Go Back   Home > Forums > >
Home Forums Rules Articles diyAudio Store Blogs Gallery Wiki Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

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

Grossly parallel multibit DAC adventures
Grossly parallel multibit DAC adventures
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 October 2019, 01:47 PM   #21
abraxalito is offline abraxalito  United Kingdom
diyAudio Member
 
abraxalito's Avatar
 
Join Date: Sep 2007
Location: Hangzhou - Marco Polo's 'most beautiful city'. 700yrs is a long time though...
First two boards of the new improved groundfill 36 DAC PCB. Tested for shorts (one DAC chip had one, out of 72) but not for SQ yet. Eagle eyed readers will notice they're slightly enlarged vertically to allow more area for the fill.
Attached Images
File Type: jpg 1515823489.jpg (218.5 KB, 290 views)
__________________
I know you think you understand what you thought I said but I'm not sure you realize that what you heard is not what I meant - Alan Greenspan
  Reply With Quote
Old 19th November 2019, 07:46 AM   #22
abraxalito is offline abraxalito  United Kingdom
diyAudio Member
 
abraxalito's Avatar
 
Join Date: Sep 2007
Location: Hangzhou - Marco Polo's 'most beautiful city'. 700yrs is a long time though...
Default Progress update

Four boards with the new layout have been now listened to and don't suffer from the low-level hiss the old layout does, which is good news. I was going to release the gerbers for that but in the meantime we decided to try out another experiment - using thicker copper to get a lower impedance, both to the supply rail and to the output pins of the DACs. When paralleling a large number of (over 100) chips the effective impedance the transformer presents is going below 100mohm hence the resistance of the tracks to the chips starts to become significant. A minor change was made to the layout for this 2oz copper experiment - the output pads are now more conveniently placed to connect twisted pairs to.

In other news the PCB layout of the main analog board has begun. For this board the aim is to have almost all through-hole components - just the NP0 capacitors for the LC filters will be SMT - that hopefully makes building a kit less daunting for those whose eyes or dexterity (or both!) is/are challenged. The buffers (see my sketch in first post) are being built out of SMT parts as there are just too many components to fit within the 10*10cm window (for a cheap PCB) if the resistors for those cannot be SMT. The buffers are going to be plug-in modules so anyone who shies away from SOT-23 package soldering can order them ready-made. Having modules also means others can experiment with their own designs for the buffers.
Attached Images
File Type: jpg 2ozr2.jpg (227.9 KB, 212 views)
__________________
I know you think you understand what you thought I said but I'm not sure you realize that what you heard is not what I meant - Alan Greenspan
  Reply With Quote
Old 23rd November 2019, 08:47 AM   #23
abraxalito is offline abraxalito  United Kingdom
diyAudio Member
 
abraxalito's Avatar
 
Join Date: Sep 2007
Location: Hangzhou - Marco Polo's 'most beautiful city'. 700yrs is a long time though...
Just listened to the four boards in a stack which were built with the latest layout and 2oz copper. One chip was found to be broken amongst the 144 on those boards. Tracking a dud chip down isn't too difficult so long as the failure of the chip is such that the pin7 multiplying effect isn't also broken (which so far I haven't found). The faulty DAC normally gives its presence away when digital volume is set very low (below -50dB) and sounds like static breakthrough modulated by the music at a level which is louder than the music itself. Shorting each DAC's pin7 to 0V in turn eventually reveals which chip contributes the hash.

Since this new layout is now good to go, I've attached the gerbers and also the schematic. The RC network values on the schematic diagram are just place holders for now and depend on the characteristics of the transformers which are attached - the higher the step-up ratio, the lower the Rs and higher the Cs.
Attached Files
File Type: pdf Schematic_DenaDAC r2.pdf (202.1 KB, 26 views)
File Type: zip Gerber_PCB new dena r2.zip (156.9 KB, 4 views)
__________________
I know you think you understand what you thought I said but I'm not sure you realize that what you heard is not what I meant - Alan Greenspan
  Reply With Quote
Old 25th November 2019, 01:09 AM   #24
matt_garman is offline matt_garman  United States
diyAudio Member
 
Join Date: Feb 2008
Location: Chicago, Illinois, USA
Grossly parallel multibit DAC adventures
That's pretty cool! Did you ever happen to try different numbers of DAC chips to find the optimum? I suspect it's a matter of diminishing returns at some point, if true, it would be nice to find the optimal point.

With all these DAC chips, you have a good amount of current available. What are your thoughts on passive I/V with so many DACs in parallel? Or other, simpler I/V solutions, such as a basic op-amp?

Need pics of the final product!
  Reply With Quote
Old 25th November 2019, 02:00 AM   #25
abraxalito is offline abraxalito  United Kingdom
diyAudio Member
 
abraxalito's Avatar
 
Join Date: Sep 2007
Location: Hangzhou - Marco Polo's 'most beautiful city'. 700yrs is a long time though...
Hi Matt - thanks for joining in here! I can't say (so far) that there really is an optimum, I've built various DACs up over the past few months with chip counts ranging from 36 (single board) up to 288 (eight board stack) and the subjective differences have been pretty subtle. The measurable change when going to higher chip counts is the variation of voltage on the DAC output pins goes down - i.e. with more chips the I/V stage input impedance goes down, and faster than the increase in output current from the DACs.

Passive I/V (direct into a resistor) with so much current available certainly works though I can't really see the point unless a very low output impedance is the only target. There must be some noise cancellation going on in such a situation though I would guess the advantage of such is undone by the increased sensitivity to PSU noise from having so many paralleled chips. I'd not recommend using many paralleled DACs into an opamp for I/V - the opamp's classAB output stage's modulation of the PSU rails would be the likely bottleneck subjectively.

Pics will certainly come in due course - today I'm going to be checking the first draft layout of the analog board then I can put up an interim pic of what that's going to look like.
__________________
I know you think you understand what you thought I said but I'm not sure you realize that what you heard is not what I meant - Alan Greenspan
  Reply With Quote
Old 27th November 2019, 07:19 AM   #26
abraxalito is offline abraxalito  United Kingdom
diyAudio Member
 
abraxalito's Avatar
 
Join Date: Sep 2007
Location: Hangzhou - Marco Polo's 'most beautiful city'. 700yrs is a long time though...
Here's a pic to whet your appetites - those of you who yearn for DAC chip counts into 3 figures that is. This is a stack of 8 PCBs - the original layout PCBs - so the total number of chips is 288. I built the stack as an experiment to see how low a supply voltage a TDA1387 could tolerate and still deliver untarnished audio. Turns out, around 2V or about 1V below its datasheet minimum. (That can't be useful with passive I/V though as there's no output compliance to speak of then.) Lots of chips in parallel are needed when running such low supplies as the output current scales with supply voltage. Hence the 8 board stack. The advantage of lower supply voltage is much lower operating power - power scales as (supply voltage)^2.
Attached Images
File Type: jpg 1932934976.jpg (221.7 KB, 116 views)
__________________
I know you think you understand what you thought I said but I'm not sure you realize that what you heard is not what I meant - Alan Greenspan
  Reply With Quote
Old 29th November 2019, 03:41 AM   #27
hollowman is offline hollowman  United States
diyAudio Member
 
hollowman's Avatar
 
Join Date: Jan 2007
Location: Los Angeles, California
Quote:
Originally Posted by abraxalito View Post
Here's a pic to whet your appetites - those of you who yearn for DAC chip counts into 3 figures that is. .
So ... why y'all be a sugar daddy and send a couple o' FREE samples this here way.
__________________
"I got banned from a Hi-Fi forum. And I liked it."
- Luv, Katy "Hollowman" Perry
  Reply With Quote
Old 29th November 2019, 08:19 AM   #28
sajunky is offline sajunky  South Africa
diyAudio Member
 
Join Date: Jun 2019
Location: Johannesburg
For those who yearn for freebies, it was said from the beginning - no frieebies - beggars may go away.

@abraxalito. Did you test a single board with 96/196kHz sample rate? Do it make sense to populate every second DAC chip and double number of boards for such application? I think of using empty space for extra decoupling.

I do consider new boards as a great progress. Good work.
  Reply With Quote
Old 29th November 2019, 08:38 AM   #29
abraxalito is offline abraxalito  United Kingdom
diyAudio Member
 
abraxalito's Avatar
 
Join Date: Sep 2007
Location: Hangzhou - Marco Polo's 'most beautiful city'. 700yrs is a long time though...
So far no, I haven't tested at 96k or 192k, just at 88k2. A little higher current draw compared to 44k1. I will have a play at higher rates just to see what happens though. Why do you want extra decoupling incidentally? You can slap extra ceramics across the top of the chips if you're so inclined but I can't see the benefit.

I've increased the height of the board stack from 8 to 10 boards now, for a grand total of 360 chips. Its working fine, sounding awesome at just 2.1V. I had a supercap board lying around so I've connected that to the supply, it has 61 * 10F EDLCs on it. Didn't notice any improvement in sound when I brought it in so I figure heroics on the supply is only of marginal benefit. But can't hurt for extra bragging rights as supercaps have a certain 'cachet'
__________________
I know you think you understand what you thought I said but I'm not sure you realize that what you heard is not what I meant - Alan Greenspan
  Reply With Quote
Old 29th November 2019, 08:40 AM   #30
sajunky is offline sajunky  South Africa
diyAudio Member
 
Join Date: Jun 2019
Location: Johannesburg
For those who yearn for freebies, it was said from the beginning - no frieebies - beggars may go away.

@abraxalito. Did you test a single board with 96/196kHz sample rate? Do it make sense to populate every second DAC chip and double number of boards for such application? I think of using empty space for extra decoupling.

I do consider new boards as a great progress. Good work.
  Reply With Quote

Reply


Grossly parallel multibit DAC adventuresHide 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

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
Denafrips discrete R2R Multibit georgehifi Digital Line Level 79 5th June 2019 09:39 PM
Schiit multibit board diy Pigeatgarlic Digital Line Level 1 8th May 2019 03:53 AM
Multi channel and multibit DAC jpak Digital Line Level 4 25th April 2017 05:19 PM
Multibit dac chips - R2R resistors. C6H5COONa Parts 2 12th February 2015 03:22 PM
WTT: Complete Pass D1 (v3 rebuild) multibit DAC tritosine Swap Meet 2 22nd May 2013 11:15 PM


New To Site? Need Help?

All times are GMT. The time now is 09:24 AM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 14.29%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Copyright ©1999-2019 diyAudio
Wiki