Matrix Element X Measurements

Discussion in 'Source Measurements' started by ultrabike, Feb 12, 2020.

  1. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Friend MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    9,618
    Dislikes Received:
    19
    Trophy Points:
    113
    Location:
    Irvine CA
    This is a relatively high end combo DAC/Amp/Streamer system from Matrix Audio that supports a wide variety of formats and data rates. As a streamer, it is capable of connecting to Apple/Android devices through WiFi. As an amp, it supports two single ended headphones. It can function as both a balanced amplifier and source.

    Under the hood there is an ESS9038PRO and a high precision clock. Based on what I've seen, it seems the clock can do multiples of 48 kHz, and 44.1 kHz through PLL. The BW of the PLL has three possible settings. For the amplifier, the system relies on the LME49600.

    The product is very well made. It is not something I would consider portable by any means. Instead, I would envision this to be the heart of a stereo system, driving high quality powered monitors or a high end amplifier. The provided remote should facilitate this use case.

    The marketing specs are as follows:

    Line Output
    XLR
    SNR: > -131dB
    THD+N: <0.00012%@1k, <0.00025%@20Hz-20kHz
    Crosstalk : >-145dB
    Output Level: [email protected] ,[email protected]+10dB
    RCA
    SNR: > -124dB
    THD+N: <0.00012%@1k, <0.00025%@20Hz-20kHz
    Channel Crosstalk : >-137db
    Output Level: [email protected] , [email protected]+10dB

    Headphone Output
    XLR4 and Dual XLR3
    SNR: >-117dB
    THD+N: <0.00055%@20Hz-20kHz 1VRMS
    Output Impedance: <1.5Ω
    Output Power: [email protected]Ω, [email protected]Ω, [email protected]Ω, 1%THD
    TRS 6.35mm
    SNR: >-112dB
    THD+N: <0.00055%@20Hz-20kHz 1VRMS
    Output Impedance: <1.5Ω
    Output Power: [email protected]Ω, [email protected]Ω, [email protected]Ω, 1%THD

    Single Ended Headphone Performance

    Measured performance using HW Vesion 1.6 / FW Version A10301 is as follows for the headphone single ended output in low gain, and the DAC rate set to 44.1 kHz (PCM Filter 1):

    Report.PNG

    MA specifications call for a THD+N < 0.00055% @ 1 Vrms. This is equivalent to THD+N < -105 dB @ 30 mW/33ohm and @ 3.3 mW /300 ohm.

    The measurement yields THD+N < -108 dB @ 20 mW/33 ohm and -102.7 dB @ 2mW/300 ohm, which is with in the ball park of the specification. These are very impressive results. But as far as I know, they are not record breakers. Compared to the good old O2, this is how THD+N results stack up:

    THD+N_Report_vs_O2.PNG

    Here are some THD+N sweeps using 44.1 kHz and the usual 33 ohm and 300 ohm loads (unbalanced HP):

    300 ohm Load

    THD+N_vs_Vout_44kHz_HP300ohm_unbalanced.PNG

    33 ohm Load

    THD+N_vs_Vout_44kHz_HP33ohm_unbalanced.PNG

    Note that performance starts to degrade for the 33 ohm case above 1 Vrms and it varies run to run. This could be partially due to the loads getting too hot. That said, it is likely that plenty of headphones will have a melt down above 1 Vrms.

    The output impedance is about 1.34 ohms, and Crosstalk is about -136 dB, all of which are in agreement with published specs. Output power into 33 ohms is about 1.3 W which is comparable to the published 1.15 W. However, output power into 300 ohms is about 157 mW which is less than the published 308 mW. The measurement was done in low gain mode, and is very likely that a high gain measurement will be much closer to the specs.

    SNR is a healthy -123.8 dB. Before we get too excited about this, note that this is referenced to 0 dBFS (DAC) / 12 dB (Amp), which for low gain mode corresponds to a little north of 7 Vrms. At 400 mVrms, SNR is a more realistic -98.3 dB. The O2 pulls out -101 dB of SNR referenced to 400 mVrms. In my opinion however, the result is pretty damn good.

    So overall, great results.

    DAC Performance

    The DAC performance is quite impressive. It does push the limits of the SR-1.

    For 44.1 kHz, the THD+N vs Vout is as follows:

    THD+N_vs_Vout_44kHz_Line_Out.PNG

    The top sweeps correspond to balanced and unbalanced performance, while the bottom one is the analyzer's internal loopback. Around 500 mVrms, the DAC catches up to the analyzer and the most we can claim is < -110 dBm in that region. Unbalanced performance seems to be slightly better in the 44.1 kHz case.

    Here are similar results for the 48 kHz case:

    THD+N_vs_Vout_48kHz_Line_Out.PNG
     
    Last edited: Feb 27, 2020
  2. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Friend MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    9,618
    Dislikes Received:
    19
    Trophy Points:
    113
    Location:
    Irvine CA
    The following plots show THD+N at 0 dBFS for the DAC output for both balanced and unbalanced outputs using both 44.1 kHz and 48 kHz.

    THD+N @ 0 dBFS / balanced / 44.1 kHz

    THD+N_Line_Out_0dBFS_A_Weighted_20kHz_AES_44kHz_balanced.PNG

    THD+N @ 0 dBFS / unbalanced / 44.1 kHz

    THD+N_Line_Out_0dBFS_A_Weighted_20kHz_AES_44kHz_unbalanced.PNG

    THD+N @ 0 dBFS / balanced / 48 kHz

    THD+N_Line_Out_0dBFS_A_Weighted_20kHz_AES_48kHz_balanced.PNG

    THD+N @ 0 dBFS / unbalanced / 48 kHz

    THD+N_Line_Out_0dBFS_A_Weighted_20kHz_AES_48kHz_unbalanced.PNG

    Both internal results using the analyzer's low distortion sine, and the Matrix Element X outputs are shown.

    It can be seen that the dominant distortion components are very close to the limit of the analyzer. However, note that for the 44.1 kHz case, there seems to be a little more of residual harmonics relative to the 48 kHz case. There also seems to be some unmitigated images in both cases. This may correlate to some observed issues this particular system had with filtering @ 44.1 kHz and 48 kHz which will be discussed later. That said I don't think there are any show stoppers, and the results are quite impressive.

    NOTE: The plots vs frequency above shows THD, but the number is THD+N at a particular Vout level. In all cases noise is dominant, and indeed quite low.

    The following are Line Out balanced IMD SMPTE results for both 44.1 kHz and 48 kHz. Note the sweep includes an analyzer loopback ("Internal") measurement to show how close the measurements are to the SR1 capabilities.

    IMD SMPTE vs Vout 44.1 kHz balanced:

    IMD_SMPTE_vs_Vout_44kHz_Line_Out_Balanced.PNG

    IMD SMPTE vs Vout 48 kHz balanced:

    IMD_SMPTE_vs_Vout_48kHz_Line_Out_Balanced.PNG

    Results are very close to the limits of the analyzer, particularly above 500 mVrms.
     
    Last edited: Feb 21, 2020
  3. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Friend MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    9,618
    Dislikes Received:
    19
    Trophy Points:
    113
    Location:
    Irvine CA
    Next we move to the Filter options of this device.

    @ 88.1 kHz, the following shows the roll-off performance of the filters:

    Frequency_Response_88kHz_All_Filters.PNG

    There is a small residual artifact shown in the sharper roll-off filter. This is supersonic and small.

    For the 44.1 kHz and the 48 kHz cases, there seems to be an error in the location of the filter corner frequency relative to the analog low pass filter (for comparison purposes, the PCM Filter 1 option for 88.1 kHz is also shown):

    Frequency_Response_44kHz_All_Filters.PNG

    I'm not sure if this is a problem fixed in later FW versions. I also don't think this is show stopper but it may correlate to the unmitigated images shown in the distortion measurements.

    After loading some the ASIO drivers for the Matrix (thanks @bilboda), I might be able to see if there is a FW update that could be uploaded to the system. However, this is not my unit and I'm hesitant to screw around with it.

    Getting this thing to play from my Iphone through WiFi was easy-peasy. Controls are very intuitive.
     
    Last edited: Feb 21, 2020

Share This Page