Filter brewing for the Soekris R2R - Page 32 - diyAudio
Go Back   Home > Forums > Source & Line > Digital Line Level
Home Forums Rules Articles diyAudio Store Gallery Wiki Blogs Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

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 6th March 2015, 04:32 AM   #311
diyAudio Member
 
Join Date: May 2005
Location: Melbourne
Blog Entries: 1
Quote:
Originally Posted by robertrowett View Post
The high point of NOS measurements is between my ears!
That isn't a measurement.
That is an extremely subjective opinion dressed with copious amounts of expectation bias and seasoned to taste with tube induced third harmonics...


Last edited by spzzzzkt; 6th March 2015 at 04:35 AM.
  Reply With Quote
Old 6th March 2015, 09:03 AM   #312
diyAudio Member
 
oneoclock's Avatar
 
Join Date: Oct 2010
Could you provide me wav of those signals? I can measure in oscilloscope or Tektronix am700. Or describe these signals so that I make with Audition.
Square wave clipping is third harmonic distorsión. I have seen different third harmonic distorsión in different filters.

I has made measurements with fuzzmeasure all filters. I can put frequency response, phase, impulse, impulse in dB and distortion of each filter with an analog measure on my audio card.

I think interesting to see distortion measurements. Varies with each filter, Some filters saturated because they increase about 10-20 db distortion of third harmonic in certain frequency bands. Also interesting impulse responses. More interesting impulse response in dB that shows the pre and pos ringing duration in milliseconds.
There are many graphs that occupy much space on the thread. Probably 80 graphs. I dont know if you interested in these graphics in thread or would be better to make a pdf document Web hosted downloadable. Can anyone give me an opinion.
I could put it all this weekend, it takes me time to capture all those screens.
  Reply With Quote
Old 6th March 2015, 09:13 AM   #313
diyAudio Member
 
oneoclock's Avatar
 
Join Date: Oct 2010
Quote:
Originally Posted by spzzzzkt View Post
That isn't a measurement.
That is an extremely subjective opinion dressed with copious amounts of expectation bias and seasoned to taste with tube induced third harmonics...
Quote:
Originally Posted by robertrowett View Post
The high point of NOS measurements is between my ears!
Filter NOS impulse response is the most perfect of all filters without pre pos ringing energy.
  Reply With Quote
Old 6th March 2015, 03:07 PM   #314
zfe is offline zfe
diyAudio Member
 
Join Date: Dec 2014
Quote:
Originally Posted by spzzzzkt View Post
zfe,
Would you do me a huge favour and take a look at a sawtooth with the stock filter and your reduced filter. It appears that the stock filter - and two of my filters that I've checked - are clipping the peak of the sawtooth with FS input and V+00, but I'd appreciate verification from someone with decent gear.
It would be my pleasure, but I am away from my test equipment till probably end of the month.

But there is no real need for measurement, I think. The done measurements (of the worst case) shows that the signal looks as predicted by the theory (within the limits of my equipment), especiallay there is no hardware induced clipping and the only hardware effect is the final analog filter (so we can calculate shape of the output signal sample per sample if needed).
... That is if we assume there are no software impementaion faluts and that we stay in the computaional savety overhead of the "3-4 bits" with the intermediate signal levels.

In this case we do not even to simuate the output, but can simulate the filter in the digital domeain only, to decide if there is clipping.

This can perhaps be done with sox (I think) as you did.

Or "by hand", especially for simple test signals. Assume the clipping free signal level is -1 to 1.
Let s[t] be the samples of your test signal, f[i] your filter taps (already multiplied by the multiplier or gain), then the output o[t] is the convolution of s and f.

As example the square wave
First without oversampling (zero insertion)

1.) A square wave of amplitude 1, longer than the filter:
So for entering the wave we are in the situation
s[t]=0 for t<=0; s[t]=1, t>0
The output at time t is the sum of the first t filter taps f[i], i.e o[t]= Sum_{i=1}^t f[i]
For exiting the square wave we are in the situation
s[t]=1 for t<=0; s[t]=0, t>0
The output at time t is the sum of the all but the first t filter taps f[i], i.e o[t]= Sum_{i=t+1}^l f[i].

Conclusion the filter does not clip with a square wave of amplitude 1, longer than the filter, if
neither the absolute value of the sum of the first k taps nor the last k taps exceeds 1 (for all k from 1 to the filter length).

For shorter square waves (with long runs of zero levels betweens the squares), the condition would be that no sum of any consecutive filter taps is greater 1 or smaller -1.

With 8x oversampling we have as new signal s[]= z[0],0,0,0,0,0,0,0,z[1],0,0,0,0,0,0,0,z[2]....
where z[] is the original signal.
Doing analog to the above reasoning, we see we get now conditions, each on filter taps with offset 8.
i.e. no sum f[i]+ f[i+8]+ ... + f[i+8j] for some i,j, is greater 1 or smaller -1.

We can handele FIR1 + FIR 2 similarly. I have not yet given a thought what happens with an intermediate IIR filter.

As I mentioned elsewhere, the condition for that a filter can not clip with ANY signal is that the sum of the absolute values of the filter taps is at most 1.
With 8x oversampling the condition that it can not clip with ANY signal is that non of the eight sums
abs(f[0])+ abs(f[8])+ ...
abs(f[1])+ abs(f[1+8])+ ...
...
abs(f[7])+ abs(f[7+8])+ ...
is greater than 1.

Triangle waves should also be no problem. I can write a little script doing these computations and giving the maximum clip-free gain for these waveforms, in the next days.

The ultimate goal would be a DAM-simulator for the digital output, taking also in account the fixed point arithmetic in the FPGA. To an certain extend this could be done, but at some point we would need implementation details of Soren.
  Reply With Quote
Old 6th March 2015, 05:23 PM   #315
diyAudio Member
 
Join Date: Jan 2002
Location: Cheltenham
Quote:
Originally Posted by zfe View Post
I stand corrected

I used the formula for charging also for discharging the capacitor.
Discharging is as U(t)= U exp(-t/(R C)).
With that we get from the data then C=1200pF.
And with that capacity we get thus as internal resistance 625 Ohm, so the contribution of the DAC ladder only.

So we do not need to think about doing the power supply better.

It is really hard to find a flaw in Sorens design or specs
Hello zfe, I'm confused here ..... what has the analog low pass filter to do with the power supply? From what I can see, there is plenty of room to optimize some of the PSU elements.
  Reply With Quote
Old 6th March 2015, 05:54 PM   #316
zfe is offline zfe
diyAudio Member
 
Join Date: Dec 2014
Quote:
Originally Posted by Spartacus View Post
Hello zfe, I'm confused here ..... what has the analog low pass filter to do with the power supply? From what I can see, there is plenty of room to optimize some of the PSU elements.
You can compute the equivalent resistance of the DAC (as voltage source) by the measurement of the charge data of the capacitor of the analog low pass filter.

My first calculation was based on a wrong capacity of the capacitor which I had found in some early post of this thread. This calculation indicated that the resistance was higher as the resistance of the DAC ladder alone. This would have meant, that the power supply (i.e. the ensamble of the final voltage regulator and the according bypasses and the shift register), feeding the DAC ladder, had a non neglectable internal resistance. If that would have been the case then it could have been useful to reduce this extra resistance to improve the impulse response of the DAC.

Measuring also the capacity showed that the assumed value was wrong and the internal resistance of the DAC is essentially only the value coming from the DAC ladder, so there is nothing to improve (resistance wise) with the power supply.

Last edited by zfe; 6th March 2015 at 06:20 PM.
  Reply With Quote
Old 6th March 2015, 07:45 PM   #317
diyAudio Member
 
oneoclock's Avatar
 
Join Date: Oct 2010
Default DAM1021 Filters

Measurements made to DAM1021 DAC with Motu Ultralite MK3. Fuzzmeasure program.
Attached Files
File Type: pdf DAM1021distor.pdf (632.7 KB, 107 views)
File Type: pdf DAM1021-impulse.pdf (615.4 KB, 84 views)
File Type: pdf DAM1021-impulse2.pdf (444.9 KB, 76 views)

Last edited by oneoclock; 6th March 2015 at 08:00 PM.
  Reply With Quote
Old 6th March 2015, 07:48 PM   #318
diyAudio Member
 
Join Date: May 2005
Location: Melbourne
Blog Entries: 1
Quote:
Originally Posted by oneoclock View Post
Filter NOS impulse response is the most perfect of all filters without pre pos ringing energy.
NOS uses no filtering so it is an oxymoron to call it the "most perfect of all filters".
You can hardly describe something as a perfect filter when it does no filtering.
  Reply With Quote
Old 6th March 2015, 07:49 PM   #319
diyAudio Member
 
Join Date: May 2005
Location: Melbourne
Blog Entries: 1
Quote:
Originally Posted by oneoclock View Post
.
These are done using the impulse test function of Fuzzmeasure?

What are the traces in the HD charts? Frequency response is fairly obvious but what about the two around -100dB range?

Last edited by spzzzzkt; 6th March 2015 at 07:57 PM.
  Reply With Quote
Old 6th March 2015, 07:54 PM   #320
diyAudio Member
 
oneoclock's Avatar
 
Join Date: Oct 2010
Quote:
Originally Posted by spzzzzkt View Post
NOS uses no filtering so it is an oxymoron to call it the "most perfect of all filters".
You can hardly describe something as a perfect filter when it does no filtering.
I refer to the different practical measures I have place in the PDF. NOS concentrating more power in less milliseconds.
  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
Soekris 's DAC implementations Eldam Digital Line Level 1035 2nd March 2017 03:01 PM
Sold: Soekris DAM1021 R2R DAC Eldam Swap Meet 6 6th February 2015 07:23 AM
24-bit R2R DAC using miltiple 16/18/20-bit R2R chips Marek Digital Source 21 1st April 2011 09:05 PM
project brewing in my head, challenging, but seem like a good way to try new concepts lemans23 Multi-Way 26 11th October 2005 12:29 PM


New To Site? Need Help?

All times are GMT. The time now is 08:00 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