saulius
  Reply With Quote
Old 19th April 2011, 07:13 PM   #7
jazz is offline jazz  Netherlands
diyAudio Member
 
Join Date: Dec 2001
Location: leiden, physically that is...
Hi Saulius,

Well i did find the 74hc590 which would allow me to keep clock division on one chip and keep it synchronous as well. Fit's the bill nicely.

Guess i'm lagging behind because I havn't given propagation delay much consideration yet (new to digital design, though i have build various dac's from others schematics). I'd have to compare the 74hc590 to the 161 to get an idea yet have little to no clue about it's real effect.

Still thinking about reclocking. I don't want to use 3 quad flipflop's for that so i'm gonna search for a single or dual dip style flipflop. I take your point about placing it after the i2s splitter. If i do it, i'll do it that way.

I'm curious about your reason for using two different recievers and the way you plan to do the switching between the various inputs. I've stalled that issue for now and will probably start with a pin header and some jumpers. If i decide to do the switching i'll probably do it later.

Because i've recently started a new job and just started on my first assignment i'm rather busy with a lot of travel time. Thus time is sparse and mostly spent on doing research and learning. The first concept design is expected in the fall in the most positive scenario. will probably do some breadboarding tests with the dac/iv stage (which is ready) first and do some testing of power supplies.

Kind regards,
Joris
__________________
oh, what a beautifull day
I wanna go out and play!
  Reply With Quote
Old 23rd April 2011, 08:56 AM   #8
Zoran is offline Zoran  Serbia
diyAudio Member
 
Join Date: Jan 2004
Location: Belgrade
You do not need to divide MCK this vale You have is for clock in at reckl.
But if You going to master mode on receiver chip You have to somehow to divide MCK make SCK and BCK. Mind the time delay of dividng lines and try to compensate this...
I am running the I2S splitter but not in balanced mode rather the L/L and R/R
working without any clicks with dual mono dac-s. I did not try the balanced splitter konfiguration.
with this rearranged circuit, which is predecessor of all same kind circuits
cheers
__________________
###
  Reply With Quote
Old 23rd April 2011, 09:08 AM   #9
Zoran is offline Zoran  Serbia
diyAudio Member
 
Join Date: Jan 2004
Location: Belgrade
he scematic was published in
hi-fi news & record review, september 1990
part is balanced I2S splitter circuit
the chips are not marked but those are the same chips like in other similar cir.
HEF4517, 74HC157
Attached Images
File Type: jpg I2S balanced splitter 1990.jpg (139.5 KB, 236 views)
__________________
###
  Reply With Quote
Old 23rd April 2011, 09:04 PM   #10
noizas is offline noizas  Lithuania
diyAudio Member
 
Join Date: May 2009
Location: Vilnius, Lithuania
Quote:
Originally Posted by Zoran View Post
he scematic was published in
hi-fi news & record review, september 1990
part is balanced I2S splitter circuit
the chips are not marked but those are the same chips like in other similar cir.
HEF4517, 74HC157
Hi

here are attached few items:
1.the same picture published in 1990 HiFi News, but in better quality - easier to see number of input/outputs.
2. Second picture is same circuit but balanced I2S splitter part only.
3. the article of Pedja Rogic, where is mentioned this 1990 HiFi News schematic.

Please take into account 1990 HiFi News shematic is ok only for 32 bit word (all L and R bits transferred per 32 BCK, as after SAA7210 and SAA7310).
So, if you are going "to steal" I2S directly from CD player based of SAA7210 or SMD SAA7310, this is realy operational schematic.

For 64 bit word (most of SPDIF receivers) we need to use schematics proposed by Pedja.

Those two shematics should operate ok, but with 5V power suply for HEF4517 only for 44.1kHz sample rate. And mainly for NXP HEF4517, other clones of 4517 can be simply to slow.
For higher sample rates here is a way to increase power supply to 10 - 15V, almost proportionally decreasing propagation delay, but then you need to play with pull up resistors and diodes to protect 74HC157 inputs from overvoltage. The you can freely use not only NXP IC.

I'm going to use in my project both schematics as I have both types of I2S source - 32 and 64 bit frame.
Attached Images
File Type: png dsm_bs.png (136.6 KB, 215 views)
File Type: gif I2S split 32 bit word.gif (8.9 KB, 203 views)
Attached Files
File Type: pdf I2Sbal_split.pdf (27.0 KB, 45 views)
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Kwak clock not working - pls advise Assyst1606 Digital Line Level 31 1st December 2008 01:48 AM
Dividing power between different speakers from single source & Combining channels Fusion-One Parts 0 10th April 2007 02:18 AM
Accuphase DF-45 Digital Frequency Dividing Network oceanic30 Multi-Way 6 4th September 2006 08:05 AM
CS8412, PMD100, PCM63, onboard clock + [word] clock output implementation stolbovoy Digital Source 7 3rd December 2004 08:18 PM
Dividing heatsinks? AudioAngel Pass Labs 2 12th November 2004 06:14 PM


New To Site? Need Help?

All times are GMT. The time now is 02:30 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2