Heat and longevity
I own the Gustard X26 Pro. My advice is to be cautious if you don't own a 3d printer to create an external fan (I did). The DAC gets way too hot and will not survive in the long-term. The inside temps get at least 66°C and the capacitors are rated 85°C for 1000 hours. Heat is the enemy of transistor and capacitor longevity which is why vintage products like McIntosh last 40-50 years before needing a capacitor change. Also, not sure how all the heat affects the solder and modern PCB boards as it expands and contracts from heat over time. I will be sharing my 3d printed fan later for this DAC.
Class A is not the problem. Poor thermal management is the problem. Pictured is my custom heat management 3d print
Sonics
Sonically it's quite good. I can hear the difference between the filters Vivid and Gentle/Composite. However, I found it sounds best when using the PC to upsample to 384/768khz or 352.8/705.6khz when listening to 44.1khz. When comparing at 44.1khz and 48khz to my older 2014 McIntosh using Burr brown and cirrus logic chips the Gustard sounds noticeably more full. However, this McIntosh did not have a PCM1792A like other older McIntosh products. Compared to the McIntosh MHA150 which also uses a Sabre DAC the Gustard removes a bit of the edginess (classic ESS IMD hump?) but the increase in sonics is not worth it for the cost, very subtle and I would be very happy with the MHA150 DAC. I have only compared low bit rates.
Upsampling, filters and NOS mode
When upsampling to 352.8 khz+ everything is clearer/smoother and is most noticeable with 44.1khz and 48khz music. Also, when upsampling the DAC filters are no longer audible and I can hear no difference when NOS mode is enabled. NOS mode sounds horrible on lower sample rates (as it should). I found that while subtle upsampling 44.1khz to 352.8/705.6khz did sound better than upsampling to something divisible by 48khz.
DSD or PCM
I could only get Jriver to upsample to DSD256 (10.2Mhz). Whenever I tried DSD512 it stuttered which might be more of a problem with my laptop CPU struggling and not the DAC. Didn't sound too different than upsampled PCM and I couldn't reliably tell the two apart.
Headphones vs speakers
The Sennheiser HD650 headphones could not hold a candle to the Jamo R909 in detail retrieval and sound stage. A superior DAC didn't do much to improve the sound for headphones and upsampling was almost inaudible (unlike speakers which were very noticeable). I assume the pinna of the ear, superior high frequency performance of tweeters, and lower THD/IMD speakers plays a role here.
The Sennhieser HD650's have good THD performance but I believe modern speaker drivers outperform it. Like most headphones 10khz+ performance is bad or non-existent. According to Rtings.com measurements soundstage is not large due to the headphone not reacting with the pinna of the ear. Perhaps the HD800S might be a better choice as it reacts with the pinna much better, confirmed by measurements.
Jamo R909 open baffle speakers which used the best available DIY drivers at the time. Uses the Scanspeak Revelator 9700, same motor of the 9900 but has the dome of the 9500. Midrange is a modified version of the SEAS hexadyne W15CH001 for open baffle use, their top motor. The four 15-inch bass drivers are custom made for open baffle use, likely by BMS. Modern day SBAcoustics Satori, Illuminator or Purifi drivers are a tad better but when making a 3-4 way speaker these differences are minor or non-existent as drivers are used in their most linear band.
Amplification
Luxman CL-38, Luxman MA-88 tube monoblocks, and McIntosh MHA100 (used to own MHA150 too). I have tried all combinations along with the Gustard X26 pro acting as a preamplifier. The best combination was the Gustard directly driving the MA-88's or acting as a DAC into the MHA-100 with XLR cables. The lower "90dB or more" signal to noise ratio of the MA-88's did not in any way affect sound quality or detail retrieval. The greatest difference in sonics was upsampling PCM to high bitrates. DAC quality is audible whether using tubes or solid state however I am using top tier products.
My take away
- Top tier speakers are more revealing than headphones.
- The Gustard is no worse than other DAC's at 44.1khz and 48khz. Sounds excellent when upsampled.
- Upsampling sounds better but requires SOTA speakers in regards to THD/IMD/crossover quality
- DAC quality is likely a result of filters and upsampling as opposed to THD/SINAD.
- Careful attention to 44.1khz upsampling might be good for future measurements on audiosciencereview, DAC's which upsample 44.1khz to a frequency divisible by 44.1khz might be better. I could hear the difference.
- NOS mode sounds horrible unless upsampled very high. Sounds excellent when upsampled.
- Upsampling to very high frequencies shifts the noise to higher frequencies. Upsampling very high likely moved this noise beyond 20khz, I assume my Revelator tweeters could pick this up.
- A computer does a better job than the upsampler built into the DAC. I wonder if this is Gustard thing or a ESS Sabre thing, maybe DAC's are not completely solved when it comes to CPU power. After all my laptop seemed to have struggled.
- Tube amplifiers are revealing of DAC's despite their inferior SN ratio and THD. However, I assume this is only if your tube amplifier is engineered to be objectively pure with TOTL output transformers and not subjectively "tubey"