Go Back   Home > Forums > Source & Line > Analog Line Level

Analog Line Level Preamplifiers , Passive Pre-amps, Crossovers, 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 31st July 2013, 06:40 AM   #11
ds23man is offline ds23man  Netherlands
diyAudio Member
 
Join Date: Oct 2006
Quote:
Originally Posted by jean-paul View Post
On the contrary: this is correct. Cirrus decouples rails and between V+ and V- which is both old fashioned and good. Data sheet also mentions this in text. Please check their reference board where they did it like that on a real PCB. Their engineers would not make the same "mistake" twice I hope:

http://www.cirrus.com/jp/pubs/rdData...DB3318_DB2.pdf
Ok, if some noise enters for example the positive rail, where should it go? To the ground and not to the other rail........
  Reply With Quote
Old 31st July 2013, 06:43 AM   #12
diyAudio Moderator Emeritus
 
jean-paul's Avatar
 
Join Date: Sep 2002
Location: Germany
Wrong thinking, it's about closing current loops. Noise should go to the point with lowest impedance to GND. Noise does not have much impact on the other rail as the opamps itself has a certain PSRR.

As said it is an old but good technique to keep some high speed opamps performing optimally. If memory serves me right it also is a way to compensate for a technical imperfection in the opamps output transistors but I tried to google it but could not find it. Anyway I decouple high speed opamps like this since ages and some won't perform good without this form of decoupling (like OPA627/637). One can also add series resistors in supply lines but I don't like to see my output signal modulated on those resistors so I omit these if the deice is "silent". Ferrite beads are very effective too (before the caps) but this is another subject.
__________________
It's only audio. Official member of the Norske Brillegeit Gang.

Last edited by jean-paul; 31st July 2013 at 07:01 AM.
  Reply With Quote
Old 31st July 2013, 07:04 AM   #13
ds23man is offline ds23man  Netherlands
diyAudio Member
 
Join Date: Oct 2006
If it is wrong thinking, all major audio designers have been wrong. Can not think of any pro audio console ( pa or studio) with decoupling between the supply rails. Always between rail and ground.....
  Reply With Quote
Old 31st July 2013, 07:10 AM   #14
diyAudio Moderator Emeritus
 
jean-paul's Avatar
 
Join Date: Sep 2002
Location: Germany
It is not "wrong" to add decoupling from V+ to V- nor is it "wrong" to forget it with relatively slow devices. BTW when decoupling between rails one must also decouple rails to GND.

It just is a good way for keeping high speed opamps performing best. You don't need them with the darn NJM4580s as used in equipment (cheap and slow, only used because of their low price).

Cirrus does know best what kind of opamps they use in the chip and if they "prescribe" how to decouple their opamps the best way I would follow that advice.
__________________
It's only audio. Official member of the Norske Brillegeit Gang.

Last edited by jean-paul; 31st July 2013 at 07:14 AM.
  Reply With Quote
Old 31st July 2013, 07:15 AM   #15
ds23man is offline ds23man  Netherlands
diyAudio Member
 
Join Date: Oct 2006
Quote:
Originally Posted by jean-paul View Post
It is not "wrong" to add decoupling from V+ to V- nor is it "wrong" to forget it with relatively slow devices. BTW when decoupling between rails one must also decouple rails to GND.

It just is a good way for keeping high speed opamps performing best. You don't need them with the darn NJM4580s as used in equipment (cheap and slow).

Cirrus does know best what it uses in the chip and if they "prescribe" how to decouple their opamps the best way I would follow that advice.
Do not follow datasheets blindly these days, they are full of faults and ommisions.... The CS3318 datasheet also, ommissions in powersupply sequencing and mistakes in pin listings.....

Proper opamp decoupling, see figure 2:

http://www.analog.com/static/importe...als/MT-101.pdf
  Reply With Quote
Old 31st July 2013, 07:20 AM   #16
diyAudio Member
 
jan.didden's Avatar
 
Join Date: May 2002
Location: Great City of Turnhout, Belgium
Blog Entries: 7
As a long-time user of the 3318, I urge you to also bring out the reference (or ground) pins of each input and output next to the particular i/o pin.
If you lump every ref pin together to a central ground, you get much less performance from this chip then possible.

Keeping each i/o paired to an actual input/output signal as long as possible improves the S/N ratio and the distortion performance.
There is a very good reason why they went to this expensive route to almost double the number of pins!

jan
__________________
Never explain - your friends don't need it and your enemies won't believe you anyway - E. Hubbart
Check out Linear Audio Vol 7
!
  Reply With Quote
Old 31st July 2013, 07:22 AM   #17
diyAudio Moderator Emeritus
 
jean-paul's Avatar
 
Join Date: Sep 2002
Location: Germany
Thanks for posting that link: it is a good document but please see the second (!) reference to a much older but even more valuable document. The link in the document is broken but I regoogled it.

It is here :

http://www.gmee.deit.univpm.it/bibli...amp/AN-202.pdf

It is not that hard to get used to a technique one does not know (yet). Again: think high speed opamps, think current loops. Since high speed devices are more often used decoupling should get more attention than it got till now. Let's say that one can use a 741 without any decoupling but try this with a high speed opamp...

The third reference in the document you linked is a good read too (same author as the other document):

http://www.analog.com/static/importe...1066AN-342.pdf
__________________
It's only audio. Official member of the Norske Brillegeit Gang.

Last edited by jean-paul; 31st July 2013 at 07:31 AM.
  Reply With Quote
Old 31st July 2013, 02:25 PM   #18
Dr_EM is offline Dr_EM  United Kingdom
diyAudio Member
 
Dr_EM's Avatar
 
Join Date: Oct 2006
Location: Swindon
Good discussion here, it certainly shows how much more sophisticated a 'universal' design can need to be compared to an application specific one.

To implement these changes we will need to accept a larger board size and eventual higher cost but I will try to be as efficient as I can with it. Here's what I propose:
  • Input discrage/terminating resistors, probably around 33k to avoid degrading performance with Johnson noise.
  • So is the decoupling used acceptable? The decoupling to ground is back at the voltage regulators so is it too far away? However there is no analogue ground pin as such, only signal grounds, so I'm assuming this is the correct location for it.
  • I'm doubtful about output coupling capacitors as I think few will need those and as you say, if the pads are added we all need to use jumpers, plus additional board area.
  • In/out connectors I think I'll change to two sets of 8 2-pin Molex style, rather than the IDE/cluster type, which needs ribbons ideally. RCA on-board would easily double board area and personally I wouldn't want that as neither input or output would be directly interfaced from the chassis. Panel-board wiring is always possible with the Molex.
  • I could integrate the control board to it, certainly most manufacturers do this, but for DIY people may prefer the modular approach. It'd even be perfectly acceptable to produce your own control boards using veroboard if people prefer to save money and roll their own. Depends if people agree that my control board offers all the functionality they could need? I wasn't particularly planning to include menus for multi-channel level balancing (rather I'd set it in the firmware) but certainly some people might like this?
  • A watchdog chip shall be added to monitor the 3.3V rail, and if it fails, put the CS3318 into reset, disabling the analogue systems. Good call ds23man!
  • The MUTE control input shall be linked to GND with a resistor, say 10k?
  • It's a challenge but I can try to bring out the individual GND references to those input and output Molex headers since you say it can bring about improved performance.

Thanks all, I will update you when I have made the changes!

Last edited by Dr_EM; 31st July 2013 at 02:30 PM. Reason: Added bullets
  Reply With Quote
Old 31st July 2013, 07:06 PM   #19
Dr_EM is offline Dr_EM  United Kingdom
diyAudio Member
 
Dr_EM's Avatar
 
Join Date: Oct 2006
Location: Swindon
With regard to bringing out the Input/Output GND references to their respective headers, it's curious to note that on their own evaluation board Cirrus have lumped the grounds together underneath the IC package exactly as in my current revision (rotated to match the orientation of my board):

Click the image to open in full size.

where the measured THD+N plot shows as the IC specification level of -112dB. I'm gussing the main benefit of seperating the ground paths out to the headers is in reducing crosstalk by modulation of grounds by strong signals on adjacent channels? It does call into question why they have gone to the effort to provide all those seperate references if they then go on to bond them directly together on their own reference board, but regardless I suspect an external ground tie point is lower impedance than connecting inside the device package.
Attached Images
File Type: png CS3318 demo board.png (38.3 KB, 224 views)
  Reply With Quote
Old 1st August 2013, 01:37 PM   #20
Nisbeth is offline Nisbeth  Denmark
diyAudio Member
 
Nisbeth's Avatar
 
Join Date: Jan 2001
Location: Denmark
My totally unscientific opinions
Quote:
Originally Posted by Dr_EM View Post
  • I'm doubtful about output coupling capacitors as I think few will need those and as you say, if the pads are added we all need to use jumpers, plus additional board area.
  • In/out connectors I think I'll change to two sets of 8 2-pin Molex style, rather than the IDE/cluster type, which needs ribbons ideally. RCA on-board would easily double board area and personally I wouldn't want that as neither input or output would be directly interfaced from the chassis. Panel-board wiring is always possible with the Molex.
Input caps are probably a good idea, but output caps I think should be left out. Normally I would expect that you have more control over which poweramp(s) are being used than over which sources are connected. If the input side is capacitor coupled, all you need to ensure is that the subsequent power amp can handle the offset (or has capacitors in the input) generated by the CS3318 itself.

RCAs onboard is generally annoying for me, because not only does it cost more in terms of board area, it also ties you to specific connectors and to a specific case mounting.

Quote:
Originally Posted by Dr_EM View Post
  • I could integrate the control board to it, certainly most manufacturers do this, but for DIY people may prefer the modular approach. It'd even be perfectly acceptable to produce your own control boards using veroboard if people prefer to save money and roll their own. Depends if people agree that my control board offers all the functionality they could need? I wasn't particularly planning to include menus for multi-channel level balancing (rather I'd set it in the firmware) but certainly some people might like this?
Again, I think it should be left out. This board could be used in many different applications, so it makes sense to allow some flexibility in the control method as well (PIC, AVR, Arduino, Raspberry PI etc.)

/U.
__________________
DIY projects blog
  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
Eagle vs. Sprint-Layout for PCB design/layout hollowman Parts 11 12th January 2014 09:01 PM
cs3318 and arduino linuxworks Analog Line Level 72 26th March 2013 11:56 PM
Digital Volume Contol - CS3318 - MAX5440 Ridin '24'z Digital Source 5 8th June 2011 01:36 PM
Who can do PCB layout d3imlay Solid State 3 23rd November 2007 11:55 PM
help for a PCB layout pencoat Parts 0 4th October 2006 01:52 AM


New To Site? Need Help?

All times are GMT. The time now is 08:36 PM.


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