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

[amirm]

But why test USB->S/PDIF using a superior analyzer and an inferior DAC ?

Because it is the only DAC I have on my bench that takes S/PDIF. My two other higher fidelity ones are USB only. And I don't want to tear up my music system to use that DAC.

I am interested in buying another DAC that lets me do this testing. See my other thread on Bargain DACs.

 

[Blumlein 88]

Some more upclose looking at the result I had from the Tact RCS 2 and TC Impact Twin. There are some small differences in the two SPDIF converters. There is really good consistency between the two DACs. BTW I played all my files in Foobar 2k using WASAPI. So Windows issues don't crop up.

First up I am showing an overlay of a 128k fft looking at a 1000 hz band centered on 11,025 hz sine wave. You'll see in close the Vlink surprisingly has a sharper peak which presumably means better close in timing. Look close and you can see the clock in the Vlink is closer to the ADC clock and faster than the Audiopilleo by 130 ppm or so. The difference in Coax and Toslink is a few spurs in the Toslink that are absent in the coax connection.
Red is Tact fed via Audiophilleo
Blue is Tact fed via Vlink coax
Green is Tact fed via Vlink toslink.

Next is the TC Impact Twin. Everything is surprisingly close to the same.
Red is TC fed via Audiophilleo
Blue is TC fed via Vlink coax.
Green is TC fed via Vlink toslink.

Here is an overlay of Tact and TCIT both fed from the Audiophilleo SPDIF converter.
Tact in red.
TC Impact Twin in blue.

Tact and TC IMpact Twin both fed from Vlink SPDIF converter over coax cable.
Tact in red.
TC Impact Twin in blue.

Finally Tact and TC Impact Twin both fed from Vlink SPDIF converter over toslink optical. You can see in red the extra spurs using the Tact that aren't there with the TC Impact Twin.
Tact in red
TC Impact Twin in blue

 

[Blumlein 88]

Okay investigating the Tact RCS 2.0 digital oddness. On the Tact 93.9 on the volume is actually full digital level. 99.9 is with 6 db of digital gain. Amir mentioned issues, and I have experienced them too. If you keep volume at 93.9 on the Tact you have no issues. It simply works like any good DAC.

I ran these tests again with all the connections I have posted earlier. Here I am showing only one set of results, but they were the same other than tiny differences mentioned in my other posts up thread. I adjusted the Focusrite Forte ADC -6 db to match resulting levels recorded.

First up is -4 db white noise at 44khz playback and 192 khz recording. Red is volume 99.9 while green is 93.9. Ultrasonic filtering or something is very different.

Here is 99.9 white noise, but now reduced to -24 db in the file. Note other than level difference it looks okay. Green is volume 93.9 while blue is volume 99.9

Here is a full scale 19,100 hz tone. Blue is volume 93.9 while red is volume 99.9. As you can see a whole grassy forest of spiky tones that should not be there at 99.9 volume setting.

Here is 99.9 playing 19,100 hz at a minus -20 db level. Compared to full scale 93.9 volume. Green is 93.9 and blue is 99.9 on the volume.

So pretty simple. Just don't use an RCS 2.0 Tact above 93.9 on the volume and it works like any good DAC. Go above that and odd things do occur.

 

[Mivera]

OK, here it is. The unit Mike sent me is the Melodious-Audio MX-U8. It retails for $270 on Amazon: https://www.amazon.com/Melodious-Ultimate-Digital-Interface-Ver-1-7/dp/B01CHXYAV2

It is a hefty little box:

I tested it with the same FiiO Taishan $20 DAC and compared the results to the best we had so far which was the Audiophilleo:

View attachment 5577

The result was identical performance with the two graphs right on top of each other. This says that this is a high quality implementation and that we are now limited by the performance of the FiiO Taishan DAC than the S/PDIF input to it.

Note that installation was a bit tricky. They just dump you in a folder with two versions of XMOS drivers. I tried the newer version and the installation would fail at the end. Then tried the older version that was also on the site and it worked.

Wanted to compare all of this to my Berkeley Alpha USB but unfortunately I can't get it to work on my laptop anymore . It completely screws up the audio samples on the way out. Maybe doesn't work well with Dshow interface anymore. Will mess with it later.

Thanks Mike for sending this to me.

Cool. No problem. I guess it wasn't a total waste of money. They are about 3 generations ahead now though

 

[Sal1950]

So pretty simple. Just don't use an RCS 2.0 Tact above 93.9 on the volume and it works like any good DAC. Go above that and odd things do occur.

D, Don't understand much of what I'm looking at, but your graphics have much purdier colors than Amir or Rays.

 

[RayDunzl]

D, Don't understand much of what I'm looking at, but your graphics have much purdier colors than Amir or Rays.

His charts above are in WaveSpectra - maybe WS151.

http://efu.jp.net/soft/ws/ws.html

 

[Blumlein 88]

D, Don't understand much of what I'm looking at, but your graphics have much purdier colors than Amir or Rays.

Yes Ray, WaveSpectra WS 151 (latest version to my knowledge). Free Japanese software with some neat features. Especially neat as some of them were not available in freeware back when it was first available. In some ways a bit hairshirted, but in others for particular purposes pretty nice. Windows software that runs very well in WINE. Help files being in Japanese is a bit of bother, but I think I know I how it works. Of course I mainly use it because I am a cheap a$$ and it is free. I still feel wronged that Adobe bought out CoolEdit, and didn't honor my paid for CoolEdit license for what is now Adobe Audition.

 

[RayDunzl]

WS has its charm, with its Wave Display and Lissajous, beyond the main display FR and THD features.

I moved to REW as it is more complete and I became more interested in in-room measurement, and even sent Mulcahy a donation a while back because I was feeling guilty. I may have to do that again sometime.

 

[March Audio]

Okay investigating the Tact RCS 2.0 digital oddness. On the Tact 93.9 on the volume is actually full digital level. 99.9 is with 6 db of digital gain. Amir mentioned issues, and I have experienced them too. If you keep volume at 93.9 on the Tact you have no issues. It simply works like any good DAC.

I ran these tests again with all the connections I have posted earlier. Here I am showing only one set of results, but they were the same other than tiny differences mentioned in my other posts up thread. I adjusted the Focusrite Forte ADC -6 db to match resulting levels recorded.

First up is -4 db white noise at 44khz playback and 192 khz recording. Red is volume 99.9 while green is 93.9. Ultrasonic filtering or something is very different.
View attachment 5588

Here is 99.9 white noise, but now reduced to -24 db in the file. Note other than level difference it looks okay. Green is volume 93.9 while blue is volume 99.9
View attachment 5589

Here is a full scale 19,100 hz tone. Blue is volume 93.9 while red is volume 99.9. As you can see a whole grassy forest of spiky tones that should not be there at 99.9 volume setting. View attachment 5591

Here is 99.9 playing 19,100 hz at a minus -20 db level. Compared to full scale 93.9 volume. Green is 93.9 and blue is 99.9 on the volume.
View attachment 5592

So pretty simple. Just don't use an RCS 2.0 Tact above 93.9 on the volume and it works like any good DAC. Go above that and odd things do occur.

Just wondering about this and your volume levels. You probably need to be careful due to ADC distortion when driving at high levels. You will usually see measurements taken at -60dB FS (ADC input) to overcome this.

If we are specifically looking for usb-spdif converter issues -60dB will probably be a more appropriate level to test at, should minimise any measurement ADC issues.

This is how my PCM4222 measurement ADC looks, good to about -18dB

 

[Blumlein 88]

WS has its charm, with its Wave Display and Lissajous, beyond the main display FR and THD features.

I moved to REW as it is more complete and I became more interested in in-room measurement, and even sent Mulcahy a donation a while back because I was feeling guilty. I may have to do that again sometime.

Did you know how to get it into analyzer mode? Some other perks hidden in that mode.

 

[RayDunzl]

Did you know how to get it into analyzer mode?

Maybe not...

 

[Blumlein 88]

Just wondering about this and your volume levels. You probably need to be careful due to ADC distortion when driving at high levels. You will usually see measurements taken at -60dB FS (ADC input) to overcome this.

If we are specifically looking for usb-spdif converter issues -60dB will probably be a more appropriate level to test at, should minimise any measurement ADC issues.

No the problem is not the ADC. It is fine and operating well within its abilities. It can handle much larger inputs with no problem. The problem is with the Tact in that it does something odd if you push the Tact volume to the point it has digital gain. Unless I misunderstand what you are saying.

 

[March Audio]

No the problem is not the ADC. It is fine and operating well within its abilities. It can handle much larger inputs with no problem. The problem is with the Tact in that it does something odd if you push the Tact volume to the point it has digital gain. Unless I misunderstand what you are saying.

No, just pointing out it can be an issue

sorry, what was the adc?

 

[Blumlein 88]

Maybe not...

Well when you pause a file being played it freezes the FFT of course and allows you to go forward or backward either by a sample amount or time increment. In Linux at one time this didn't work I think with version 1.4o. Anyway under play/rec in settings there was a check box to go into "analyzer mode" if a format or playback is incompatible. In Linux you would see Pulse or Alsa and Wavemapper. Choose wavemapper which wasn't actually available and it automatically dumped you into analyzer mode as soon as you opened a file. Where you can leave markers with comments and a few other goodies.

 

[Blumlein 88]

No, just pointing out it can be an issue

sorry, what was the adc?

The ADC was a Focusrite Forte. Little two channel USB recording interface. Has microphone, instrument, and line level inputs. Line level like much recording gear has plenty of headroom. Plus you can adjust gain as needed.

The issues I posted about regarding the Tact I have seen with a few different ADCs and other measuring gear. I never figured out what Boz was doing with the digital gain that was messed up. I just didn't go above the proper level once I learned about it. I have wondered how many people assumed it should just be put at max volume and used an analog pre-amp. Ran into several people who swore the Tact sounded terrible. Maybe that is why. While not playback sota it is by no means terrible when properly used.

 

[Wayne]

@ Amirm: I’m a newby who would like to reproduce the results you shared in this thread concerning the testing (evaluations) of different USB to S/PDIF converters. I am not in the same league as you and the others who contributed to the thread, but I would like to try; if you would assist me with step by step instructions. Maybe in the future I could actually return some test results.

The question is basically, what was your set up? Did you use two computers, one to produce the test tone that was sent out the USB and converted to S/PDIF and one to display the spectrum? What program did you use for the frequency analyzer? and get two frequency spectra on top of each other? And how did analyze the S/PDIF signal, convert it back to UBS and into the computer?

I notice you used Foobar2000 for the test signal, is this critical, or can the test tone(s) be generated by other test tone programs or even from iTunes, Amarra, etc? Also I am not familiar with the Directshow filter.

Any assistance would be appreciated. Thanks

 

[amirm]

Great to see the initiative Wayne. I will answer the easy part first, you can use any program or source you like to play the tone. You just have to make sure it is "bit exact." For example if you just use Windows Media Player to play those tones, it will go through the operating system which will modify it. I use Foobar2000 because it supports "ASIO" which bypasses the operating system modifications. iTunes also resamples audio so it is no good for this use. Amarra should be fine.

The bad news is that the analyzer I use is hardware based ("Audio Precision Analyzer") and retails for $25,000. What you see is the user interface which runs on Windows.

The not so bad news is that you don't need my fancy hardware to do this. You can just use any analog to digital converter (ADC) interface to capture the output of the first computer. You can record that output and analyze it using the free Audacity.

There is also software called RightMark which can do this in real-time, ie. not having to save the file first. I am not a big fan of it in the limited experience I had with it.

Blumlein, what software and hardware are you using for your analysis?

 

[Blumlein 88]

Blumlein, what software and hardware are you using for your analysis?

Firstly I don't care for Rightmark either though it has its purposes.

My hardware is a Focusrite Forte recording USB interface. It cost me about $350. There are other such interfaces that work fine. The Forte has a lower noise floor than most which is useful for this kind of testing. There are other such things from Focusrite, or Steinberg, or Behringer (even the $79 unit Amir reviewed could be used for this), or other brands that will do fairly well for $500 or less.

With these devices you communicate over ASIO. You will get better results recording with some DAW software (such devices include such software with them normally) or you can use Reaper. My reasoning for this is you are assured of getting 24 bit bit-perfect input to them via ASIO. You can record with Audacity using these same devices, but if you aren't very careful you'll get some conversion to 16 bit going that route which will corrupt best possible results. This is because Audacity cannot use WASAPI or ASIO for input or output.

For playback I too use Foobar. Depending upon the playback devices I use Foobar via ASIO or WASAPI. Both are bitperfect via Windows. You may have to mess with settings in Foobar preferences, but it is pretty simple and easy to do.

Once you have the test tones recorded I export them from the DAW as 24 bit wav files. You can then analyze them with Audacity or other software (Audacity is fine for this). Now most of my posted FFT's were done in WaveSpectra. It is free FFT software from Japan with an interface that reminds you of Windows 98, but it works quite well. I could send a copy as finding the latest English version is sometimes difficult. If someone has Adobe Audition it is good software for these purposes too.

I have a few tips about dealing with unlocked drifting clocks for some tests, but those are minor refinements.

One other possibility on the ADC. If you are using a desktop machine the Asus Zonar Essence STX PCI cards are supposed to be very quiet with shielding against noise in the PC. They claim and more or less achieve a 124 db SNR. They come in a few flavors most costing around $200. I haven't used one, but they should do the job.

 

[Wayne]

@amirm: You have overloaded my brain.... never knew digital audio was so complicated. I have lost the better part of a nights sleep trying to understand what you and Blumein 88 said. I need to take baby steps, if you will bear with me.

RE: For example if you just use Windows Media Player to play those tones, it will go through the operating system which will modify it. I use Foobar2000 because it supports "ASIO" which bypasses the operating system modifications. iTunes also resamples audio so it is no good for this use. Amarra should be fine.

1. If I play a test tone, how does the OS modify the tone? does it res-ample it? i.e. if test tone recorded at 96khz does it re-sample it to 44.1khz? or?
2. When you refer to ASIO, it appears that there are two types: "FL Studio ASIO" and "ASIO4ALL". Does it make a difference which is used?
3. Prior to my current setup (Computer (WIN10) > USB > S/PDIF > powered speakers I had WIN XP > external Creative Sound Blaster Card with a S/PDIF output, just curious, do you know was this a ASIO based setup?

I'll deal with WASAPI later, my wife wants me to shut down for the night.

Thanks again.....

 

[RayDunzl]

Amateur/Starter method:

Room Eq Wizard - free software - generates test tones and can measure results about as well as anything else.

In my case, an Optical output (or USB, but I don't use it) for PC output, UMIK-1 USB microphone for capturing sound in air ($100 or so), and just a long RCA cable from PC input to preamp output to capture electrical signal (but it is noisy with 25ft length).

I'd consider this a minimal starter system to get started and see how interested you are. No mic needed for electrical only.

The limitation of the above is capturing and analyzing the lowest energies due to noise/non-optimized interfaces. That's where you start decoding what Blumlein 88 is saying. But it is cheap, and works well enough for most of my purposes.

Another thing I use is Audacity (free software) - can generate and record - letting you examine audio waveforms.