Review and Measurements of WesionTEK Khadas Tone Board DAC

[BYRTT]

Regarding what PSU route is best for Khadas TB its probably good enough have peace in mind with some overdoing here and there : ) that said i'm not shure designers didn't take care some USB ports can be crap and designed unit to be more or less immune, pages back there was a few trials shared about external power supply use where one of them included pictures and none noticed any audiable effect, so quistion is if any low noice LT3045 is really needed else Khadas guys probably had implemented it from start on. Think it looks ab fabric be a high performance design seen in Amir's measurements for naked Khadas TB and boxed March Audio dac1 plus myself tested boxing process looking into a high resolution FFT and there was not a piece of grass difference between naked board / boxed floating board non GND / boxed GND via one brass standoff / boxed GND via two brass standoffs / boxed GND via three brass standoffs. Only thing FFT test revealed was far down low 50Hz spike plus its harmonics from power line at left side of 1kHz stimulus was present when Laptop was powered from external power line and switcing to batteri left side was clean, but that can change from Laptop to Laptop if they don't add other noice spectra powered form batteri. Some day will try external power myself but in all objective data seen so far looks so great and sound is sensed excelent powered via USB buss is a bit relaxed it will make a difference.

 

[Herbert]

That´s true but sometimes the USB port is simply not needed and it faces into the same direction like the RCA-jacks.
Good if you plan to use USB, bad if you simply do not need it besides firmware upgrades -
because for a DIYer, the square hole for the USB in your self-made-case is not as easy to drill as the holes for the RCA jacks.
So no matter whether you use good old 7805 regulator or super-super TPS7A4700, 5 & 21 can become handy...

 

[maxxevv]

Thank you for taking the time to do the AB test
I am hoping the Khadas will have better depth, because even though I like BAL out of DA3, when I use SE out, depth gets totally screwed. First, I was thinking, yea, the balanced sounds really good, but as I don´t have balanced amp, I use my full-size headphones 90% with SE out through my Atom. The depth of DA3 on SE is terrible, making almost everything sound "in your face" with only separation going left & right, like only 1 layer of the sound if you know what I mean.. I hear this not only with my Edition X, but also with iems (FH5, Noble X) and I´m really hoping this is just example of bad implementation of SE out on DA3, because for example Mojo excells at this dept-layering even on SE out..

Actually, you should have no problems running your 2.5mm TRRS to split into the L / R RCA inputs of the ATOM.

TRRS is 4 pole anyway, not (3+3 = )6 pole like XLR. So you're outputting to 2x2 poles of the RCA. Not an issue.

The reason you probably have issue with the SE is the very low power output. Its probably not enough to drive the input sufficiently.

 

[jaynyc]

@Herbert @BYRTT do you have any advice of a suitable finished external power supply for the Khadas board? (assume using SPDIF input only)

 

[Nango]

My guess: https://www.ebay.com/itm/TeraDak-DC...3-A-A-B-B-3-Linear-Power-supply-/222531593968

 

[BYRTT]

@Herbert @BYRTT do you have any advice of a suitable finished external power supply for the Khadas board? (assume using SPDIF input only)

Myself intend try a diy PSU with good old LM317 or 7805 because from graphs so far it looks board is so well engineered and and as for now running USB buss power from laptop it simply sounds more than good. Have anybody noticed board itself feels quite heavy even components is so tiny a part/volume, would guess there is a ton of copper layers and clever layout routes into that PCB to get it weight so relative much.

About SMPS with USB C power, think seen somewhere Amir recommend the ones from Apple is good performers and rather low cost, so for SPDIF input only this is probably a good and easy solution and also a unit that is pratical in it can probably be used for many other tasks/gear.

The one Nango link to above is probably a high performer but also quite expensive.

If one really like to feed Khadas TB clean power can't stop think about Jan Didden's silent switcher either feeded via USB but probably better as Jan do for his Autorange LAB unit feed silent switcher from a power bank to get DAC 100% isolated from mains, there is graphs over there that look quite stunning in comparison to traditional stuff (LINK https://linearaudio.nl/silentswitcher ).

 

[unklez]

I'm getting very bad results with the Khadas using Rightmark audio analyzer vs my onboard audio on the PC. Any idea what's going on guys?

 

[Veri]

I'm getting very bad results with the Khadas using Rightmark audio analyzer vs my onboard audio on the PC. Any idea what's going on guys?

You are testing 16-bit, any reason for that? Can you use 24-bit, 32-bit? That should help results for one thing..

 

[unklez]

You are testing 16-bit, any reason for that? Can you use 24-bit, 32-bit? That should help results for one thing..

Don't think it changed much but after lowering microphone input level I got proper results.

Also I changed USB-C cables from Khadas to Blitzwolf and got better results each time. Left side on screenshot is Khadas usb-c cable and right side Blitzwolf.

Even different USB ports has effect on measurements. :S

I'll measure tomorrow with SPDIF when my toslink to coax convert arrives

 

[JohnYang1997]

Don't think it changed much but after lowering microphone input level I got proper results.

Also I changed USB-C cables from Khadas to Blitzwolf and got better results each time. Left side on screenshot is Khadas usb-c cable and right side Blitzwolf.

Even different USB ports has effect on measurements. :S

I'll measure tomorrow with SPDIF when my toslink to coax convert arrives

A few questions.
Are all devices using asio/wasapi?
Have you tried other software like rew or arta? Spectrum says better than just a number.
Is the output level of khadas making your input distorting?
Tried another computer? like a battery powered laptop?

 

[unklez]

I don't take my results serious as the input device isn't up to task but however the Khadas measurements should be at least better than the onboard Realtek soundcard? Btw my onboard soundcard also has a ESS sabre dac. Mobo is a Asus z390 maximus code.

I reckon I have some grounding problem going on causing noise and stuff. I'll measure tomorrow with SPDIF hoping to see gains.

 

[Herbert]

Could we do a roundup on the ESS hump / intermodulation distortion vs level?

Why the hump probably happens was explained here

https://www.audiosciencereview.com/...hadas-tone-board-dac.4823/page-19#post-122993

and here someone states it as being audible in comparison to a Wolfson DAC

http://www.itsonlyaudio.com/audio-hardware/wm8741_vs_ess9023/

But what is the bottom line?

Definately audible?
Can the hump be adressed with firmware updates or minimized
with certain filter charactaristics?

Thanks,
Herbert

 

[JohnYang1997]

I believe everything is audible. Apart from my opinion. It's better to assume everything is audible and tackle it instead of omitting/overlooking it. Sure it is probably not our responsibility or need to correct this phenomenon but it is what it is.
Good old cs4398 when implemented correctly can produce -120db thd at a lower level. And doesn't have weird behaviour of more modern chips especially from ess. I'm not bashing them. For such advanced designed chips it's very likely that we still don't quite see some of the potential behaviors and potentially do something to the sound.

 

[caguilar91]

From what I've investigated, distortions aren't that audible until above 1% for THD and in certain cases not even until above 20% for mid-bass at 280 Hz and lower, not to mention 20 and 40 Hz:

https://www.axiomaudio.com/blog/distortion

And 0.5% of Transient Intermodulation Distortion seems to be perceived momentarily:

https://ieeexplore.ieee.org/document/1170453

In reality, most distortions shouldn't be easily audible below 0.1% and DACs and amps usually aren't the worst performers of the chain, but the headphones or speakers which are the ones that measure the worst.

 

[somebodyelse]

and here someone states it as being audible in comparison to a Wolfson DAC

http://www.itsonlyaudio.com/audio-hardware/wm8741_vs_ess9023/

That post seems more about the huge rise in IMD above -1.5dB FS than about the hump. That isn't shown in the measurements of the Tone Board, or most other ESS DACs. It is similar to the SMSL SU-8 where the output stage saturated on the original version, and was fixed in version 2. My guess is that the 'dirt cheap' ESS board he's testing has a similar output stage design flaw that's to blame.

 

[unklez]

I got the toslink to coax converter and did some test. Left side in screenshot is USB and right side is SPDIF. I'm using Asus z390 motherboard -> Khadas -> Fiio K5 amp.

SPDIF wrecks it. Might try a ifipurifier3 usb stick to test USB like that.

 

[unklez]

I had the Toslink to Coax converter connected to the computer USB in the above test. After connecting the toslink converter to a USB wall charger much better result.

 

[BYRTT]

@unklez,

Thanks those interesting graphs where SPDIF on paper looks better than USB and even better when using external PSU for SPDIF converter, guess for test loop you use same computers input of Realtek soundcard and can't stop thinking how it would look on a electrical input not sharing that same computers PSU, think balance the two between for your case is larger than what Emir's comparison showed.

Real world also little better numbers should be expected using ASIO or WASAPI than direct stream, example for that is my old PCI card below using its SPDIF I/O (none DAC) loopback in a 96kHz rate test, the one at left is ASIO and right one is direct stream and while REW can't use WASAPI will guess or expect it follows spirit of ASIO graph.

 

[Megatronic]

Hello. Has anyone had difficulty ordering ktb through official site? A week has passed, and the order has not been sent. Through the mail they do not respond.

 

[BYRTT]

@Megatronic,

Can't see your location via profile, but should you be in EU region its easy order at Audiophonics France so customs and VAT is taken care of and they have units in stock for now : )