Reference DAC Module - Discrete R-2R Sign Magnitude 24 bit 384 Khz - Page 316 - diyAudio
Go Back   Home > Forums > Commercial Sector > Vendor Forums > Vendor's Bazaar
Home Forums Rules Articles diyAudio Store Gallery Wiki Blogs Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

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

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 26th July 2015, 12:26 PM   #3151
alecm is offline alecm  Ukraine
diyAudio Member
 
Join Date: Feb 2011
Location: Kyiv
Quote:
Originally Posted by zfe View Post
Zooming in, shows that the main oscillation pattern repeats (twice?) with the shift register clock input (upper signal, ~24.5MHz, in second picture) but has a faster oscillation of ~ 4nS or 250MHz.
thanks for real catch! I have fq limitation of my oscilloscope and not saw it.
Now we going to the dark hole area where chip leg length has a big matter. Unfortunately, sizes of dam1021 components are big and not suitable for such fq.s and resolving problems in Mhz area are raising many other problems underway.
i have proposition to mod this. this is not ideal approach but...
resistor ladder drawing current around 5mA. So, try to connect the 3 kOhm resistor to the legs of Vcc-Gnd. possibly it will decrease the oscillation. please see the attachment.
Attached Images
File Type: png IMG_26072015_142205.png (125.8 KB, 984 views)
  Reply With Quote
Old 26th July 2015, 12:34 PM   #3152
alecm is offline alecm  Ukraine
diyAudio Member
 
Join Date: Feb 2011
Location: Kyiv
Quote:
Originally Posted by rickmcinnis View Post
Have you come to a conclusion of what the optimum amount of capacitance would be?
now i working on several new mods and waiting for certain results. I wouldn’t recommend you the optimum amount of capacitance.
i will get back later with my ideas on this...
  Reply With Quote
Old 26th July 2015, 04:07 PM   #3153
diyAudio Member
 
Join Date: Jun 2006
Quote:
Originally Posted by alecm View Post
now i working on several new mods and waiting for certain results. I wouldn’t recommend you the optimum amount of capacitance.
i will get back later with my ideas on this...
Thanks again most intrepid one!

I also wonder when one is paralleling multiple caps - can ESR get too low?

I continue to like the idea of multiple higher capacity caps at each shift register but wonder if we should consider using something other then ultra low ESR caps? Of course, I then remind myself that they are already bypassed by MLCCs but their low impedance becomes meaningless at anything other than very high frequencies.

Can't wait to see (and hear) what you come up with.
__________________
Rick McInnis
  Reply With Quote
Old 27th July 2015, 12:53 PM   #3154
DIYBras is offline DIYBras  Brazil
diyAudio Member
 
DIYBras's Avatar
 
Join Date: May 2015
Location: Parana
Default make sure to read "LATE NOTE" at end of my post

Quote:
Originally Posted by zfe View Post
Before modding I first did some measurements of the current configuration (plus some additional bypass capacity). Which lead to a remark and a request:

Firstly, it is possible to measure vref with a sound card if it supports Pro audio levels or is AC cupeled (I however take no responsibility for fried cards).
This gives you, in the audio frequency range, much more detailed information compared to the mV you get from a scope.
For example here the Vref with a white noise signal (fs=384kHz) with additional 1000uF (red) compared to the stock configuration (green).
Attachment 495514


Secondly a request to someone with a fast scope to verify the following measurements (my scope is getting old and I do not trust it entirely when pushed to the limits).
The measurements were made between ground- and Vcc-pin of the MSB shift register. Measuring Vref somewhere else works too but the "signal" is weaker.

I measured a 96kHz sine recorded with fs=384 kHz and the NOS filter (this does not realy matter, at least for the two last pictures) without applying the bandwidth limiter of the scope. And I see peeks of 100-200 mV magnitude but oscillating fast.
Attachment 495515

Zooming in, shows that the main oscillation pattern repeats (twice?) with the shift register clock input (upper signal, ~24.5MHz, in second picture) but has a faster oscillation of ~ 4nS or 250MHz.
Attachment 495516Attachment 495517
Putting a fast capacitor (Wima FKP, Mica) over the ground- and Vcc-pin tames the magnitude but the signal is still there.
The big bypass capacitor has no influence on this signal. The fast capacitor has no influence in the audio range.
Tomorrow I'll have time to play with the DAM, I'll complete some measurements as suggested by alecm but inspired in zfe measurements. As I already make my DAM mods, and I'm not willing to undo (if I forget to save some, I don't redo). With oscilloscope, probably I don't need publish... see that zfe here shows a huge ripple, more than in my own original measurements, and my mod result I no longer detect the ripple with my oscilloscope; It is well below 5 mV. So, you might just find that I turned off the oscilloscope input to impress, but it's true, try for yourself and see if in doubt.
I've realized the FFT in 4V line, with interesting results. I hope I have saved the before and the after ... I tend to be very disorganized in my DIY measurements saving tasks... When I return to the lab/home, I publish some measurements (finally).

LATE NOTE: I noticed that the return of original feedback 4V supply resistor in DAM comes a point after one of the shift registers, which makes the opamp amplify the noise that it induces, like an remote DC sensing. Maybe the ceramic cap is barely working here... My 47μF capacitor mod takes the signal direct from the "official" output of the regulator, totally bypassing the noise becoming from the "680" resistor, and this prevents undesirable voltage drops that can be amplified by the supply. I suspect this is what caused the biggest difference, because to me made no sense THAT huge ripple size with such low drain circuit like this. Soon I'll get some pictures of the DAM and try to explain my suspicion.

Last edited by DIYBras; 27th July 2015 at 01:08 PM.
  Reply With Quote
Old 27th July 2015, 01:18 PM   #3155
DIYBras is offline DIYBras  Brazil
diyAudio Member
 
DIYBras's Avatar
 
Join Date: May 2015
Location: Parana
Figures:

DAMexp4 is my "suspicious" (buffer contamination)

DAMexp3 is my actual 47µF mod

Is the not most correct mod but helps to explain some things

The Soekris mod also make the "680" res inoperative with low impedance jumpers in cap position so will have even better results, probably...
Attached Images
File Type: png DAM exp4.png (45.6 KB, 860 views)
File Type: png DAM exp3.png (134.5 KB, 856 views)
  Reply With Quote
Old 27th July 2015, 07:41 PM   #3156
zfe is offline zfe
diyAudio Member
 
Join Date: Dec 2014
Some "sound-card measurements" of Vref (4Vpp would be -7.5dBFS).

First, to compare, Vref when playing white noise in the stock config (green) and stock plus 1000uF (red).
wnoise1000mu over wnoise.png

Next the "factory mod" and variants:
The red curve is Vref when playing white noise, the green one (if nothing else said) Vref when no signal played (which coincides more or less with the noise floor).

1. a) factory mod, b) factory mod +1000uF, c) factory mod +6800uF (over 1000uF green), d) factory mod +17000uF (over 6800uF green)
1a Mod wnoise over nosig.png1b mod wnoise1000u over nosig.png1c mod wn 6800u over 1000u.png1d mod wn 17000u over 6800.png

2. A variant of the factory mod using 1 Ohm as load resistor instead 0.1 Ohm
a) 1 Ohm factory mod, b) 1 Ohm factory mod +1000uF, c) 1 Ohm factory mod +6800uF
2a Mod 1O wn over nosig.png2b mod 1O wn 1000u over nosig.png2c mod 1O wn 6800u over nosig.png

P.S. the high frequency behavior on the oscilloscope does not changed due to the factory mod. But as I said, I am mot sure if this not an problem of my oscilloscope. The high frequency behavior needs to be verified by independent measurements.
  Reply With Quote
Old 27th July 2015, 09:19 PM   #3157
zfe is offline zfe
diyAudio Member
 
Join Date: Dec 2014
And "sound-card measurements" of Vref with a 1kHz signal playing:

First a) stock (red) over no signal (green), b ) stock+1000uF (red) over stock (green), c) stock+6800uF (red) over stock+1000uF (green)
3a 1k over nosig.png3b 1k 1000mu over 1k.png3c 1k6800mu over 1k1000mu.png

Then a) factory mod (red) over no signal (green), b) factory mod+1000uF (red) over factory mod (green), c) factory mod+6800uF (red) over factory mod+1000uF (green), d) factory mod+17000uF (red) over factory mod+6800uF (green)
4a mod 1k over nosig.png4b mod 1k 1000u over 1k.png4c mod 1k 6800u over 1000u.png4d mod 1k 17000u over 6800u.png
  Reply With Quote
Old 27th July 2015, 09:43 PM   #3158
alecm is offline alecm  Ukraine
diyAudio Member
 
Join Date: Feb 2011
Location: Kyiv
Quote:
Originally Posted by zfe View Post
And "sound-card measurements" of Vref with a 1kHz signal playing:
many thanks for inputs!
You giving us lots of food for reflection.



Please try to check the interchannel crosstalk - send a signal to one channel and check spectral recording in the different one
__________________
Soekris DAM 1021 MODS #1,#2,#3 and multi-channel Soekris project
  Reply With Quote
Old 28th July 2015, 02:10 AM   #3159
diyAudio Member
 
Join Date: Aug 2012
Quote:
Originally Posted by rickmcinnis View Post
Thanks, alecm,

Trying to find out where the boundaries are.

Have you come to a conclusion of what the optimum amount of capacitance would be?

I was not sure if all of the shift register's bypass caps are in parallel or just feeding the individual shift registers. I did wonder if that might put quite a strain on the regulator at start up if they were. And I do not know if that would make any difference, anyway. Haven't done anything yet.

Bought 0.1R resistors with the conductive strip going in the wrong direction! Had to order the correct ones. So now I have removed the components and will have to wait until the correct resistors arrive on Tuesday to finish this.

Do you have an opinion on HIFIDUINO's idea of caps at each shift register? Seems like a good idea to me not having to have the extra caps stacked at the regulator. "Good practice" seems to be to have the capacitance at the load instead of at the output of the regulator so this seems to be the best way.

Now if I could figure out what would be the best capacitors to use.

There are those who think the polymers are suspect in the "analogue" part of a DAC and I have no idea if a shift register could be construed as being analogue in any way even though it is directing the resistor stack which does seem to be "analogue". So I wonder if one should use a conventional "audio" cap there - like the SILMICS or any of the other highly regarded electrolytics recommended for use in power supplies for analogue circuits.

Needless to say, your opinion would be welcomed!

One thing after another ...

After posting I see zfe's post - very interesting but discouraging. I guess perfection will not be achieved? Bittersweet realization. Again!
silmics have very distinct signature like very rounded sound and unless and untill the system is very harsh silmics sound blurred and lost attack and resolution even after 500hours of burning.
As a decoupling caps in my experience over the past 3 years i never found silmic sound with great resolution and in fact its overly rounded sound even wit mix and match wit nichicon FG or muse they still sound artificial.
  Reply With Quote
Old 28th July 2015, 02:14 AM   #3160
diyAudio Member
 
Join Date: Aug 2012
Quote:
Originally Posted by zfe View Post
And "sound-card measurements" of Vref with a 1kHz signal playing:

First a) stock (red) over no signal (green), b ) stock+1000uF (red) over stock (green), c) stock+6800uF (red) over stock+1000uF (green)
Attachment 495821Attachment 495822Attachment 495823

Then a) factory mod (red) over no signal (green), b) factory mod+1000uF (red) over factory mod (green), c) factory mod+6800uF (red) over factory mod+1000uF (green), d) factory mod+17000uF (red) over factory mod+6800uF (green)
Attachment 495824Attachment 495825Attachment 495826Attachment 495827
why there is a strong peak of 50Hz?
  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
e18 DAC - 8 channels at 32bit /384 kHz exa065 exaDevices 30 29th June 2012 06:11 PM
384 Khz DAC? SunRa Digital Source 8 2nd October 2009 12:14 AM
24 bit/192 kHz via USB? gentlevoice Everything Else 3 22nd December 2008 07:24 AM
sign magnitude DAC Bernhard Digital Source 0 30th January 2007 02:40 PM
24 bit / 192 kHz Tube DAC questions Overlord Digital Source 4 29th April 2003 06:14 PM


New To Site? Need Help?

All times are GMT. The time now is 07:48 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright ©1999-2017 diyAudio

Content Relevant URLs by vBSEO 3.3.2
Wiki