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

Vendor's Bazaar Commercial Vendors large & small hawking their wares

Reference DAC Module - Discrete R-2R Sign Magnitude 24 bit 384 Khz
Reference DAC Module - Discrete R-2R Sign Magnitude 24 bit 384 Khz
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 30th December 2017, 04:02 PM   #6831
soekris is offline soekris  Denmark
diyAudio Member
 
Join Date: Jun 2009
Location: KorsÝr, Denmark
Reference DAC Module - Discrete R-2R Sign Magnitude 24 bit 384 Khz
Quote:
Originally Posted by ynmichael View Post
Thanks! I would definitely not want to short my boards this way... This really put an end to the age-old discussion!

Just a quick question while you're here, how much of an improvement would a fully discrete solid state pre-amp / head-amp be over the 0.5% parts used currently...? And could we expect such a fine add-on to be coming soon?...

Also, any comments on the digital/analogue separation in dam1021?...

Thanks for the great product as is!
The used opamps are good, a good discrete amplifier will be better but it will of course depend on how good it is....

As I mentioned before, I actually have a prototype here of a discrete amplifier, had some problems with ultra low frequency noise, not something you can hear but something you can measure. Couldn't remove it complete, but now have it under control. The final version basically ended up in the dac1541.... Might offer it at some point, but don't know when.

The dam1021 analog power is basically filtered by a CRC network, I believe it's fine, other people want other power solutions....
__________________
SÝren - www.soekris.dk
  Reply With Quote
Old 31st December 2017, 02:41 AM   #6832
derekr is offline derekr  Barbados
diyAudio Member
 
Join Date: Mar 2009
Quote:
Originally Posted by soekris View Post
The used opamps are good, a good discrete amplifier will be better but it will of course depend on how good it is....

As I mentioned before, I actually have a prototype here of a discrete amplifier, had some problems with ultra low frequency noise, not something you can hear but something you can measure. Couldn't remove it complete, but now have it under control. The final version basically ended up in the dac1541.... Might offer it at some point, but don't know when....
Yes, please! Think thatíd be of interest to enough folks to make it worth your while.

On another topic - any progress on a replacement U.S store?

Happy New Year!
  Reply With Quote
Old 31st December 2017, 04:29 AM   #6833
ynmichael is offline ynmichael
diyAudio Member
 
Join Date: Nov 2015
Quote:
Originally Posted by soekris View Post
The used opamps are good, a good discrete amplifier will be better but it will of course depend on how good it is....

As I mentioned before, I actually have a prototype here of a discrete amplifier, had some problems with ultra low frequency noise, not something you can hear but something you can measure. Couldn't remove it complete, but now have it under control. The final version basically ended up in the dac1541.... Might offer it at some point, but don't know when.

The dam1021 analog power is basically filtered by a CRC network, I believe it's fine, other people want other power solutions....
Also looking forward to the release of the head-amp! I would definitely get one if it can be stacked on the dam1021...would be even better if itís offered in balanced setup and easy to integrate with existing power supplies for the dam1021 (e.g. same +-9V or 12V DC supplies if thatís good enough..) ! A US distributor will certainly help as well!

I have a couple more questions:
(1) Are the ground returns for digital and analog isolated, at least with a 0R resistor? If not can you explain why it isnít needed? It would be nice to know for sure that weíre not missing out too much not using batteries...
(2) If youíre right and that the CRC filtering provides clean enough analog power rails, and I guess this is partly corroborated by the lack of improvement in providing externally regulated 5V DC directly to the vrefs, power-wise I now see a few more options:

a. Bypass main regulators and provide +-5V DC(?) directly. Based on reported experiments, this doesnít seem to power the entire board, at least the muting circuits properly. Also, please correct me if Iím wrong, Iím not sure if anyone reported meaningful improvements..
b1. Bypass the 3.3V regulator for the oscillator (and perhaps the FPGA?) and use a quality external regulator, such as NewClassD, Sparko or muzgDIY. Multiple reports of improvements can be found in this thread.
b2. Supply externally regulated (or battery powered) 3.3V DC to the rail. Only reported attempt seems to be with batteries by randytsuch (in his blog) and other battery tweakers. Not sure how this compares to the off board regulator approach.
c. Bypass 1.2V regulator for the vref switches. Again this seems to be done only with batteries and little improvement was reported.

Did I miss anything? Can I understand the power regulation onboard as:

AC/DC -> main regulators -> +-5V DC
+5V DC -> 3.3V reg -> Digitals (FPGA and oscillator) -> 1.2V reg -> vref switches
+-5V DC -> 4V reg -> vref
+-5V DC -> buffer op-amps

Where does the muting circuit fit in? Where is it supposed to draw 6V DC from?..

Thanks for your input and everyone elseís here as we advance towards the best sounding dam1021 setup thatís also easily accessible to all!
  Reply With Quote
Old 31st December 2017, 04:02 PM   #6834
ynmichael is offline ynmichael
diyAudio Member
 
Join Date: Nov 2015
It's amazing how little I actually know about the DAM DAC... There's just so much hidden in the electrical circuits and components that could affect the performance, completely hidden away from non-majors behind the pretty abstractions... Let alone the fact that we don't really know the full dam1021 schematics...

I found out after studying all 77 pages of Soekris Dam Dac - Tir Na HiFi where Nigel and friends went for (and got VERY far) the ultimate dam1021, destroying a few boards now and then of course.

First of all, correcting my last post, the power regulation on the dam1021 is more likely:
AC/DC -> Bridge rectifier -> PWR A+/-
PWR A+ -> 3.3V Regulators-> Digitals (FPGA, oscillator and microprocessor)
PWR A+ -> 1.2V Regulator -> ?
PWR A+/- -> +/-5V Regulator -> +/- 5V DC -> Op-amps -> 4V vref -> shift registers
PWR A+/- -> Output buffer op-amps
[Probably still missing some but better than last...]

With this, now I can answer my own questions...:
(1) Digital/analog separation in the ground plane, or power in general, is probably not perfect. There is no conclusive evidence but the fact that Nigel and friends believe the vref battery mod was the most significant among all the mods which they A/B tested gives some hint. The mod basically provides independent power supply to the analog circuits (with output buffer already removed...).
(2) One could go all out on battery/externally regulated power supplies for all voltages/major components on-board, as a few did. And replace the 1.2V switch-mode regulator which seems to be noisy by an LDO counterpart... Here's my current idea of a reasonable and accessible mod, if the output buffer is not needed:

i. Supply PWR A+ with 5-7V DC, replace 3.3V regulator with Sparko discrete regulator, wire the VCC on the oscillator directly to the regulator for improved performance.
ii. Replace 1.2V switch-mode regulator with LDO.
iii. Supply ~4V (such as 3.3V with LiFePo4 battery) to shift registers, bypass op-amps. If done with batteries capacitors can be left out, but if done with (here I'm very much not sure...) regulators, supercapcitors might be needed. A possible regulated supply is perhaps Regulated DC -> 4* 4V Sparko Regulators. Sparkos are rated at 3uV ripple, 1.5% voltage precision, and 1A output, which should be enough for the shift registers (??).
iv. Supply future Soekris add-on buffer/headphone amp with the same regulated DC as the Sparkos for the shift registers. Or use an existing buffer kit.
v. (optional?) Swap out certain on-board ceramics for better caps / bridge more caps at places if there's room.

This should provide a higher level of digital/analog separation, better regulated voltage for critical components, and doesn't go as far as using battery supplies. Ideally, this should only require one transformer with two sets of secondaries for digital and analog respectively. one dual-rail linear regulator for analog and one single-rail linear regulator for digital. The enclosure would also have a quality buffer/head-amp built-in.

Does this sound reasonable? If so I might consider doing it next year when I have more time... also ideally after Soren releases the head-amp board so I can stack it with the balanced dam1021's....

This is just a naÔve attempt at a decent mod... please offer critiques of any kind!

Happy New Year!
  Reply With Quote
Old 1st January 2018, 05:02 AM   #6835
ynmichael is offline ynmichael
diyAudio Member
 
Join Date: Nov 2015
As a summary to my last post, the proposed setup would be along the lines of:
Transformer AC 15V -> Dual-rail quality regulator 12V -> buffer/head-amp board
-> +/-5V DC-DC quality regulator -> onboard 5V (then 4V opamp regs -> shift reg)
[OR -> +/- 4V extremely high quality regulator -> shift regs]
-> Single-rail quality 5V regulator -> PWR A+ -> Sparko 3.3V reg to replace onboard reg (or similar) -> all digitals
-> 1.2V LDO reg to replace onboard switch mode reg -> something....

One would also need to remove the onboard buffer circuits, main filter caps, bridge rectifier and 5V regs.

The problem here without using batteries is the feasibility of providing accurate enough 4V vref and the performance of compromising and only feed +-5V dc to the onboard opamps.

The current vref + and - rails track to 0.01% according to Soren, and a 0.1% difference would result in 12dB increase in second harmonics. So providing two rails of 4V accurately enough is at least something to worry about. Could we do meaningfully better than the onboard opamp approach without LiFePO4, I.e. external voltage regs? If not, it would still be nice to know that the Soekris approach is only second to batteries.

This is somewhat beyond what I can find out with google searches effectively. Some expert input would be greatly appreciated. In any case, Iíll try to document what Iíve found out soon to provide a starting point for the more knowledgeable that may stumble upon this great DAC!

Thanks!!

P.S. as i typed this up it occurred to me that Soren said the vref is already better than any external regs.... still would be nice to have some confirmation since significant improvement was reported by using batteries to provide vref.

P.P.S. The formatting of the first part seems off... attached a screenshot of what I meant..
Attached Images
File Type: png proposed_mod.PNG (45.8 KB, 451 views)

Last edited by ynmichael; 1st January 2018 at 05:10 AM.
  Reply With Quote
Old 2nd January 2018, 04:25 PM   #6836
fedde is offline fedde  Netherlands
diyAudio Member
 
Join Date: Jul 2002
Location: Eindhoven, the Netherlands
Happy new year to everybody!

I finally finished going through the 684 (!!!) pages of this thread during my Christmas holiday!

I started reading this thread already a year or two ago at a time where the dam1021 board received still quite some complaints, but I finally decided to go for this platform after reading the much more enthusiastic recent reactions...

I currently use the MHDT Paradisea DAC (Philips TDA1545 with tube buffer) with USB connection to PC, two Ncore mono-block amps and Supravox 8" Bicone based open baffles. I would like to try a more modern DAC now. I appreciate the higher maximal sample rate, higher bit depth, lower jitter and especially the possibility for low loss digital volume control of the DAM design!

I ordered a dam1021-12 and the advised optical TORX receiver and 1:1 S/P-DIF Murata transformer. I plan to start very simple with off-my-shelf components (2x15V toroid with LM7812/7912) and optical an/or coaxial S/P-DIF.

Later on, I plan to upgrade to low-noise supplies (using a toroid with lower voltage naturally) and an XMOS 384k USB board. I may even consider to go for the multi-channel board. Long term it would be great to do a crossover using PC or onboard filtering.

Anyhow, I really look forward to try this beast out!
__________________
http://www.fedde.nu, gainclone, non-oversampling DAC and open baffle audio projects...
  Reply With Quote
Old 2nd January 2018, 04:37 PM   #6837
fedde is offline fedde  Netherlands
diyAudio Member
 
Join Date: Jul 2002
Location: Eindhoven, the Netherlands
I could imagine a workaround for loss of bit exactness due to the requirement of lower than 1 multipliers. Try a multiplier of 0.5. This would result in a (signed) right shift of exactly one bit which can be exactly compensated by applying a factor of two (more) digital gain. You will lose one LSB of intermediate internal precision, but given the long internal word lengths I expect this to have no impact. Question of course is whether this change for bit exactness would bring any audible improvement at all...

Quote:
Originally Posted by zfe View Post
@Soren
If the first step of filtering/signal processing is done by the computer, it would be nice to have the following features in a future firmware. Would this make too much trouble?

1.) For 352.8, 384 Ksps an (optional) real bypass of FIR1.
I.e. that the bits sent will be the input of FIR2 (if the volume is set to 0dB). Currently this is not the case due to the multiplicator (that can not be set 1) and possibly due to the DC blocking filter and the De-emphasis.
__________________
http://www.fedde.nu, gainclone, non-oversampling DAC and open baffle audio projects...

Last edited by fedde; 2nd January 2018 at 05:03 PM.
  Reply With Quote
Old 2nd January 2018, 05:05 PM   #6838
Rodeodave is offline Rodeodave  Austria
hates ground loops
diyAudio Member
 
Rodeodave's Avatar
 
Join Date: Dec 2004
Location: in the alps
Reference DAC Module - Discrete R-2R Sign Magnitude 24 bit 384 Khz
Hey guys, I'm a bit out of the loop. I built the dam1021 two years ago, I believe it's a rev2 board. See post #864 here: Soekris' DAC implementations

I did the Vref mod with polymer caps as can be seen in post #4603 here: Reference DAC Module - Discrete R-2R Sign Magnitude 24 bit 384 Khz

I have a rev4 from the xmas special on the way and was wondering what mods have proven worthy of implementing. From what I've read the Vref mod is not as crucial anymore, but I still have a bag of those fpcap polymers from back then, so I guess I'll install them just because.

Anything else? Maybe polymer caps at the output of the Vref generating chips? Filters? My rev2 is still equipped with the stock FW and filters, maybe I should update these as well. Like I said, I'm a bit out of the loop here...
Attached Images
File Type: jpg 15149161286711604877839.jpg (985.7 KB, 389 views)
__________________
Gravity - Making the G since 13.7 billion B.C.
  Reply With Quote
Old 2nd January 2018, 07:08 PM   #6839
zfe is offline zfe
diyAudio Member
 
Join Date: Dec 2014
Quote:
Originally Posted by fedde View Post
I could imagine a workaround for loss of bit exactness due to the requirement of lower than 1 multipliers. Try a multiplier of 0.5. This would result in a (signed) right shift of exactly one bit which can be exactly compensated by applying a factor of two (more) digital gain. You will lose one LSB of intermediate internal precision, but given the long internal word lengths I expect this to have no impact. Question of course is whether this change for bit exactness would bring any audible improvement at all...
The problem is that the gain is set in dB so you do not get an exact factor 2 this way.

For listening it probably does not matter, but for some measurements control of the actually set bits can matter.
  Reply With Quote
Old 2nd January 2018, 07:13 PM   #6840
randytsuch is offline randytsuch
diyAudio Member
 
Join Date: Jan 2003
Location: LA
Reference DAC Module - Discrete R-2R Sign Magnitude 24 bit 384 Khz
Quote:
Originally Posted by ynmichael View Post
As a summary to my last post, the proposed setup would be along the lines of:
Transformer AC 15V -> Dual-rail quality regulator 12V -> buffer/head-amp board
-> +/-5V DC-DC quality regulator -> onboard 5V (then 4V opamp regs -> shift reg)
[OR -> +/- 4V extremely high quality regulator -> shift regs]
-> Single-rail quality 5V regulator -> PWR A+ -> Sparko 3.3V reg to replace onboard reg (or similar) -> all digitals
-> 1.2V LDO reg to replace onboard switch mode reg -> something....

One would also need to remove the onboard buffer circuits, main filter caps, bridge rectifier and 5V regs.

The problem here without using batteries is the feasibility of providing accurate enough 4V vref and the performance of compromising and only feed +-5V dc to the onboard opamps.

The current vref + and - rails track to 0.01% according to Soren, and a 0.1% difference would result in 12dB increase in second harmonics. So providing two rails of 4V accurately enough is at least something to worry about. Could we do meaningfully better than the onboard opamp approach without LiFePO4, I.e. external voltage regs? If not, it would still be nice to know that the Soekris approach is only second to batteries.

This is somewhat beyond what I can find out with google searches effectively. Some expert input would be greatly appreciated. In any case, Iíll try to document what Iíve found out soon to provide a starting point for the more knowledgeable that may stumble upon this great DAC!

Thanks!!

P.S. as i typed this up it occurred to me that Soren said the vref is already better than any external regs.... still would be nice to have some confirmation since significant improvement was reported by using batteries to provide vref.

P.P.S. The formatting of the first part seems off... attached a screenshot of what I meant..
I don't know anyone that actually compared a properly implemented vref circuit to a good external voltage regulator to lifepo4 batteries. I never did, I went from the very original (flawed) vref circuit to lifepo4 and never felt the need to try anything else. IMHO, it was better, but I don't know how it would compare a modified vref circuit. I have one of the original dac's, so a vref mod was required for the best sound, it's well documented in this thread. Vref was fixed in the later revisions of this dac.

From what I remember, vref and the power to the isolated section made the most difference to sound, but obviously ymmv.
__________________
My system is here
randytsuch system
  Reply With Quote

Reply


Reference DAC Module - Discrete R-2R Sign Magnitude 24 bit 384 KhzHide 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
e18 DAC - 8 channels at 32bit /384 kHz exa065 exaDevices 30 29th June 2012 05:11 PM
384 Khz DAC? SunRa Digital Source 8 1st October 2009 11:14 PM
24 bit/192 kHz via USB? gentlevoice Everything Else 3 22nd December 2008 06:24 AM
sign magnitude DAC Bernhard Digital Source 0 30th January 2007 01:40 PM
24 bit / 192 kHz Tube DAC questions Overlord Digital Source 4 29th April 2003 05:14 PM


New To Site? Need Help?

All times are GMT. The time now is 11:37 PM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2020 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 15.00%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2020 DragonByte Technologies Ltd.
Copyright ©1999-2020 diyAudio
Wiki