DAC recommendation

Jump to Latest
EUVL

EUVL

Member
Joined 2003
xen-audio.com
This is your definition of quiet ?

Murata 7805SR.jpg





Patrick
 
EUVL

EUVL

Member
Joined 2003
xen-audio.com
M

Markw4

Member
Joined 2016
audiostar said:
To those of you who have abraxalito dacs, have you compared the abraxalito to commercial dacs some that cost more than $5000?
Click to expand...
I have one of his early dacs. Its pretty good with a good power supply. Of course, I can plainly hear that its a 16-bit dac. Have I heard better dacs that cost less than $5,000? Of course. But isn't the reason for using an abraxalito dac mostly because they sound surprisingly good for being so low cost? They're a great value. For people who have $5k to spend on a dac, they might find a Holo Spring 3 to be quite good for the money.
 
Last edited:
rhthatcher

rhthatcher

Member
Joined 2018
Paid Member
Between this thread and this other one
https://www.diyaudio.com/community/...ra-analog-d20400a-who-have-heard-both.347298/ I decided I want to try the Pass D1 and another classic PCM63 based DAC. I found Theta GenV for a good price, and also placed a WTB ad for the Pass D1. After some time I found one and the price seemed fair. I also looked at what else from the era had the PCM63: Wadia 15, EAD, etc.

Once the D1 arrived I did some A/B testing. It needs to be stated that both are killer sounding. Theta has higher voltage out. In the end, the Pass won, and I sold the Theta this week. If I never got a response for my Pass WTB ad I could have been very happy keeping the Theta.

Maybe one day I'll try my hand a DIY on a DAC, but there are too many other projects on my plate already.
 
abraxalito

abraxalito

Member
Joined 2007
kazap said:
I dont know what to listen for to tell a 16bit DAC from a "24bit" DAC when using a preamp with no digital attenuation. How do you tell?
Click to expand...

Normally its the noise floor - the DAC @Markw4 listened to was a single TDA1387 which has baseline noise around -95dB. The noise gets better with paralleling multiple chips even though its still just a 16bit DAC. Even though when playing music the noise may not be directly audible, its presence masks low-level details so the subjective effect is that resolution is lacking.
 
  • Like
  • Thank You
Reactions: 1 users
kazap

kazap

Member
Joined 2011
abraxalito said:
Even though when playing music the noise may not be directly audible, its presence masks low-level details so the subjective effect is that resolution is lacking.
Click to expand...

Please indulge me with some speculation with test conditions for unmasking low level detail loss, from inadequate bit depth?

Assume the test file has gone through Replay Gain or even better PGGB for full signal encoding free of any over shoots and there's a preamp for gain control.

Im just guessing but say:
  • A subject has great ears and can actually hear a 15dB signal in a special listening room eg a sound proofed basement with a super low noise floor of 20dB
  • The system goes very loud with 110dB clean peaks at the listening position (recommended for less then two minutes before hearing damage occurs but the subject astonishingly hasn't gone deaf yet)
Thats a 95dB dynamic range. Isn't that within 16bit encoding and a 16bit DAC capability?

I must be missing something here. Please help :(
 
abraxalito

abraxalito

Member
Joined 2007
I don't believe 95dB DNR does fit within a 16bit wide signal chain, I think with the correct dither the number's closer to 92dB. But if you're hinting at something like 'We shouldn't be able to hear this difference' then I'm with you. I don't understand why reducing the noise within a 16bit DAC below -92dB (which I take to be the best case noisefloor on a 16bit recording) has an audible effect, given that there's no way my system is able to play at 110dBA peaks. But this is rather going off topic so I'll not speculate further about reasons for this.
 
  • Thank You
Reactions: 1 user
juanitox

juanitox

Member
Joined 2006
i have lost a thousands of money searching the perfect dac during the last 20 years , ( Weiss , Wadia , Pass D1 , EAD , holo , Berkeley , Burl , T+A DSD , emprical , Lessloss , altman , TotalDAc , Metrum , EC design , JADIS , Chords , etc....) running in circle during years to finally accept that the sound i like is the sound of the 90' (Cello reference Dac , Wadia 9 , PAss D1 , EAD DSP 9000 ) so i end up and will die with an Old French PCM63 based Dac at 500$ and can"t be more happy . when it's good it"s good , to get better is an illusion costing time and money same for speakers , amps and women :cool:
 
  • Like
Reactions: 1 users
Rewind

Rewind

Member
Joined 2009
Markw4 said:
Had a low end Denafrips here for a few days that a friend brought over. Wasn't impressed compared to Topping D90. I would say that of the commercial dacs that have been auditioned at my place, D90 was best and was preferred over some dacs costing 3+ times more. However, my custom AK4499 dac has so far been preferred by all listeners over D90 and all the other auditioned dacs. Haven't had a high-end Chord or a Holo May here yet, although it would be great if that could happen sometime.
Click to expand...
Well, tell us about the custom AK4499 dac. I think most of us would want to build a custom Dac instead of supporting The Regime by buying chifi.
 
M

Markw4

Member
Joined 2016
kazap said:
Im just guessing but say:
  • A subject has great ears and can actually hear a 15dB signal in a special listening room eg a sound proofed basement with a super low noise floor of 20dB
  • The system goes very loud with 110dB clean peaks at the listening position (recommended for less then two minutes before hearing damage occurs but the subject astonishingly hasn't gone deaf yet)
Thats a 95dB dynamic range. Isn't that within 16bit encoding and a 16bit DAC capability?

I must be missing something here. Please help :(
Click to expand...
The type of calculation you are trying to do is mistaken. First of all, 'thresholds of audibility' are not hard limits, they are estimates of an average for a population. In other words, 50% of people should be able to hear below the threshold. Also, I would say that whether or not a low level signal is audible depends greatly on whether or not it is strongly correlated with a higher level signal one is focused on listening to. Thus a brass band may be hidden at -60dB below some other music and be completely inaudible. Yet failure to dither a CD before truncation leaves quantizing distortion at -93dBFS. It is uncontroversial that some people can hear that distortion. Yes, not all of music is constantly at 0dBFS, but the distortion that can be heard is audible to at least some people at normal listening levels. The difference between that distortion and the brass band is the degree of correlation with the louder signal being listened to. This type of phenomenon has been found to hold at levels well below -93dBFS. It doesn't even have to be a distortion, it can be noise that is correlated with the audio signal being listened to.

The other thing with trying to make a calculation is that hearing is not linear, time-invariant, nor stationary. Therefore you can't just add up dB numbers and get a correct result. That sort of calculation implicitly assumes that hearing must be no more than weakly nonlinear, which is not the case. So far as I can tell what people can or can't hear is very dependent on the exact time-domain waveform (as well as correlation). It means typical audio FFTs which discard phase information are no more than a pretty rough estimate of what someone might be able to hear. Maybe listen to the test files at: https://purifi-audio.com/2019/12/07/amfm/ which have the exact same audio FFT spectra. Do they sound the same?

Typically when trying to measure numbers like thresholds of audibility, test signals are used that the test subject is not expertly familiar with. Clicks, noise burst, sine waves, a few sine waves sat once, etc. OTOH if you take a trained percussionist who, say, is an expert at setting up and tuning snare drums, that person can possibly hear each rattle of the snares against the bottom head. Listening to a complex sound one knows so well makes it much easier to hear the when details are distorted or missing. Also, in modern perceptual testing, it is considered proper to train the test subject to progressively hear smaller and smaller changes in a sound until that person's limit is found. However long it takes to complete that training and development of ability is how long a researcher has to spend on each test subject. IIUC that's not the way they did it back in the day when a lot of psychoacoustics research was done. The new field that continues to study human hearing is now called 'auditory scene analyisis.' That's probably the best place to check for more up to date research. https://www.frontiersin.org/articles/10.3389/fnins.2016.00524/full
 
Last edited:
  • Like
Reactions: 1 user
You must log in or register to reply here.
Top Bottom