The problem lies in the 'expectation' of how human hearing is tuned to sound perception and how the digitized signal is able to deliver the transients of analog sound.
Pre and post ringing obviously changes the way of how sound is expected to be perceived naturally. Although the frequency lies beyond the audible range the envelope can be perceived. This part is true for every ADC/DAC .
If I am not completely wrong, the filter-brewing thread tries to find the best way of how to combine certain parameters to the best sounding filter. And the final test for each filter is always up to which degree it gives the listener the feeling of being 'real', meeting the underlying expectation of how we are used to hear in the real world.
That we can try different filters (and that there are some nice people who do this job for all of us!) and come step by step closer to something that sounds 'real' to our ears is the biggest advantage of this R2R DAC. That outweighs by far theoretical calculations and THD numbers in my opinion.
Linear phase filters have no time smearing, and have been a part of your humble CD player since the beginning of time.
Yeah, back when there was only one sample rate to worry about, most constructive reply to the whole thread here folks....
Yeah, back when there was only one sample rate to worry about, most constructive reply to the whole thread here folks....
Digital filters don't care about sample rate. Run the sample rate 10% higher, and the filter corner moves up 10%.
I found an example where the linearity error of a DAC is corrected by using the low order bits. It is done dynamically, by digital feedback, so capturing also the thermal and other errors. The DAC are two 16-bit hybrid R2R DACs (with a "thermometer section") with 8 bits overlap, so resulting in a 24-bit DAC. The achieved linearity is 20 bits. The specs are not for audio but for lab purposes, so to be taken serious
There is a very extensive and instructive discussion about how to verify the accuracy and of possible sources of errors/noise.
Linear Technology Application Note 86
There is a very extensive and instructive discussion about how to verify the accuracy and of possible sources of errors/noise.
Linear Technology Application Note 86
This is so not true. Digital filters, no matter linear, intermediate or minimal phase results with ringing (pre, post) which energy is stolen from impulse energy which results with lower amplitude of orginal impulse. So impulse response of DF is smeared over time.
An externally hosted image should be here but it was not working when we last tested it.
Ringing False: Digital Audio's Ubiquitous Filter Page 2 | Stereophile.com
Here is a simple linearity test I did some time ago...
It's just a simple proposal and the levels have to be taken carefully...
Hp
Attachments
With all due respect, you fellows have gone quite far off topic. You have, repeatedly, made a point about how you feel about theoretical THD levels in this DAC vs others. Fine. Most of us, however, just don't care. We're interested in _this_ DAC, and know how to search if we care to read about the 1541A or the 1704. No need to, repeatedly, denigrate the work of Soekris or to proselytize for other chips.
Last edited:
I said several times, but since it was not enough I say again:
I have nothing to denigrate, I have no commercial interest (Soekris has), I'm on a diy forum to debate about audio device, including technical choices.
My replies following what other members claim... for example
someone claims: accuracy doesn't matter...
well, I reply: then ask for a 10% resistors tolerance, they are very cheap, so you save money
someone claims: time is more important..
well, I reply: get a good master clock, not a noisy SI5xxx
someone claims: thanks of a "sliding window", magically accuracy grows....
I reply: not true, accuracy is always 13-14 bit with 0.01% resistors tolerance
Debating around technical question should be the core of this forum.
Finally, I say again:
everyone is free to buy this dac, I have never suggested to not buy this commercial product (nothing to do with diy).
Funny how people read things.
From looking at the previous posts -
NO ONE said accuracy does not matter (well now you said it)
NO ONE said timing was more important (they just referenced an article) (again, you're the one claiming it now)
Calculations and measurements already show the 0.05% version is able to deliver the 14-15 bits. If anything LordByron has just made it perfectly clear it's about the luck of the draw on how all of those things pan out.
From what you were arguing about. It's already proven the DAC has a dynamic range of 108dB... a noise floor of about 120dB and THD+N of 84dB (using the terrible 0.37%)... quite different from the 14-bit DAC picture that you're continually yammering about.
If you can agree with the above, let's just leave it as the above and let everyone else judge the product. Continually pushing your point on the accuracy isn't doing you or the thread any favors.
Edit: It seems Soren has updated his main site with measurements. In case others have missed it like me:
dam1021 | -01 | -02 | -05
THD @ -1db | 0.005% | 0.008% | 0.015%
THD @ -60dB | 0.03% | 0.05% | 0.09%
S/N 20 Khz Bandwith 127 dB unweighted
From looking at the previous posts -
NO ONE said accuracy does not matter (well now you said it)
NO ONE said timing was more important (they just referenced an article) (again, you're the one claiming it now)
Calculations and measurements already show the 0.05% version is able to deliver the 14-15 bits. If anything LordByron has just made it perfectly clear it's about the luck of the draw on how all of those things pan out.
From what you were arguing about. It's already proven the DAC has a dynamic range of 108dB... a noise floor of about 120dB and THD+N of 84dB (using the terrible 0.37%)... quite different from the 14-bit DAC picture that you're continually yammering about.
If you can agree with the above, let's just leave it as the above and let everyone else judge the product. Continually pushing your point on the accuracy isn't doing you or the thread any favors.
Edit: It seems Soren has updated his main site with measurements. In case others have missed it like me:
dam1021 | -01 | -02 | -05
THD @ -1db | 0.005% | 0.008% | 0.015%
THD @ -60dB | 0.03% | 0.05% | 0.09%
S/N 20 Khz Bandwith 127 dB unweighted
Last edited:
I claim accuracy and time are both important. So, IMHO, 13 bit accuracy is not a great precision (accuracy) and SI5xx is not low jitter oscillator (time), see datasheet.
Said several times.
Mathematically speaking 0.05% resistor tolerance is 11 bit accuracy: 1/2 ^11 * 100 = 0.0488%.
Statistically speaking you could add 1 or 2 more bit (error distribution), but you forget:
- resistance of the switches
- thermal drift
- double ladder network that means double error.
-84 db, 0.006%, exactly 14 bit: 1/2^14 * 100 = 0.0061%.
Said, said and said again (see post #2417)
So, I cannot agree.
But obviously I let everyone to judge the product. Never forced someone to judge the product with my point of view.
Or maybe I have to judge the product with your point of view?
How convenient... You're basically making stuff up in your head to get your points across.
You decide to change what you stated before. You mentioned some imaginary person contradicted you before to push your point. And you've conveniently dropped how, according to you again, someone said accuracy isn't important... to push your point. For the record, I agree with you both time and accuracy are important.
It's more like you can't see beyond your ego.
I've put in those figures because those are what we all can agree on. So 84dB because of 0.006% THD+N (not because I don't think your point is valid - try that for a change). That's exactly what the product measured with the measurements on the first page. In my previous post I've posted updated measurements from Soren... so this is in fact no longer accurate.
So far, you're cherry-picking issues you want to answer to. You haven't acknowledged that the product measures better than your stated 14-bit DAC. You haven't addressed the point raised by Marek in post #2457
At -60dBFS the THD+N of your favored "17-bit precise" PCM1704 is at 0.6% and 0.9% and the DAM1021 is a "crappy" 0.37%... so how crappy is the precision again?
So the question with regard to discussion and debate... can you move beyond your own personal ego, or should I say, agenda?
Again, I stated the specs based on measurements and calculations from LordByron - those are factual data points not something from my point of view. If you want to perceive that as needing to see it from my point of view, that's just sad... because it's your own head talking.
If anything, my point of view is per my previous post. State the measurements that you think will make it fair for people to judge (which I believe I've already summarized) and stop the discussion of building other DACs in this thread as it's not doing you, or the thread, any good. Look at how many people have asked you to stop repeating already.
Last edited:
andrea_mori,
you can't judge DAC SQ utill you try it by yourself. Your opinion about TDA1541A gloried by you many times in this thread was probably made AFTER that you listened to? Or you posted 100 posts that this is worse DAC than ex. AD1865 because it has 2 bit lower resolution, and for sure worse than PCM1704 and all modern S-D because it has far worse specs?
Guys who risked buying DAM (no one knew how it really sounds before) now reports how good it is (for me it is something I didn't hear before although I have heard many DACs) and you enter here and you cultivate some theorizing.
Like with TDA1541A take a risk, buy it, judge it and come back with your results. No one will blame you if you make here a negative opinion AFTER you really hear it! If you dont like it - sell it on swap meet! Anyway this is not toy for a few thousands $$..
Otherwise stop this thread littering - every one interested in this thread GOT IT that DAM has about 14bit effective *MONOTONICITY* but it seems that you can't really get it that despite this fact it sounds incredible.
you can't judge DAC SQ utill you try it by yourself. Your opinion about TDA1541A gloried by you many times in this thread was probably made AFTER that you listened to? Or you posted 100 posts that this is worse DAC than ex. AD1865 because it has 2 bit lower resolution, and for sure worse than PCM1704 and all modern S-D because it has far worse specs?
Guys who risked buying DAM (no one knew how it really sounds before) now reports how good it is (for me it is something I didn't hear before although I have heard many DACs) and you enter here and you cultivate some theorizing.
Like with TDA1541A take a risk, buy it, judge it and come back with your results. No one will blame you if you make here a negative opinion AFTER you really hear it! If you dont like it - sell it on swap meet! Anyway this is not toy for a few thousands $$..
Otherwise stop this thread littering - every one interested in this thread GOT IT that DAM has about 14bit effective *MONOTONICITY* but it seems that you can't really get it that despite this fact it sounds incredible.
Last edited: