AMB Ɣ2 technical measurements

Discussion in 'Source Measurements' started by atomicbob, Feb 6, 2022.

  1. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,652
    Trophy Points:
    113
    Location:
    On planet
    amb Ɣ2 technical measurements

    Standard Prologue
    If you are unfamiliar with audio measurements please use a search engine with the query:
    "audio measurements" or "audio measurement handbook"
    Look for publications by Richard C. Cabot and also by Bob Metzler, both from Audio Precision. There are other useful publications as well. These will provide basic knowledge.
    Interpretation of the following measurements is beyond the scope of technical measurements posts.

    The data presented were collected as follows:
    1. PrismSound dScope III, picoscope 5243B
    2. Tecnec 75R spdif cable
    3. Single Ended cables Mogami 2964 1 meter with Amphenol RCA connectors
    4. 100 Kohm load used for measurements
    5. dScope analyzer sample rate 48 KHz unless otherwise noted
    6. DAC 44.1 KHz sample rate, 24 bit depth unless otherwise noted
    7. Audioquest Forest and Schiit Pyst USB cables
    8. Vaunix Lab Brick USB hub
    9. Shielded 14AWG and 16AWG power cables
    10. A filter mode unless otherwise noted
    11. Soft clip ON unless otherwise noted

    Measurements are made in accordance with AES17:2015

    Measurements for record commenced after warmup of at least 48 hours.
    Measurements were performed over a period of several months.

    Index
    Post 1 - measurement setup description, highlights
    Post 2 - spdif input - SE RCA outputs
    Post 3 - USB ASIO input - SE RCA outputs
    Post 4 - Filter response part 1 - step response (square wave)
    Post 5 - Filter response part 2 - Imaging
    Post 6 - Anticlip feature
    Post 7 - External Power Supply
    Post 8 - Comparison with ODAC (Original Generation)
    Post 9 - reserved for additional data and corrections

    Notable highlights:
    Ɣ2 is one of the better WM8741 DAC implementations.
    spdif performance is decidedly better than USB.
    Very low distortion.
    Power supply noise spectrum is notably low for a SE output at -118 dBFS with spdif and external power supply.
    SE output Dynamic Range of 121 dB.
    SE output Cross-talk is approaching dual mono with > 125 dB isolation.
    Very low jitter, especially with spdif input.

    The AMB Ɣ2 preceded the ODAC by three years. Considering how NwAvGuy denigrated other DACs, the Ɣ2 embarrasses the ODAC both technically and in listening evaluations. See post 8 below for details.

    2009 AMB Ɣ2 introduced
    2012 ODAC introduced

    amb Ɣ2:
    DSC_0630_small.jpg

    Measurement setup example:
    DSC_0631_small.jpg

    Listening evaluation example:
    DSC_0632_small.jpg

    Ɣ2 is eminently listenable. USB implementation is dated, however spdif + external power supply allow the Ɣ2 to perform its best. A far better auditory experience than the Original Generation ODAC.
     
    • Like Like x 13
    • Epic Epic x 4
    • List
    Last edited: Feb 6, 2022
  2. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,652
    Trophy Points:
    113
    Location:
    On planet
    spdif input - SE RCA outputs

    Ɣ2 Dynamic Range spdif input SE output
    20210621 gamma2 dynamic range FFT spdif SE.png
    Excellent performance, especially considering SE outputs, size and price of this DAC.

    Dynamic range, in simplified terms
    1. find maximum output voltage at 0 dBFS
    2. find residual noise with a -60 dBFS 1 KHz stimulus, removed from analysis by window notch
    3. Dynamic Range is the difference between maximum output and residual noise
    Specifics are in AES17 section 9.3 (measurebators do your homework please)
    Example here:
    https://www.maximintegrated.com/en/design/blog/spec-dynamic-range.html

    Why we have a -60 dBFS stimulus
    Some clever codec designers include an output gate that shuts off when there is no signal present. This has the dual benefit of lowering output idle noise while also gaming the measurements. spdif became wise requiring a -60 dBFS stimulus so any such gates are open during measurement and true Dynamic Range performance measured.

    Ɣ2 A04 THD+N THD nth-HD FFT spdif input SE output
    20210621 gamma2 A04 THD+N THD nth-HD 4+HD+N 60Hz FFT SE 0dBFS - spdif.png

    Ɣ2 50 + 7000 Hz spdif input SE output - Left Channel
    FFT_2_L1T13_1_A.png

    Ɣ2 Gain Linearity spdif input SE output - Left Channel
    G2_L1T3_1_A.png

    Ɣ2 THD+N vs Frequency spdif input SE output - Left Channel
    G2_L1T5_1_A.png

    Ɣ2 Residual Noise SE spdif input SE output
    20210621 gamma2 residual noise FFT spdif SE - 180 dB range.png

    Ɣ2 Inferred Jitter spdif input SE output
    20210621 gamma2 inferred jitter spdif SE.png
    This is an exceptionally clean jitter measurement!

    Complete spdif input SE output analysis report pdf attached
     

    Attached Files:

    • Like Like x 6
    • Epic Epic x 1
    • List
    Last edited: Feb 6, 2022
  3. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,652
    Trophy Points:
    113
    Location:
    On planet
    USB ASIO input - SE RCA outputs

    Ɣ2 Dynamic Range ASIO input SE output
    20210623 gamma2 dynamic range FFT ASIO SE t2.png

    Ɣ2 A04 THD+N THD nth-HD FFT ASIO input SE output
    20210623 gamma2 A04 THD+N THD nth-HD 4+HD+N 60Hz FFT SE 0dBFS - ASIO.png

    Ɣ2 50 + 7000 Hz ASIO input SE output - Left Channel
    FFT_2_L1T13_1_A.png

    Ɣ2 Gain Linearity ASIO input SE output - Left Channel
    G2_L1T3_1_A.png

    Ɣ2 THD+N vs Frequency ASIO input SE output - Left Channel
    G2_L1T5_1_A.png

    Ɣ2 Residual Noise ASIO input SE output
    20210623 gamma2 residual noise FFT ASIO SE - 160 dB range.png

    Ɣ2 Inferred Jitter ASIO input SE output
    20210623 gamma2 inferred jitter ASIO SE.png
    Again a very clean jitter response, especially from USB power!!

    Complete ASIO input SE output analysis report pdf attached
     

    Attached Files:

    • Like Like x 6
    • Epic Epic x 2
    • List
    Last edited: Feb 6, 2022
  4. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,652
    Trophy Points:
    113
    Location:
    On planet
    Filter response part 1 - step response (square wave)

    Measurement conditions
    1. Ɣ2 gain set to A, B or C filter as noted
    2. 44.1 KHz sample rate except where noted
    3. USB ASIO input
    4. SE output to oscilloscope
    5. Ɣ2 anticlip on

    Filter step responses are independent of inputs and outputs.

    filter A Linear phase, slow rolloff
    filter B Minimum phase, slow rolloff
    filter C Linear phase, brickwall

    Ɣ2 20Hz square -3dBFS ASIO input SE output - Filter A
    01 20210624 SigGen gamma2 20 Hz square -3dBFS 100uS div 5MHz filter ASIO SE - fltr A.png

    Ɣ2 20Hz square -3dBFS ASIO input SE output - Filter B
    02 20210624 SigGen gamma2 20 Hz square -3dBFS 100uS div 5MHz filter ASIO SE - fltr B.png

    Ɣ2 20Hz square -3dBFS ASIO input SE output - Filter C
    03 20210624 SigGen gamma2 20 Hz square -3dBFS 100uS div 5MHz filter ASIO SE - fltr C.png



    At 0dBFS ringing is softly clipped:

    Ɣ2 20Hz square 0dBFS ASIO input SE output - Filter A
    04 20210624 SigGen gamma2 20 Hz square 0dBFS 100uS div 5MHz filter ASIO SE - fltr A.png

    Ɣ2 20Hz square 0dBFS ASIO input SE output - Filter B
    05 20210624 SigGen gamma2 20 Hz square 0dBFS 100uS div 5MHz filter ASIO SE - fltr B.png

    Ɣ2 20Hz square 0dBFS ASIO input SE output - Filter C
    06 20210624 SigGen gamma2 20 Hz square 0dBFS 100uS div 5MHz filter ASIO SE - fltr C.png
     
    • Like Like x 4
    • Epic Epic x 1
    • List
    Last edited: Feb 6, 2022
  5. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,652
    Trophy Points:
    113
    Location:
    On planet
    Filter response part 2 - Imaging

    Measurement conditions
    1. Ɣ2 filter set to A, B or C filter as noted
    2. dScope signal generator set to -6 dBFS white noise
    3. 44.1 KHz sample rate except where noted
    4. spdif or USB ASIO input
    5. SE output to dScope
    6. Ɣ2 anticlip on

    filter A Linear phase, slow rolloff
    filter B Minimum phase, slow rolloff
    filter C Linear phase, brickwall

    Ɣ2 imaging spdif 44K input SE output - Filter A

    01 20210624 gamma2 imaging 44KHz spdif SE - fltr A.png

    Ɣ2 imaging spdif 44K input SE output - Filter B
    02 20210624 gamma2 imaging 44KHz spdif SE - fltr B.png

    Ɣ2 imaging spdif 44K input SE output - Filter C
    03 20210624 gamma2 imaging 44KHz spdif SE - fltr C.png


    Ɣ2 imaging ASIO 44K input SE output - Filter A
    04 20210624 gamma2 imaging 44KHz ASIO SE - fltr A.png

    Ɣ2 imaging ASIO 44K input SE output - Filter B
    05 20210624 gamma2 imaging 44KHz ASIO SE - fltr B.png

    Ɣ2 imaging ASIO 44K input SE output - Filter C
    06 20210624 gamma2 imaging 44KHz ASIO SE - fltr C.png
     
    • Like Like x 3
    • Epic Epic x 1
    • List
    Last edited: Feb 6, 2022
  6. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,652
    Trophy Points:
    113
    Location:
    On planet
    Anticlip feature

    Ɣ2 anticlip appears to reduce the output by 2 dB.
    It does not appear to influence the soft clipping of filter ringing.

    Ɣ2 SigGen gamma2 440 Hz sin 0dBFS 500uS div 5MHz filter ASIO SE - fltr A - anticlip off
    20220204 SigGen gamma2 440 Hz sin 0dBFS 500uS div 5MHz filter ASIO SE - fltr A - anticlip off.png

    Ɣ2 SigGen gamma2 440 Hz sin 0dBFS 500uS div 5MHz filter ASIO SE - fltr A - anticlip on
    20220204 SigGen gamma2 440 Hz sin 0dBFS 500uS div 5MHz filter ASIO SE - fltr A - anticlip on.png

    Ɣ2 SigGen gamma2 20 Hz sqr 0dBFS 200uS div 5MHz filter ASIO SE - fltr A - anticlip off
    20220204 SigGen gamma2 20 Hz sqr 0dBFS 200uS div 5MHz filter ASIO SE - fltr A - anticlip off.png

    Ɣ2 SigGen gamma2 20 Hz sqr 0dBFS 200uS div 5MHz filter ASIO SE - fltr A - anticlip on
    20220204 SigGen gamma2 20 Hz sqr 0dBFS 200uS div 5MHz filter ASIO SE - fltr A - anticlip on.png

    Ɣ2 SigGen gamma2 20 Hz sqr 0dBFS 200uS div 5MHz filter ASIO SE - fltr B - anticlip off
    20220204 SigGen gamma2 20 Hz sqr 0dBFS 200uS div 5MHz filter ASIO SE - fltr B - anticlip off.png

    Ɣ2 SigGen gamma2 20 Hz sqr 0dBFS 200uS div 5MHz filter ASIO SE - fltr B - anticlip on
    20220204 SigGen gamma2 20 Hz sqr 0dBFS 200uS div 5MHz filter ASIO SE - fltr B - anticlip on.png

    Ɣ2 SigGen gamma2 20 Hz sqr 0dBFS 200uS div 5MHz filter ASIO SE - fltr C - anticlip off
    20220204 SigGen gamma2 20 Hz sqr 0dBFS 200uS div 5MHz filter ASIO SE - fltr C - anticlip off.png

    Ɣ2 SigGen gamma2 20 Hz sqr 0dBFS 200uS div 5MHz filter ASIO SE - fltr C - anticlip on
    20220204 SigGen gamma2 20 Hz sqr 0dBFS 200uS div 5MHz filter ASIO SE - fltr C - anticlip on.png
     
    • Like Like x 3
    • Epic Epic x 1
    • List
    Last edited: Feb 6, 2022
  7. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,652
    Trophy Points:
    113
    Location:
    On planet
    External Power Supply

    Measurement conditions
    1. Ɣ2 filter set to A
    2. 44.1 KHz sample rate except where noted
    3. spdif input
    4. SE output to oscilloscope or dScope

    An external power supply is only needed when the Ɣ2 is used with an input other than USB and is not connected to USB power. However Ɣ2 performs better with an external power supply and not connected to USB. When using USB input it appears that the Ɣ2 doesn't benefit from the external power supply.

    I expected spdif operation with external power to behave better than USB powered operation. When considerably inconsistent measurements were obtained in succession, I knew something was amiss. This is why one doesn't just push the button, run a measurement, then post the result.

    Here is an animation of Ɣ2 inconsistent dynamic range measurements acquired in rapid succession:
    01 20210621 gamma2 dynamic range FFT spdif SE animation.gif
    Those are wildly varying from 91 to 123 dB.

    Ɣ2 residual noise also measured inconsistently:
    02 20210621 gamma2 residual noise FFT spdif SE - 180 dB range - t1.png

    03 20210621 gamma2 residual noise FFT spdif SE - 180 dB range - t2.png
    -84 to -107 dBFS should make one suspicious that something is seriously wrong.

    I began to suspect the external power supply. Attaching an appropriately sized resistor to create a load similar to the DAC, which is approximately 150 mA, power supply spectrum and time domain ripple measurements were observed.

    Jameco power supply with 150 mA load - spectrum:
    04 20210623 Jameco 150mA load FFT 2a.gif
    Clearly the spectrum is changing over time.

    Jameco power supply with 150 mA load - time domain ripple:
    05 20210623 Jameco 150mA load 100mV div 5mS div 1MHz fltr animation.gif
    Ripple varies over time from nearly ok to a decidedly horrible 100 mV.

    It is likely that some component(s) in the Jameco power supply has deteriorated but not sufficient to cause complete failure. Keep in mind the Jameco power supply may be over 12 years old. The deficient behavior did not occur immediately but after approximately 15 minutes of warm up. Another good reason to warm up components before taking measurements for the record.

    Switching to an iFi X power supply, DAC results improved greatly, yielding consistent results in successive measurements as observed below:

    Ɣ2 dynamic range spdif in SE out
    06 20210621 gamma2 dynamic range FFT spdif SE.png

    Ɣ2 residual noise spdif in SE out
    07 20210621 gamma2 residual noise FFT spdif SE - 180 dB range.png

    Gain Linearity was also highly influenced by the power supply

    Ɣ2 gain linearity spdif in SE out - Jameco power supply
    08 G2_L1T3_1_A_Jameco.png

    Ɣ2 gain linearity spdif in SE out - iFi X power supply
    09 G2_L1T3_1_A_iFi.png
    Much improved performance when the compromised, aging power supply was replaced with the iFi X.
     
    • Like Like x 6
    • Epic Epic x 1
    • List
    Last edited: Feb 6, 2022
  8. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,652
    Trophy Points:
    113
    Location:
    On planet
    Comparison with ODAC (Original Generation)

    Comparison of AMB Ɣ2 to ODAC (original generation, not necessarily representative of current ODAC versions)
    NwAvGuy made considerable noise about the superiority of ODAC via design by measurements. The following comparisons reveal AMB Ɣ2 clearly measures better and preceded ODAC by several years. Someone didn't do their homework, or worse, purposely misled a large number of people.

    Jitter

    ODAC inferred jitter ASIO input SE output
    01 20210621 ODAC inferred jitter ASIO SE.png

    Ɣ2 inferred jitter ASIO input SE output
    02 20210623 gamma2 inferred jitter ASIO SE.png
    Ɣ2 demonstrates superior jitter performance than ODAC

    Ɣ2 inferred jitter ASIO input SE output
    03 20210621 gamma2 inferred jitter spdif SE.png
    Ɣ2 jitter is better still with spdif and iFi X external power supply


    Step Response (low frequency square wave)

    ODAC step response 0dBFS ASIO input SE output
    04 20210621 SigGen ODAC 50 Hz square 0 dBFS 100uS div 5MHz filter ASIO SE BW calc.png
    I have never observed this egregious behavior from any other DAC of the hundreds I have measured. ODAC oscillates between the two states during state transitions.

    Ɣ2 step response 0dBFS ASIO input SE output
    05 20210621 SigGen gamma2 20 Hz square 0 dBFS 100uS div 5MHz filter spdif SE BW calc.png
    Ɣ2 behaves as expected with a single transition and the usual pre and post transition ringing consistent with a linear phase filter.

    1KHz sine -90dBFS 16 bit performance

    ODAC 1KHz sine -90dBFS 16 bit performance
    06 FFT_0_L1T19_1_odac.png
    Discrete bit levels are noisy and indistinct

    Ɣ2 1KHz sine -90dBFS 16 bit performance
    07 FFT_0_L1T19_1.png
    Each of the discrete bit levels are apparent and with much lower noise than ODAC
     
    Last edited: Feb 6, 2022
  9. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,652
    Trophy Points:
    113
    Location:
    On planet
    reserved for additional data and corrections
     
  10. gixxerwimp

    gixxerwimp Professional tricycle rider

    Pyrate
    Joined:
    Sep 30, 2015
    Likes Received:
    5,773
    Trophy Points:
    113
    Location:
    small island claimed by China
    Have you measured an ODAC before? Maybe (also due to age?) its performance has deteriorated over time ...

    Measuring another unit would provide confirmation (and the final nail in the coffin ;))
     
  11. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,652
    Trophy Points:
    113
    Location:
    On planet
    I will be publishing measurements from the OG ODAC I own in the near future. Along with Mousai msd192 which is another WM8741 implementation.

    I have measurements from years ago and recent. They are not significantly different. I presented the step response oscillation years ago at an AES meeting.
     
  12. rhythmdevils

    rhythmdevils MOT: rhythmdevils audio

    Pyrate
    Joined:
    Apr 15, 2020
    Likes Received:
    12,239
    Trophy Points:
    113
    Location:
    Bay Area, CA
    Home Page:
    His name is NwavDouche just for the record ;)

    Thanks or the measurements! Very cool to see proof of a conman.
     
  13. Cspirou

    Cspirou They call me Sparky

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    8,200
    Trophy Points:
    113
    Location:
    Northwest France
    This post is a great example that just looking at a certain set of data doesn’t tell you the full story.

    From the FFT it appears that the γ2 is much worse than the ODAC. The time domain however shows a different story.
     
  14. Michael Kelly

    Michael Kelly MOT: Pi 2 Design

    Pyrate
    Joined:
    Apr 22, 2016
    Likes Received:
    4,112
    Trophy Points:
    113
    Location:
    Rhode Island, USA
    Home Page:
    Was this unit with the optional SRC4192 ASRC?
     
  15. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,652
    Trophy Points:
    113
    Location:
    On planet
    Yes, this Ɣ2 has a BB SRC4192, as does the Walnut X.3, also an WM8741 based DAC as measured here.
     
    Last edited: Feb 8, 2022
  16. Michael Kelly

    Michael Kelly MOT: Pi 2 Design

    Pyrate
    Joined:
    Apr 22, 2016
    Likes Received:
    4,112
    Trophy Points:
    113
    Location:
    Rhode Island, USA
    Home Page:
    That goes a long way to explaining the low jitter.

    Excellent work as always!
     

Share This Page