oatmeal769
500+ Head-Fier
- Joined
- Mar 18, 2008
- Posts
- 511
- Likes
- 14
These are the results of experiments I've been doing lately to resolve some of my own questions about different levels of gear, price differences, and so on. I recently purchased a PS Audio DL-III and for several reasons, my curiosity then got out of hand. Sometimes I felt like it rendered D/A conversion as the smoothest cleanest most revealing sound I'd yet heard in a DAC. Other times, I wasn't so sure it was really all that much of an upgrade.
I started to wonder about the relation between cost and quality in my audio chain.
I was aware of diminishing returns in audio, but if I was to spend around ten times the amount on one DAC vs. another, I wanted it to at least be demonstrably better - Meaning I could at least be sure I and others could definitely hear the value difference. So without further rambling, I've put together -
THE GREAT "PSA DL-III with Cullen level 3 Modification DAC vs. Tianyun ZERO DAC with Burr/Brown OPA-627 OpAmps-on-a-Brown-Dog-Adapter DAC" SHOOT OUT
====================================
Equipment:
Computer
The PC used was one of my home built PC's Running MS Vista.
Spec's -
Main Board : GIGABYTE with onboard audio for the SPDIF digital output coaxial.
Processing : Intel P45 Chipset with Intel Core 2 Duo 'Conroe' Processor @ 2.4 GHz
Audio Chipset : Realtek ALC888, Azalia audio chip
Audio Driver : R-2.31Realtek Function driver for Realtek Azalia audio chip (Including Microsoft UAA Driver)
Volumes were set at 100%
Playback:
Tianyun ZERO DAC with dual OPA-627 Op-Amps on Brown-Dog Adapter, purchased new, March 2008
PS Audio DL-III with Cullen Circuits level 3 modification, purchased used, Summer 2009
Foobar2000 with the WASAPI plug-in, and all processing etc. turned off. Volumes were set at 100%
( Data is verified to be "Bit Perfect" to the input of the DAC using the familiar method of analog recording of DTS data, decoded correctly by a home theater DTS receiver. )
For complete-ness, I used the exact same power cord, outlet and coaxial cable on each DAC. Nothing else was hooked up to the DAC's other than the same output cables. These then were connected to the E-Mu 1212m soundcard at it's recording inputs. The connections used were quality, name branded cables. Coaxial cable was used to connect the Dac's to the PC's Coaxial/SPDIF output.
Capture:
Hardware based recording was done with the E-Mu 1212M PCI System
Software based file capture and editing was done using Soundforge version 9.0c and the E-Mu Patchmixer v.2
The Music:
Modest Mussorgsky - "Promenade 01" - Pictures at an exhibition [DDD] 1999
Carmen Gomes - "Let's Go Get Stoned" - Marantz High End Audiophile Test Demo SACD 1st Edition
Fantômas - "4-03-05 Megalisia" Suspended Animation, [ADD] 2005
All music is from lossless files which have been ripped error free using E.A.C. and converted to FLAC. Each file is under 30 seconds in length to avoid copyright issues.
Analysis:
RightMark Audio Analyzer V. 6.2.3
A/B/X testing was done using Foobar2000 with it's built in ABX comparator.
=============================
PART I:
The R.M.A.A. Tests
Download and unzip the folder, then click on the .htm file to see how these two measured up against each other, as well as the E-Mu 1212m I used to record the tracks.
(The folders in the .zip contain the images for the .htm files to use in the browser.) The RightMark Audio Analyzer was used to indicate measureable differences between the two DAC's.
Download the RMAA files here
and check out the results.
=============================
PART II:
The A/B/X Double Blind Tests
To do this, you will need an A/B/X tool, which will allow you to compare the samples in a double blind environment. There are several tools available like :
WinABX, and Foobar2000
Recording input was done with an E-Mu 1212m sound card. Earlier experimentation using the RealTek onboard inputs for analog recording produced non-linear and inaccurate results, so I upgraded to a dedicated studio recording card - E-Mu 1212m.
E-Mu Patchmixer software was used to set inputs to the correct analog signal levels. All other inputs and outputs were left at 100% / 0dB (unity) No gain change occured anywhere in the analog signal path.
For recording, I first played a lossless recording of a 0 dB 1 KHz sine wave through my system, the only variable being the DAC. Using this, I set the recording level at -3.50 Db, for each to recording to ensure no chance of clipping.
File recording and editing was done using Soundforge 9. The files were clipped at each end so that they were exactly the same length and time. I normalized as necessary to further ensure exactly identical volume levels after editing the clips. This did not change the file's structure or dynamics at all.
Two files pre-editing in Soundforge. Editing consisted of clipping the files to the exact starting and stopping points, and checking for exact volume matching only.
File s-t-r-e-t-c-h. The screen here spans only 2 thousandths (0.002) of a second. Resolution is precise enough to edit to the exact sample point.
=========================================
Listening
You didn't think I was going to just GIVE you MY results, did you? Besides, if I gave the answers, it wouldn't be 'double blind' would it?
Download the Audio Files here
and A/B/X them yourself.
More about how to do blind tests HERE:
I have an idea which is to post the files as unidentifiable by name (hopefully) and allow others to be the research team. I hope others sharing their own conclusions will shed more light on this topic than just me posting what I think.
I have renamed each file by color. I hope this will help preserve the objectivity. Can anyone can hear differences, and if so, how much? Is the test too easy? Too hard? which sounds 'better'?
All I ask is that IF you choose to post your results, please be honest. There's no need to skew someone else's experience by trying to prove a point. Each result should have at least 6 to 8 attempts to be considered 'statistically viable.'
Foobar2000's ABX comparator gives a nifty little log at the end, which can be copied and pasted.
If there is enough interest, I will post which DAC is responsible for each file later.
My fundamental query remains unanswered, so I'm leaving it to the reader. The PSA DL-III with Cullen mods is roughly eight to ten times the cost of the Tianyun ZERO DAC, both used and new.
Is the cost worth it? Why or why not?
I started to wonder about the relation between cost and quality in my audio chain.
I was aware of diminishing returns in audio, but if I was to spend around ten times the amount on one DAC vs. another, I wanted it to at least be demonstrably better - Meaning I could at least be sure I and others could definitely hear the value difference. So without further rambling, I've put together -
THE GREAT "PSA DL-III with Cullen level 3 Modification DAC vs. Tianyun ZERO DAC with Burr/Brown OPA-627 OpAmps-on-a-Brown-Dog-Adapter DAC" SHOOT OUT
====================================
Equipment:
Computer
The PC used was one of my home built PC's Running MS Vista.
Spec's -
Main Board : GIGABYTE with onboard audio for the SPDIF digital output coaxial.
Processing : Intel P45 Chipset with Intel Core 2 Duo 'Conroe' Processor @ 2.4 GHz
Audio Chipset : Realtek ALC888, Azalia audio chip
Audio Driver : R-2.31Realtek Function driver for Realtek Azalia audio chip (Including Microsoft UAA Driver)
Volumes were set at 100%
Playback:
Tianyun ZERO DAC with dual OPA-627 Op-Amps on Brown-Dog Adapter, purchased new, March 2008
PS Audio DL-III with Cullen Circuits level 3 modification, purchased used, Summer 2009
Foobar2000 with the WASAPI plug-in, and all processing etc. turned off. Volumes were set at 100%
( Data is verified to be "Bit Perfect" to the input of the DAC using the familiar method of analog recording of DTS data, decoded correctly by a home theater DTS receiver. )
For complete-ness, I used the exact same power cord, outlet and coaxial cable on each DAC. Nothing else was hooked up to the DAC's other than the same output cables. These then were connected to the E-Mu 1212m soundcard at it's recording inputs. The connections used were quality, name branded cables. Coaxial cable was used to connect the Dac's to the PC's Coaxial/SPDIF output.
Capture:
Hardware based recording was done with the E-Mu 1212M PCI System
Software based file capture and editing was done using Soundforge version 9.0c and the E-Mu Patchmixer v.2
The Music:
Modest Mussorgsky - "Promenade 01" - Pictures at an exhibition [DDD] 1999
Carmen Gomes - "Let's Go Get Stoned" - Marantz High End Audiophile Test Demo SACD 1st Edition
Fantômas - "4-03-05 Megalisia" Suspended Animation, [ADD] 2005
All music is from lossless files which have been ripped error free using E.A.C. and converted to FLAC. Each file is under 30 seconds in length to avoid copyright issues.
Analysis:
RightMark Audio Analyzer V. 6.2.3
A/B/X testing was done using Foobar2000 with it's built in ABX comparator.
=============================
PART I:
The R.M.A.A. Tests
Download and unzip the folder, then click on the .htm file to see how these two measured up against each other, as well as the E-Mu 1212m I used to record the tracks.
(The folders in the .zip contain the images for the .htm files to use in the browser.) The RightMark Audio Analyzer was used to indicate measureable differences between the two DAC's.
Download the RMAA files here
and check out the results.
=============================
PART II:
The A/B/X Double Blind Tests
To do this, you will need an A/B/X tool, which will allow you to compare the samples in a double blind environment. There are several tools available like :
WinABX, and Foobar2000
Recording input was done with an E-Mu 1212m sound card. Earlier experimentation using the RealTek onboard inputs for analog recording produced non-linear and inaccurate results, so I upgraded to a dedicated studio recording card - E-Mu 1212m.
E-Mu Patchmixer software was used to set inputs to the correct analog signal levels. All other inputs and outputs were left at 100% / 0dB (unity) No gain change occured anywhere in the analog signal path.
For recording, I first played a lossless recording of a 0 dB 1 KHz sine wave through my system, the only variable being the DAC. Using this, I set the recording level at -3.50 Db, for each to recording to ensure no chance of clipping.
File recording and editing was done using Soundforge 9. The files were clipped at each end so that they were exactly the same length and time. I normalized as necessary to further ensure exactly identical volume levels after editing the clips. This did not change the file's structure or dynamics at all.
Two files pre-editing in Soundforge. Editing consisted of clipping the files to the exact starting and stopping points, and checking for exact volume matching only.
File s-t-r-e-t-c-h. The screen here spans only 2 thousandths (0.002) of a second. Resolution is precise enough to edit to the exact sample point.
=========================================
Listening
You didn't think I was going to just GIVE you MY results, did you? Besides, if I gave the answers, it wouldn't be 'double blind' would it?
Download the Audio Files here
and A/B/X them yourself.
More about how to do blind tests HERE:
I have an idea which is to post the files as unidentifiable by name (hopefully) and allow others to be the research team. I hope others sharing their own conclusions will shed more light on this topic than just me posting what I think.
I have renamed each file by color. I hope this will help preserve the objectivity. Can anyone can hear differences, and if so, how much? Is the test too easy? Too hard? which sounds 'better'?
All I ask is that IF you choose to post your results, please be honest. There's no need to skew someone else's experience by trying to prove a point. Each result should have at least 6 to 8 attempts to be considered 'statistically viable.'
Foobar2000's ABX comparator gives a nifty little log at the end, which can be copied and pasted.
If there is enough interest, I will post which DAC is responsible for each file later.
My fundamental query remains unanswered, so I'm leaving it to the reader. The PSA DL-III with Cullen mods is roughly eight to ten times the cost of the Tianyun ZERO DAC, both used and new.
Is the cost worth it? Why or why not?