The Audio-GD DI-20HE is a fairly expensive DDC at around $1000. It has a fairly extensive feature-set including I2S, SPDIF, BNC output, 10mhz clock input support, isolated USB, adjustable I2S pinout, and 256fs clock output support.
It also features Audio-GD’s ‘regenerative’ power supply. A beefy class-A design which supposedly gives the best possible PSU performance.
But, how does the device actually perform?
Well….in a word: Insane!
I’m hugely impressed with this, and will definitely be doing a full review. I’d also love to buy one for myself once I’m able.
The only thing that I wish this did differently was having 768khz (or 1.536mhz) support.
If you’re wondering why a DDC is helpful in the first place you can watch my video on the topic here:
– Audio Precision APx555 B-Series Analyzer
– AudioQuest Carbon SPDIF, BNC and AES Cables (all 1.5m)
– AudioQuest NRG-Z3 power cable, direct to mains.
– AMD PC USB -> ifi iGalvanic USB Isolator -> ifi iUSB 3.0 hub feeding both APx555 and DI-20HE
– 44.1khz and 48khz real music played through device during measurement
To measure jitter, the device is connected to the digital input of the APx555, and the analyzer is set to analyze jitter, not audio content.
These measurements do not show audio/analog info, but instead show the spectrum of jitter, ie: time-domain inconsistencies.
This is the primary factor that a good streamer or DDC will seek to improve.
NOTE: I tried the DI-20HE in both ‘serial’ and ‘parallel’ modes, and jitter didn’t seem to change in any noticeable fashion.
This is the best DDC I’ve come across thus far, and it’s not far off the APx555’s own internal loopback in terms of RMS/Peak jitter levels.
I actually had to move the graph down from where I had it for previous measurements because it’s so low….
This is super impressive.
Some odd stuff going on in the higher freqs that isn’t present on other devices such as the pi2aes, but everything in the audible band is EXTREMELY impressive.
What about the other outputs?
The BNC output, which uses Kingwa’s ‘ACSS’ design, does not perform as well as the coax out.
We see the same four spikes in the upper frequencies poking out, but the random jitter is much higher, and RMS/Peak jitter levels are around 7x higher than coax out.
It’s likely that this may be different with an Audio GD dac if the BNC input there has something specially designed for this ‘ACSS’ approach, but for any other dac, stick to Coax out.
AES/EBU performance is also crazy impressive, though a touch behind the coax output again. However, AES does come with the advantage that it should be galvanically isolated. So whether Coax or AES is preferable is probably just going to come down to your dac or personal preference.
Unfortunately I can’t measure the I2S output as the APx555 does not support differential/LVDS I2S, but as shown in my DDC video, the performance is absolutely stellar, and it did show measurably lower jitter in a J-Test than the already great performance coax output did. If you have I2S, definitely use it. The DI-20HE also features an adjustable I2S pinout so you should be able to use it with any DAC.
The second issue that a good DDC or Streamer will seek to address is electrical noise. You don’t want noise from your source causing your DAC to perform poorly. Noise can have a direct, audible effect, such as hearing GPU-whine through your headphones/speakers, or it can have indirectly-audible effects. For example, causing clocks in your dac, or other circuitry, to perform sub-optimally.
Some dacs are more immune to this, and some will even have full galvanic isolation to in theory prevent any noise getting through entirely. But many smaller dacs are much more susceptible to it.
Some interesting behaviour here. Audible band noise content is very low, but with a rising level leading up to and above 20khz.
Overall this is pretty good, though it’s not as good as the pi2aes or zen stream.
9 thoughts on “Audio-GD DI-20HE DDC Measurements – Wow….”
Very interesting. Looks like audio-gd made a proper product this time around at least 😛
Question. This is Topping D10B coax out:
doesn’t it look incredibly clean for the $140 it costs, comparing to your DI-20HE graphs? With a 5V mod you could probably control the noise of it really well, too… might grab this one.
Can’t really tell anything from that as he’s not actually playing anything through the device. The device is idle and so jitter performance there isn’t comparable to what I’m testing.
The asr graphs showed much more problematic results with it I think
Ah. I thought these graphs were comparable. I see..
Would be cool if you’d get one of these for measurement tho, pretty cheap after all 🙂
I wonder why you didn’t include the SOtM units I can see on your desk. I assume they are the SMS-200 ultra and the txUSB ultra (or maybe one is the SPS-500?) I would like to see a review on these.
Those are the sms200 ultra and SPS500. I no longer have those as I sold them to help pay for the analyzer.
But also. The sms200 is a usb streamer. Its not a synchronous protocol streamer (i2s, aes, coax spdif) and so jitter is not present like it is on synchronous protocols. The dac is the master clock when using usb and the level of jitter will not be dependant on the USB source other than noise
Excellent info about this product and your ‘bits are bits…’ video is impressive – great explanation for a novice like me!
I’ve purchased a DAC with AES and coax input and want to connect 2 x coax sources and 1 x USB. The Di20 looks like a high quality option to do this.
Can you connect all three sources to the USB, Coax and BNC clock inputs and then to the DAC using either the AES or Coax outputs?