Gustard X26 Pro Measurements (F20 Firmware + 10Mhz clock)

Content and analysis made possible thanks to support from https://headphones.com, and https://patreon.com/goldensound supporters.

This unit was loaned to me by Aoshida Audio for testing and review

This unit was previously measured, however I am posting this as an updated set of measurements with the DAC running the F20 firmware (available here) and also looking into use of a 10Mhz clock.

ChitChat:

The Gustard X26 Pro is a fully balanced digital to analog converter which whilst using an off the shelf DAC chip, contains extensive proprietary design elements, impressive objective performance, and for a relatively reasonable price too.

Design:

The X26 is using dual ESS9038PRO chips, which on the mainboard itself are actually using a heatsink, indicating that the DAC is likely running in current mode as opposed to voltage mode.
I’ve been told by another manufacturer that they subjectively preferred the sound of the 9038 when running in current mode but that it created significantly more heat. The X18 does run fairly warm.

Next there is a board containing an array of ICs and processors. These serve a variety of functions.
It’s not clear if the Arm microprocessor or the XMOS XU216 chip is actually responsible for the USB receiver circuitry. The position of the ARM microprocessor would indicate that it is, with the XMOS chip likely serving as the MQA decoding processor as the MQA implementation typically runs on XMOS chips.

The analog devices chip is used for various DSP options, including oversampling, and likely DSP volume control.
Though it’s a little unclear as to how Gustard’s upsampling/oversampling implementation on this chip differs from the internal ESS one as the filters seem to be identical.

The X26 pro also features built in bluetooth with aptX HD and LDAC support, so you can use this wirelessly with minimal loss in fidelity compared to a direct connection.

And lastly it does also have full MQA support, though as you may already know, I’m not exactly a fan of MQA so in my opinion this just means the product is more expensive than it would otherwise be as it has an MQA licensing fee attached.

Measurements:

Test Setup:

– Audio Precision APx555 B-Series Analyzer with 200kOhm input impedance set unless otherwise specified
– USB Source: Intel PC via intona 7055-C isolator
– Measurement setup and device under test are running on an AudioQuest Niagara 5000 power supply
– Audioquest Mackenzie XLR and RCA interconnects
– Intona Reference Impedance Characterized USB Cable
– Exact analyzer/filter configurations for each measurement are detailed in the full reports

Before we go further, I’d like to mention the use of a galvanic isolator in my measurement setup. USB ‘cleanup’ devices come in all shapes and sizes, and whilst I think that many of them and their marketing claims are complete nonsense, it is a simple fact that noise on the USB connection can have a negative impact on a DAC, even if there is no ground loop present and regardless of grounding setup. And therefore when measuring devices I want to ensure that my USB source is not negatively impacting the measurement results.

To do this, I use an intona 7055-C galvanic isolator which completely separates the DAC from the PC, meaning no noise can pass through.
The X26 pro is affected by USB noise even from my fairly small PC and this can be shown in a few ways.

As a couple examples, firstly, looking at ultrasonic noise:

We can see that there is some additional ultrasonic noise content present which disappears when we isolate the DAC.
But noise can also have indirect effects on a DAC.
For example the clocks in your DAC are called ‘Voltage Controlled Crystal Oscillators’ and they rely on being fed by a very clean, stable voltage to be accurate. If noise from the USB source contaminates this or the ground plane, there can be a degradation in performance.
Below is a jitter test comparing direct to PC vs with an isolator:

Not a massive change, but we can indeed see some added jitter components when connected directly to the PC.
This is why I use an isolator in my measurements (though I always check to ensure the isolator does not have a negative impact on performance. Some DACs rely on grounding via USB and so it may actually worsen things in that situation, but for 90% of DACs there is either an improvement or no change.)

Full Measurement Reports:

Reports available here:

Dynamic Range (AES17): 125.4dB

SNR: 126.97dB

IMD SMPTE: -108.6dB

Noise Level RMS (20-20khz): 2.267uVrms

Noise Level RMS (20-90khz): 4.306uVrms

DC Offset: 402uV active, 371uV idle

1khz 0dBFS Sine, Balanced Out (F20 Firmware):

1khz 0dBFS Sine, Balanced Out (F1A Firmware):

Raw SINAD doesn’t seem to change between F1A and F20 firmwares. Maybe a couple minor differences in harmonic structure but it’s too close to say with certainty that it isn’t just run to run variation.

1khz -3dBFS Sine, Balanced Out (F20 Firmware):

Analog Filter:

The X26 Pro incorporates an analog low-pass filter that attenuates content after 20khz. This is meant to reduce potential unwanted ultrasonic imaging or noise at the output.

-90.31dBFS 1khz sine (96khz capture bandwidth):

The exceptional dynamic range and lack of ultrasonic noise means that low level signals like this are reproduced very cleanly!

Filter Ultrasonic Attenuation:

Idle Noise FFT:

THD+N vs Frequency:

IMD:

Linearity:

Crosstalk:

Multitone:

Wideband:

Jitter:

Jitter performance here is truly excellent! A couple small spurs but other than that absolutely excellent.
So a question that many people will have: “Is adding a 10Mhz clock worth it?”

The short answer: No, 10Mhz clocks were never intended to improve jitter performance and almost never will do so. They are a tool to solve a problem in professional environments that does not exist in home setups.

Here’s the X26 Pro jitter when using the APx555’s 10Mhz clock as a reference:

As you can see, quite notably poorer jitter performance than without the 10Mhz clock.
So why is this?
The reason why a 10Mhz clock will worsen jitter isn’t to do with the particular 10Mhz clock used (though it will have an impact of course), even if you used the most perfect 10Mhz clock in the world you’re still going to get poorer performance simply due to how 10Mhz clocks are actually implemented.

In your DAC, there will almost always be two clocks:
One at 45.1584Mhz (1024 x 44.1khz)
and
One at 49.152Mhz (1024 x 48khz)

You can use a DDC to potentially improve jitter performance, because when connecting via I2S, you are literally replacing the 45.1584mhz/49.152mhz clocks in your DAC with the (hopefully higher quality) ones in your DDC. (Though some DACs will have PLL or buffering systems but that’s a bit of a tangent). And when using SPDIF/AES the clock signal is embedded in the stream too.

You can also do this with a dedicated word clock instead of a DDC.
The dCS one as an example provides a 44.1khz and 48khz word clock to the DAC

The two above systems provide a native rate mclk or wclk directly to the DAC which is directly used to determine when to convert samples. It is effectively replacing the DACs internal clocks and as such can provide an improvement in performance if the external clock is of higher quality than the internal one.

With a 10Mhz clock, this is not the case. A 10Mhz clock cannot be cleanly divided by 44.1khz or 48khz, and it cannot and does not directly run the DAC.
Instead, the DAC uses a PLL system with the 10Mhz clock as the input reference, to output a 49.1520/45.1584Mhz (or whatever the required rate is) using its OWN clock internally.
Your DACs clock is still the clock feeding the converter, it’s just being kept in time long term with the 10Mhz clock.

10Mhz clocks were never intended to improve jitter performance. They were intended to keep many devices in sync and prevent ‘clock drift’ over time in professional situations where you may have potentially dozens of DACs, ADCs, Processors etc running simultaneously and you do not want one to be running slower than the other and causing desync over time. As this could potentially cause issues in production or recording/outputting from several devices simultaneously.

This is particularly important in situations where video is involved and audio needs to be perfectly in time with frame capture.
This was the intended purpose and it works very well, but it does not and was never intended to improve short term timing/jitter performance and will usually make it worse due to the challenges of fractional clock division.

This video has a lot of information on the topic if you wanted to know more:

Impulse Response:

Gustard also says it ‘supports NOS’. Which is confusing because…well..you can’t really run a delta sigma dac in ‘NOS’.
The NOS setting just disables the initial FIR filter. But the secondary IIR filter and modulation stage still occurs and so the impulse response looks like this:

I 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.