• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Review and Measurements of Budget ($20) DACs

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,386
Location
Seattle Area
Working through my backlog of stuff to test, I finally got to do some preliminary testing of two budget DACs:
1. FiiO Taishan
https://www.amazon.com/D03K-Digital-Analog-Audio-Converter/dp/B009346RSS


2. SIGNSTEK Mini USB DAC
https://smile.amazon.com/dp/B00FEDH...00-other-smile-us000-pcomp-feature-scomp-wm-4

My reference for comparison was the iFi iDAC2 which retails for $349 or 17 times more money.

FiiO does not have USB input. To feed it audio I used the Coax output of iFi iDAC2. SIGNSTEK does have USB input so I used that as a complete system. It also has Coax input but I could not get it to work. Nor could I find a manual for it.

I have not bothered to measure frequency response. Instead I focused on just running a couple of high frequency tones through them. One was a 14 Khz tone at 48 Khz and the other was 11.05 Khz at 44.1 Khz.

All measurements are performed using my Audio Precision analyzer using its unbalanced inputs. $5,000 AC cords were used on the AP to reduce its noise level (not really). FFTs spectrum analysis is average of eight (8) runs to lower the instrument's own noise floor.

Let's look at the 44.1 Khz response first:

upload_2017-1-31_21-33-20.png


Please note that the analog output of these DACs is not matched. Specifically the SIGNSTEK has much lower output. In that regard, its output needs to be shifted up and its noise floor would then be the worst of all three. It also has more spurious peaks. There a few around our central tone and a couple further out.

FiiO output is surprisingly good, essentially matching the iFi DAC2.

Stepping back, this is excellent performance for all three really. Even the SIGNSTEK produces distortions that are 100 db below signal level.

All three have that odd second harmonic at 22.1 Khz. I thought that may be in my file but analysis using Audition did not show it. It is above audible level so other than bothering your tweeter, it is harmless.

Next I tested 14 Khz tone using 48 Khz:

upload_2017-1-31_21-38-1.png


Good grief! Clearly neither SIGNSTEK or FiiO have proper 48 Khz clock. SIGNSTEK's distortion products are just 50 db below its signal peak!

FiiO is 10 db better with distortions 60 db below peak.

Conclusion
If you are only playing CDs these $20 DACs perform admirably. I expected garbage output but that is not what I saw at this sample rate.

At 48 Khz (and likely its multiples) these are garbage products. Wonder if there is a step up product that has proper 48 Khz support at not too much extra money.

Acknowledgements
Special thanks to members watchnerd and RayDunzi for recommending these products for test.

Note: measurements can easily show wrong data courtesy of my aging brain and poor user interface. So I welcome any corrections or comments from manufacturers and members.
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,198
Likes
16,981
Location
Riverview FL
Since the reason I bought the Signstek was to convert USB to Optical or Coax (to get a signal from REW on Laptop with only USB out into a DAC with only Optical or Coax in), could you take a look at that too?

Laptop USB -> Signstek -> Optical or Coax Out (or try both) -> a decent DAC -> Analog (measurement point)

For grins, of course.

44.1 and 48kHz of interest to me.
 
Last edited:

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
My guess is the 22.1 kHz tones are the images of the 11.05 kHz input signals when sampled at 44.1 kS/s. It is just over Nyquist (22.05 kHz) and thus practically impossible to completely suppress (need a really steep filter with almost zero transition band). Probably goes away at 48 kS/s.

Nit-picking: k for kilo- is not capitalized; H in Hz is...
 

Jinjuku

Major Contributor
Forum Donor
Joined
Feb 28, 2016
Messages
1,278
Likes
1,180
When I see these results you stop wondering why a well appointed $200 DAC can acquit itself nicely.
 

Palladium

Addicted to Fun and Learning
Joined
Aug 4, 2017
Messages
630
Likes
769
When I see these results you stop wondering why a well appointed $200 DAC can acquit itself nicely.

Q: What's the point of a $200+ standalone DAC which doesn't even outperform the integrated one in a $200 phone with a zillion other functions?
A: Oh because I have pitifully low standards and also hate money.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,386
Location
Seattle Area
I realize that my test is not "apples to apples" comparison with yours, would love to see what the J-Tech "premium" would do in your test suite. Great site, I'll enjoy hanging out here!
Welcome aboard and thanks for the measurements and kind words! Do you have a link for it? I may buy and test it to compare.

Also, take a look at the Speaka thread as we found good performing DACs there too:https://www.audiosciencereview.com/...-speaka-usb-dac-and-headphone-amplifier.2423/
 

Wayne

Active Member
Forum Donor
Joined
May 26, 2017
Messages
172
Likes
46
Location
Los Angeles, CA
Next I tested 14 Khz tone using 48 Khz:
Good grief! Clearly neither SIGNSTEK or FiiO have proper 48 Khz clock. SIGNSTEK's distortion products are just 50 db below its signal peak!

FiiO is 10 db better with distortions 60 db below peak.

Conclusion
If you are only playing CDs these $20 DACs perform admirably. I expected garbage output but that is not what I saw at this sample rate.

At 48 Khz (and likely its multiples) these are garbage products. Wonder if there is a step up product that has proper 48 Khz support at not too much extra money.

@amirm I am confused as to what the graphs are showing and why the difference between 44.1 & 48 kHz, Would you please explain? Thanks
 

Mike S

Member
Joined
Mar 15, 2018
Messages
24
Likes
4
Welcome aboard and thanks for the measurements and kind words! Do you have a link for it? I may buy and test it to compare.

Also, take a look at the Speaka thread as we found good performing DACs there too:https://www.audiosciencereview.com/...-speaka-usb-dac-and-headphone-amplifier.2423/

Here's the link:

https://www.amazon.com/J-Tech-Digit...03HG/ref=sr_1_1?ie=UTF8&qid=1521260297&sr=8-1

BTW, I also purchased a Topping D30 based on your rec and after visiting their site. Looking forward to trying it out...I sometimes need a portable DAC that is as good as my 1820m and it appears to have the right parts in it to do that.
 
Last edited:

Mike S

Member
Joined
Mar 15, 2018
Messages
24
Likes
4
I thought my (now deleted) measurements of the J-Tech were too good to be true and it turns out I had my soundcard mixer settings wrong...my apologies. I will do it again correctly and repost them.

That said, the J-Tech has one "budget" quality for TV > good sound system usage that turns out to be useful: it has fairly muddy highs. LOL, this sounds like a bad thing but with all of the compression artifacts in broadcast and streamed digital sound, it serves as a filter. Case in point: I patched a D30 in (which is bright and clean to my ears) and the compression artifacts were unbearable. And I have 1Gig fiber service so its not some sort of auto compress due to limited bandwith issue. I literally thought the new DAC was oscillating above 15k until I checked further! The J-Tech definitely has pedestrian highs, but overall it has a smooth sound for "hi-fi" TV use imo. I didn't expect a decent DAC to be a downgrade from a $22 wonder, but for TV use it definitely is imo...
 

Mike S

Member
Joined
Mar 15, 2018
Messages
24
Likes
4
Here is a replacement post with "real" test findings for the J-Tech (again my apologies for posting wrong stuff). It comes out about how my ears expected: I was guessing -6dB at 20kHz with 10kHz corner... Was hearing no real noise to speak of and only limited distortion. The far left is the 1820m used to measure the J-Tech, old as it is, it still has solid numbers and sounds great. Both support 24 192 but I could not get RMAA software to work at that rate for some reason.

JT specs.JPG


Since they market the J-Tech as suitable for use with TV to stereo, I'm guessing the high requency rolloff might be intentionally built in after hearing what the D30 sounded like with its ability to faithfully reproduce compression artifacts in all of their glorious splendor...(I thought a new D30 was oscillating). I guessed the JT was -6dB at 20kHz with a 10K corner (i.e. where the response is -3dB). Got very close to actual, and this appears to be an intentional design to me. It definitely sounds better (less bad) for streamed TV sound.

JT-freq.JPG



The JT has pedestrian THD, which was actually pretty good a couple of decades ago. This is 24 bit 48 kHz but they all look about the same:

JT-thd.JPG


Similarly the noise is pedestrian by todays standards but actually pretty decent in real use. Even the 60Hz bump is -112dB:

JT-noise.JPG


The JT can be overdriven into distortion quite easily, though I have not heard this in hours of TV listening:

JT-thd-vs-freq.JPG


If you need something to hook a TV to a decent sound system, the JT is a keeper for $22 IMO. Seems to be designed for that purpose (the HF rolloff) and actually produces a pleasing pseudo-hi-fi sound for the artifact filled audio that comes in NetFlix et.al... Everything JMO of course, but hope it helps with making informed choices.
 
Last edited:

Mike S

Member
Joined
Mar 15, 2018
Messages
24
Likes
4
Wow, that is pretty steep high frequency roll off. I don't think I have ever seen a DAC do that.

Neither have I, and thats why I think its intentional. If you have a DAW just construct a LP filter with the same response and listen to material with lots of compression artifacts thru a good DAC and the filter. After hearing that, its pretty easy to imagine a descision to do a LP in something targeted at that market, jmo of course. Most people can't even hear 20kHz to begin with, the nastiness is about 12-16 kHz which most can hear and that response tames it for really cheap. jmo of course....
 

Mike S

Member
Joined
Mar 15, 2018
Messages
24
Likes
4
In the FWIW department, Cirrus specifies a LP filter for the CS4354 part used in this DAC. The mfg has altered the values since spec has 470 ohm / 2.2 nF and this board has 5.6K / ??. Given the unbuffered nature of the LP filter and the fact its over an order of magnitide higher impedance than spec, I retested into a high impedance imput that you would expect on an RCA input preamp and the rolloff measured -3.5 dB at 20kHz. No other measurements changed to speak of. So its either a case of intentional rolloff or Chinese mfg using "whatever" parts on hand. I vote for the former since its too much a coincidence otherwise.


CS4354.JPG
JT-board.JPG


The oscillator for the 8416 is on the other side of the board.
 
Last edited:

Mark S.

Active Member
Joined
Sep 24, 2018
Messages
124
Likes
46
Location
Somerset, England
Hi all,

post no.1, having found this thread when searching for some hard test data on the FiiO D03K.

From amirm's and Mike S's measurements, are we to take it that the FiiO is hardware re-sampling 48KHz to 44.1?

And that it does the same with 96KHz/192 KHz streams, hence isn't actually a 'hi-res' DAC?

cheers,
Mark.
 
Last edited:

Mark S.

Active Member
Joined
Sep 24, 2018
Messages
124
Likes
46
Location
Somerset, England
Well, in answer to my own question;

yes, it does look like the D03K turns everything it's sent into 44.1KHz before conversion, and it's actually implicit in their own frequency response specs.

No such thing as a free lunch, as they say.

ETA >> this was all brought about because I bought a D03K recently during a 'hiatus' while I have no source other than the on-board Realtek codec chip on my Asus P8Z77-V mother board - I immediately noticed how 'hazy' the sound of the FiiO was with hi-res files (pretty good near-field system and Senn HD580's with a decent CMoy), even compared to the rather low bar set by the latter.
 
Last edited:

Mark S.

Active Member
Joined
Sep 24, 2018
Messages
124
Likes
46
Location
Somerset, England
I'd been using Foobar and Kodi with WASAPI pass-through, but given what the FiiO seems to be doing, I've switched Kodi to DirectSound with Windows configured to 24/44.1, and in Foobar I now have the Sox resampler converting everything to 44.1 to to avoid whatever nastiness the FiiO might otherwise be inflicting on non-44.1 streams.

Also just made a battery power supply for it, 4 x D-cell NiHD (not ideal - they need to be c. 20% discharged to get the voltage down to what I consider safe).

Sounds very good indeed with all my Red Book music, and certainly no worse with other sample rates with either Windows or Sox doing the down-sampling.
 
Top Bottom