• 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!

Three USB to S/SPDIF Converter Measurements: Audiophilleo, iFi iDAC2, SIGNSTEK

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,393
Location
Seattle Area
Here is a little comparison of three different ways to convert USB to S/PDIF all driving an ultra low cost, $20 FiiO Taishan DAC. It is the only DAC on my workbench that takes S/PDIF input as the rest are all USB and I am too lazy to measure my main system :).

A simple test tone at roughly 11Khz is output from the PC using Foobar2000. I had to use the Directshow filter as SIGNSTEK would not work otherwise. So all three are going through the kernel audio stack. Windows audio output format was selected to be 16-bit/44.1 Khz.

The first device measured is my Audiophilleo USB to S/PDIF converter/bridge. I bought this from Phil back in 2010/2011 when he had first started the company. It retailed for $500 which at the time was a very reasonable price since async USB was a rare bird.

Audiophilleo.png


The test tone is the near 0 db line on the left. In an ideal situation there would be nothing else after that. But this is not ideal and we have harmonics of that tone going up way high in frequency. Your amplifier and speakers may "hear" those tones but your ears would not. In the audio band of 20 to 20,000 Hz, there are no anomalies.

Note that we are testing the combination of DAC fidelity and S/PDIF input.

Next I add the measurement for SIGNSTEK Mini USB DAC which is only used as a USB to S/SPDIF converter (its DAC ignored). Here is what we get:

Signtek.png


We see that SIGNSTEK has caused the DAC to have new distortion and noise products some of which are in the audible band (see the red spikes). The level though is at -100 db in the audio band though so it really is not an audible concern especially when we consider that this little device retails for $20 or so.

What we do see here is that bits are NOT bits. Two different digital inputs to the same DAC produce two different analog outputs.

Now let's use the S/PDIF output of the iFi iDAC2 DAC (again, ignoring its own internal DAC), driving the same Fiio Taishan:

iFi iDAC2.png


Whoa! What happened here??? This $300 device is causing our little DAC to put out far more ultrasonic noise! It also has highly increased the amplitude of our harmonic distortion products!

Its saving grace over the SIGNSTEK is that its output in the audible band is just as clean as Audiophilleo. Still, this is not good engineering especially coming from a company that touts the quality of its products, says they have measurement gear, etc. Clearly it is spitting out lots of noise over its S/PDIF port that is polluting the FiiO DAC.

So there you have it. Well engineered products like Audiophilleo pass the test of time. Buy good stuff and it will last you. Go for hype without measurements and you get hit on the face like with the iFi.

Also this validates my saying that the best digital tweak is getting the best USB/SPDIF converter. It makes even low cost DACs perform better (measurably).

As always, I welcome correction, feedback and comments. Testing audio gear in spare time is always prone to errors and I am happy to correct anything that is found.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,393
Location
Seattle Area
Oh, here is all three graphs overlaid:

All three.PNG


Edit: and addition of Melodious-Audio MX-U8 compared to Audiophilleo:

Melodiouis Audio MX-U8 vs Audiophilleo.png


As you see, the performance is identical which tells us that we have reached the limit of the performance of the FiiO Taishan DAC.
 
Last edited:

Sal1950

Grand Contributor
The Chicago Crusher
Forum Donor
Joined
Mar 1, 2016
Messages
14,073
Likes
16,607
Location
Central Fl
Great stuff Amir, even I can understand this.
Thanks.
 

Sal1950

Grand Contributor
The Chicago Crusher
Forum Donor
Joined
Mar 1, 2016
Messages
14,073
Likes
16,607
Location
Central Fl
Dumb question I know but
If I was using one of these devices to create another s/pdif source from a USB out, it would retain the sources data rate. IE 16/44 in is 16/44 out and 24/96 in is 24/96 out?
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,393
Location
Seattle Area
It would only do that if the device has either WASAPI or ASIO interfaces. Then you can bypass the Windows audio stack. Otherwise Windows will convert all the samples to high-resolution internally and then convert back to a single bit depth and sample rate which is not good. Audiophilleo and iFi both have such drivers but SIGNTEK does not. That support though is very common these days but just check to be sure.
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,524
Likes
37,057
I also have an Audiophilleo. Very nicely designed device. Have a friend's original VLink in my possession currently as well. Don't have an AP, but do have one ADC that goes to 192 khz. When I get some time I'll do a similar test on these just for added data.

I wonder how different the result might be with a different DAC. I should be able to round up a couple DACs to see if they vary in this regard. Am I right to think near 11 khz was 11,025 hz?
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,393
Location
Seattle Area
Love to see more data. I do have some measurements I did with my higher performance DAC in this article: http://www.audiosciencereview.com/f...performance-pc-server-interfaces-async-usb.8/

index.php


This is comparing the Berkeley Alpha USB against Audiophilleo. As you see the Berkeley does even better.

Please note that these are measured using a different instrument: the Prism Sound dSound analyzer so the graphs are not comparable (and use different analysis parameters).
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,393
Location
Seattle Area
Okay, any significance to that particular frequency?
No, I made a typo when I created it :). Wanted it to be quarter of the sample rate as you guessed. :) :) It nevertheless shows the issues so I am living with it for now.
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,201
Likes
16,983
Location
Riverview FL
Will 48khz conversion create additional problems in the audio band as it did when you tested the DAC output of the SIGNSTEK?
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,393
Location
Seattle Area
Will 48khz conversion create additional problems in the audio band as it did when you tested the DAC output of the SIGNSTEK?
I assume it would because I already tested the same DAC with iFi's output and it was pretty bad so I did not go there for this work.

I don't have another DAC near my workbench without that problem unfortunately so can't test 48 Khz.
 

Sal1950

Grand Contributor
The Chicago Crusher
Forum Donor
Joined
Mar 1, 2016
Messages
14,073
Likes
16,607
Location
Central Fl
It would only do that if the device has either WASAPI or ASIO interfaces. Then you can bypass the Windows audio stack. Otherwise Windows will convert all the samples to high-resolution internally and then convert back to a single bit depth and sample rate which is not good.
I don't do Windoz, remember. ;)
My Linux config of Clementine feeds a bit perfect stream out USB at its original rate whatever that is. So I take it the answer is that the converter does not resample its input to some arbitrary rate?
 

Mivera

Major Contributor
Joined
Mar 4, 2016
Messages
2,322
Likes
97
Location
West Kelowna
Why didn't you include the USB to SPDIF bridge I sent you in the test?
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,393
Location
Seattle Area
Why didn't you include the USB to SPDIF bridge I sent you in the test?
'cause your stuff is still in the box! You just reminded me about them. Do you want me to test them or send them to you? If the latter please PM me with the address and I will ship it out. Not sure how to apologize for keeping it so long! :)
 

Mivera

Major Contributor
Joined
Mar 4, 2016
Messages
2,322
Likes
97
Location
West Kelowna
Well you might as well test as you already have them. That's why I sent the stuff . It's only been a year now :) Might as well just add the same measurements of the SPDIF converter to this thread.
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,524
Likes
37,057
5624-close-up-portrait-of-a-beautiful-brunette-pv.jpg


In a nod to Thomas' suggestion. Creative Commons licensing on this one.
From here:http://www.freestockphotos.biz/stockphoto/5630

Okay, on to the measurements.

I used a Tact RCS 2.0 for one DAC, and a TC Impact Twin for another. Each was fed via an Audiophilleo 2 USB to SPDIF converter over coax. Then via Musical Fidelity Vlink first over coax and then over Toslink. The ADC is a Focusrite Forte.

Here is the only difference I found. In the noise floor of the Vlink are a few very low level idle tones around 100 hz spacing either side of 1 khz for about 1 khz. Red is the Vlink and blue is the AP2. This extra grass was present over both coax and toslink in the Vlink.

AP2 silence vs Vlink silence log scale.png


BTW, here is the noise floor of the ADC. It has rising noise starting at 50 khz and above. It also has a couple idle tones. 30,700 hz and twice that frequency. Level is around -125 db for both tones.

Audacity silence.png


I found my Tc IMpact Twin had a pair of tones at low level spaced 1500 hz apart in one channel a few places below 20 khz. The other channel was clean. I am using the clean channel for these measurements.

Other than that just not much going on. The TC Impact Twin was 3 or 4 db quieter in noise floor as I could connect over balanced cable to it and over unbalanced to the Tact. I see a few low level spurious tones show up when a high level signal is present. Those were the same frequency and rough level with any combination of devices so it seems intrinsic to the ADC I used.

Here is an overlay of the Tact AP2 in red and Vlink in blue playing 11,025 hz tone. This is a snapshot of realtime FFT so the little peaks bounce around 3 db or so. The small differences you see fluctuate, but amount to nothing.
Tact AP2 and Vlink.png


Here is an overlay of the Tact in red and Impact Twin in blue.
TactAP2 vs TCIT AP2.png


Here is an overlay using -4db white noise for the input. Red is Tact with the AP2 converter while blue is the TC Impact Twin being fed the same signal via Vlink Toslink. Just not a whole lot between them.

Tact AP2 vs TCIT Toslink noise.png


So basically other than some very low level differences of no consequence the moral of the story might be stay away from $20 DACs (yes it is still possible to go too cheap) and maybe be wary of iFi gear.
 
Last edited:
Top Bottom