Gustard X26 Pro Measurements

Discussion in 'Digital: DACs, USB converters, decrapifiers' started by GoldenOne, Jul 12, 2021.

  1. GoldenOne

    GoldenOne Friend

    Friend
    Joined:
    Jul 17, 2020
    Likes Received:
    895
    Trophy Points:
    93
    Location:
    England
    Gustard X26 Pro: https://aoshida-audio.com/products/gustard-dac-x26-pro

    20210712_182959.jpg

    Test Setup:
    - Audio Precision APx555 B-Series Analyzer with 200kOhm input impedance
    - Van Damme Star-Quad shielded SPC XLR cables with Amphenol Connectors
    - USB Source: AMD PC via ifi iGalvanic 3.0
    - Audioquest Carbon USB and SPDIF cables
    - All measurements shown in this post are taken with SPDIF input, 44.1khz sample rate and stock 'vivid' filter unless otherwise noted (other input/filter measurements available in reports)
    - Full reports containing additional data and test configurations are attached
    - Exact analyzer/filter configurations for each measurement are detailed in the full reports
    - APx Project is included in the reports folder should anyone wish to repeat this set of measurements or use it for another DAC

    ChitChat:

    This DAC is an interesting one, it's quite pricey at about $1500, and offers a few extra/unusual features. It's not quite the same 'insert ESS chip and Opamp here for 122dB SINAD' design, with 10Mhz clock input support, a discrete Class A output stage (with analog LPF) , and a fair amount of extra DSP and clock related stuff.

    As a result, this costs twice the price of something such as a D1SE. Whether that is worth it is another question. At $1500 it's knocking on the door of a lot of very well regarded DACs like the Yggdrasil, Spring 3 etc.

    If there are any other measurements you'd like to see they're most likely in the reports. But if not, let me know and I'll add them to my sequence for future tests.

    If anyone sees any issues with any measurements or configs please let me know and they'll be addressed.

    FULL REPORTS:

    Reports available here: https://drive.google.com/drive/folders/1Lvy6m79jos2vK-kiipRGUW8aSd0nMpX7?usp=sharing

    Reports contain the following measurements:

    - 1khz Sine View
    - 15khz Sine View
    - -90.31dBfs Sine View
    - Phase Response
    - Filter Ultrasonic Attenuation
    - 20hz-90khz RMS Noise Level
    - Device Output Level (Vrms)
    - 20hz-20khz Noise RMS Level
    - Idle Noise FFT
    - 1khz FFT (0dBfs and -3dBfs)
    - 50hz FFT (0dBfs and -3dBfs)
    - Effective Number of Bits (0dBfs and -3dBfs)
    - THD+N (0dBfs and -3dBfs)
    - THD+N / Frequency
    - Dynamic Range (AES17)
    - SNR
    - IMD (SMPTE)
    - 50hz/7khz IMD SMPTE FFT
    - IMD SMPTE vs Output Level
    - Linearity
    - Linearity (No Bandpass)
    - SINAD/Output Level
    - Crosstalk / Frequency
    - DC Offset when Active
    - DC Offset when Idle
    - 44.1khz Jitter
    - 48khz Jitter
    - 90khz Bandwidth test
    - 32 Tone FFT
    - 1.2Mhz Bandwidth Noise FFT
    - 1.2Mhz Bandwidth 1khz FFT (0dBfs and -3dBfs)
    - And more!


    Dynamic Range (AES17): 125.4dB

    SNR: 127dB

    Noise Level RMS (20-20khz): 2.264uVrms

    Noise Level RMS (20-90khz): 4.405uVrms

    DC Offset: 567uV Active, 461uV idle


    1khz 0dBfs:

    ezKbpTodCW.png

    Pretty crazy THD+N performance, BUT:

    1khz -3dBfs:
    SINAD/THD+N: 119.718 (L) / 119.523 (R)
    upload_2021-7-13_1-37-45.png

    Actual SINAD figure is still very good, but it is rather odd that once we stop playing a full scale signal the 3rd harmonic in particular is suddenly almost 15dB higher.....
    Not sure if this is an unintentional quirk or something done specifically to optimise for looking good in test conditions.

    THD+N / Frequency:

    upload_2021-7-13_1-40-51.png

    Linearity:

    upload_2021-7-13_1-41-15.png

    SINAD/THD+N vs output level:

    upload_2021-7-13_1-41-48.png
    upload_2021-7-13_1-41-58.png
    upload_2021-7-13_1-42-5.png

    IMD (SMPTE) vs Output Level:

    upload_2021-7-13_1-42-38.png
    upload_2021-7-13_1-42-44.png
     
    • Like Like x 6
    • Epic Epic x 1
    • List
    Last edited: Jul 12, 2021
  2. GoldenOne

    GoldenOne Friend

    Friend
    Joined:
    Jul 17, 2020
    Likes Received:
    895
    Trophy Points:
    93
    Location:
    England
    Crosstalk:

    upload_2021-7-13_1-51-44.png


    -90.31dBfs 1khz Sine (Recorded with 192khz ADC Sample Rate):

    upload_2021-7-13_1-52-0.png


    Filter Ultrasonic Attenuation:

    upload_2021-7-13_1-52-8.png

    Gustard says this is a custom/proprietary filter, but it seems to be identical to ESS Fast Linear filter.

    D1SE 'Fast Linear' vs Gustard 'Vivid' filter, can you tell which is which?
    upload_2021-7-13_1-54-8.png

    Impulse Response:

    Vivid:
    upload_2021-7-13_1-54-53.png

    Gentle:
    upload_2021-7-13_1-55-3.png

    Composite:
    upload_2021-7-13_1-55-14.png

    Gustard also says it 'supports NOS'. Which is confusing because...well..you can't really run a delta sigma dac in 'NOS'.
    Looks like the 'NOS' option is basically a VERY short minphase filter:
    upload_2021-7-13_1-56-24.png

    Was sort of expecting a Zero-Order-Hold filter on this one similar to RME's 'NOS/Super-Slow' filter on the ADI-2 which does actually produce a result similar to what you'd expect from a NOS dac.
    [​IMG]
     
    Last edited: Jul 12, 2021
  3. GoldenOne

    GoldenOne Friend

    Friend
    Joined:
    Jul 17, 2020
    Likes Received:
    895
    Trophy Points:
    93
    Location:
    England
    Idle Noise Spectrum upto 1.2Mhz:

    upload_2021-7-13_2-0-35.png

    Analog Low Pass FIlter:
    Gustard mentioned the analog low-pass filter so I checked that, 192khz input
    upload_2021-7-13_2-1-50.png

    44.1khz Jitter (USB):
    (SPDIF 44.1khz jitter was almost identical)
    upload_2021-7-13_2-2-38.png

    48khz Jitter (USB):

    upload_2021-7-13_2-3-14.png

    Great jitter performance here, though it wasn't as good on SPDIF input:

    upload_2021-7-13_2-3-55.png

    Multitone:

    No matter what I did, I struggled to get a result on this test where the <200hz region wasn't overly high. This is one of the best results, many of them had noise upto about -100dB.
    I'm not entirely sure if this is a measurement configuration issue or the DAC. I've encountered it with other DACs sometimes, but in those cases just running the test again or enabling frequency shift correction sorts it.

    With the X26 Pro, I really struggled to get this test to work properly so take this with a grain of salt as this is very much a cherrypicked screenshot:
    upload_2021-7-13_2-4-46.png
     
    Last edited: Jul 12, 2021
  4. GoldenOne

    GoldenOne Friend

    Friend
    Joined:
    Jul 17, 2020
    Likes Received:
    895
    Trophy Points:
    93
    Location:
    England
    Reserved 3
     
  5. GoldenOne

    GoldenOne Friend

    Friend
    Joined:
    Jul 17, 2020
    Likes Received:
    895
    Trophy Points:
    93
    Location:
    England
    Reserved 4
     
  6. dllmsch

    dllmsch Friend

    Friend
    Joined:
    Jan 3, 2016
    Likes Received:
    276
    Trophy Points:
    63
    Location:
    Molvanîa
    AFAIK the owner of gustard got hold of an APx555 while designing this amp, so the amp should look really good in ASR style measurements. However X26Pro have quite good of a reputation among Chinese subjectivists as well, wonder how it fares against schiit and holo audio's offerings subjectively.
     
  7. Marvey

    Marvey Super Friend

    Staff Member Friend BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,693
    Trophy Points:
    93
    [​IMG]

    That impulse response will mean a roll off in the highs of the audio passband. In essence, it would mimic that particular NOS characteristic. However, I suspect part of the NOS sound with some NOS DACs is the anti-aliasing / reconstruction filter being implemented at the headamp or headphone itself - LOL - which maybe explains the dense timbre deal.

    upload_2021-7-13_8-28-51.png

    I wonder if this is endemic to ESS chips and related to what many people hear as the "Sabre" highs or digititus - granted the latest and greatest from ESS seems to have calmer highs than their prior generation. Now I would be curious to see if how far these harmonics extend past 20kHz.

    They are different sounding. I don't see the X26 as sounding significantly different from the Matrix X-Sabre. I think @Vtory may have had both X26 and X-Sabre. They may be small differences, but these two DACs will have a strong family sound. @ChaChaRealSmooth after listening to an Yggdrasil GS said to me "why da F did you let me buy the X-Sabre?". LOL, I replied that he pretty much needed to figure it out for himself. (I did feel the X-Sabre was good sounding DAC). @Vtory was on a ESS DAC tear, until he met the Soekris.

    It's a matter of preference. I think in this community there is a strong preference toward the resistor ladder / resistor string DACs. The ESS stuff can be good. I really like the MOTU Ultralite, it's almost not an ESS sound, but I suspect their engineers may have whispered it. I should take measurements of the MOTU.
     
    • Like Like x 7
    • Respectively Disagree Respectively Disagree x 1
    • List
    Last edited by a moderator: Jul 13, 2021
  8. Walderstorn

    Walderstorn Friend

    Friend
    Joined:
    Mar 20, 2016
    Likes Received:
    1,537
    Trophy Points:
    113
    Location:
    Europe
    Very interesting review that is now up on Youtube so check @GoldenOne 's channel.
    For those, of us, with medium budgets, this can be an even more interesting review.
     
  9. Merrick

    Merrick A lidless ear

    Friend
    Joined:
    Jan 6, 2016
    Likes Received:
    10,590
    Trophy Points:
    113
    Location:
    Portland, OR
    I checked out when the timbre was described as slightly off, needing a bit more body.
     
  10. mitochondrium

    mitochondrium Friend

    Friend
    Joined:
    Sep 1, 2017
    Likes Received:
    850
    Trophy Points:
    93
    Location:
    A Cell
    Not specifally for the X26. But I am wondering what the 10 MHz Clock can bring to the table when the Jitter test via USB is already performing quite well.

    @GoldenOne you would not happen to have a REF 10 or other 10 Mhz clock lying around you could hook up to the X26 and compare jitter of USB and spdif with and without 10 Mhz clock. In some parts people swear by reclocking, I have a hard time imagining that this would make a big measurable difference in the J-test and even if it did whether it would make a noticeable audible difference. I have to admit, I do not know whether the J-test is a little like SINAD in that it does not tell you the whole story and a 10 Mhz clock may yield audible improvements without showing a difference in the J-test.
     
  11. GoldenOne

    GoldenOne Friend

    Friend
    Joined:
    Jul 17, 2020
    Likes Received:
    895
    Trophy Points:
    93
    Location:
    England
    I'd definitely like to experiment with a 10mhz clock. Both with some dacs and some ddcs.

    Unfortunately I don't have one and they're usually VERY expensive so unlikely to be able to do so unless mutec or someone is happy to lend one.
     
  12. soumya

    soumya Acquaintance

    Joined:
    Jun 24, 2018
    Likes Received:
    33
    Trophy Points:
    18
    Location:
    Mordor, Middle Earth
    Sorry for responding to an older thread.
    First off, thanks for posting these measurements knowing how much time and effort goes in to this.

    I thought I should call out on the NOS feature of Gustard ESS implementations.
    The NOS feature essentially bypasses the 8x FIR based interpolation altogether - at least for <= 192kHz sources.

    If you check out their user manual - say X16 or X18, this is what Gustard says
    "c. NOS MODE: Turn on to bypassing the internal interpolating FIR filter, sourcing data directly into the IIR filter. This function requires data oversampled by 8x externally or high-spec data."

    Page 7 from X18 manual
    http://www.gustard.cn/qfy-content/uploads/2021/12/92c8d4d34cd68beb1c2195348ed31348.pdf

    In other words they are exposing the ESS' OSF bypass feature as a menu option to the end user.
    Page 9 from ES9038 Pro datasheet
    https://www.esstech.com/wp-content/uploads/2021/03/ES9038PRO-Datasheet-v3.7.pdf

    "Oversampling Filter (OSF) Bypass The oversampling FIR filter can be bypassed using bypass_osf in Register 37: Programmable FIR Configuration, sourcing data directly into the IIR filter. The audio input should be oversampled at 8 x fs rate when OSF is bypassed to have the same IIR filter bandwidth as PCM audio sampled at fs rate. For example, a signal with 44.1kHz sample rate can be oversampled externally to 8 x 44.1kHz = 352.8kHz and then applied to the serial decoder in either I2S, LJ, or RJ format. The maximum sample rate that can be applied is 1.536MHz (8 x 192kHz)."

    The reason you see the minimum phase impulse response is because the 2nd stage oversampling (which takes the final bandwidth to MHz region before HyperStream modulator) after digital interpolation is an IIR filter. And IIR filters are minimum phase by design (well mostly).

    So there is no slow minimum phase FIR filter. In fact there is no filter at all. But that also means the necessary bandwidth needed to feed to IIR stage also gets reduced at least 8x (for 44.1 / 48 kHz sources). Which explains the early roll off from around 5kHz. Situation is better if using 176 / 192 kHz sources.
     
  13. Mista Lova Lova

    Mista Lova Lova Rando

    Joined:
    Oct 4, 2022
    Likes Received:
    0
    Trophy Points:
    1
    Location:
    UK
    As an X26 Pro owner, this is very interesting; unfortunately, I lack the technical knowledge to fully understand the implications of the above.

    I am currently using Roon's upsampling fed to my X26 Pro in NOS mode; it sounds better to me than the DAC's internal upsampling and filters. Although, for sheer body and punch I might have to give it to the DAC itself, but for more nuanced, spacious and separated sound I'd go with Roon + NOS.

    Should I be sticking to x8 upsampling with Roon, in light of what you wrote, or should I just go for the highest available rate (power of 2)?

    To my ears, NOS mode on and off sound different even when feeding the highest possible rate from Roon. Does it mean that despite the sample rate, with NOS mode off, the DAC is still doing additional processing (adding its own filter on top of Roon's?) which it does not do in NOS mode?

    Thanks
     

Share This Page