Measurement section for the article 'Review of the FiiO E07K 'Andes' / Avinity USB DAC Mobile (& FiiO L7)'

Fancy graphs (measurements)

This article is part of a bigger article, 'Review of the FiiO E07K 'Andes'  / Avinity USB DAC Mobile (& FiiO L7)' and they can be enjoyed in a better way if read side by side. I therefore recommend to click this link here to read the first part as well. ATTENTION: if an engineer reads this and finds an error, is able to offer some additional, meaningful measurements or anything else, please notify me.

Headphone output

Fig. I. FiiO 'Andes', headphone out, four different headphones
with different impedances, USB +6 dB gain, volume setting 50

Fig. II. FiiO 'Andes', frequency responses, USB +6 dB gain, volume setting 50

Fig. III. FiiO 'Andes', channel separation, USB +6 dB gain, volume setting 50

Have a look at fig. II. where you can witness (or not) the effect several headphones, featuring impedances from 16-300 Ohm, have on the headphone output of the Andes. The small load of 16 Ohm of the Sony MDR-W08 is the most challenging, yet the FiiO doesn´t display any frequency deviations, indicating a fabulously low output impedance. Please note that I connected the Sennheiser HD-600 just for comparison, the E07K is unsuited for these demanding headphones (sounding thin, flat and 'tiny' driving them). Channel separation (fig. III.) is at a very good -70 dB for the 16 Ohm load and consistent over the whole frequency band. The chart (fig. I., pictured just for completeness) indicates distortions driving the smallest load, and, oddly enough, stronger intermodulation distortions on the 60 Ohm load than the 32 Ohm load.

Fig. IV. FiiO 'Andes', intermodulation distortion, USB +6 dB gain, volume setting 50

Fig. V. FiiO 'Andes', total harmonic distortions + noise, USB +6 dB gain, volume setting 50

Low frequency IMD (fig. IV.) are strongest with the 60 Ohm Koss PortaPro; I don´t know the reason for it and I´d be very glad if someone could explain it to me. If I understood it correctly, these are THD really, not related to the interaction of the 60 Hz / 7000 Hz testtones. Much more interesting should be the intermodulation products above 7 kHz which are below -100 dB. Total harmonic distortions (fig. V.) are below -90 dB at all times, this should be inaudible. Generally, odd order distortions are dominating, if they´d be audible you´d probably listen to a more aggressive and crisper sound signature.

Line-out (using FiiO L7)

Fig. VI. FiiO 'Andes', RMAA interpretation chart, line-out, 96 kHz, +6 dB gain, 0 dBfs

Fig. VII. FiiO 'Andes', frequency response, line-out, 96 kHz, +6 dB gain, 0 dBfs

Fig. VIII. FiiO 'Andes', noise floor, line-out, 96 kHz, +6 dB gain, 0 dBfs

Fig. IX. FiiO 'Andes', THD + noise, line-out, 96 kHz, +6 dB gain, 0 dBfs

Attaching the FiiO L7 one gains a true line-out for the DAC function. The voltage level of this output is at supposedly 1.35 Volts. As you can see on the chart where RMAA interprets the measured values (fig. VI.) the performance of the FiiO E07K 'Andes' is excellent throughout when used as a DAC. Frequency response (fig. VII.) misses only 1 dB at 20 Hz and 0.2 dB at 20 kHz, inaudible and certainly not responsible for its sound signature. All the same, I would love to know what those distortions from 120 Hz to 4 kHz are (fig. VIII.). They are present on the measurments for the headphone out too (are they USB related?), only buried by a higher noisefloor. Speaking of the noisefloor... it is at a respectable -108 dB, the Wolfson DAC has a typical signal-to noise ratio of 116 dB for 96 kHz but it isn´t the responsible part here; the operational amp MX97220 with its SNR of 112 dB is. Considering this I´d say that -108 dB are pretty damn good.

FiiO ASIO driver

Fig. X. FiiO 'Andes', RMAA interpretation chart, ASIO driver
configured to 44.1 kHz, playback at 96 kHz

Fig. XI. FiiO 'Andes', frequency response, ASIO driver
configured to 44.1 kHz, playback at 96 kHz

The RMAA interpretation chart doesn´t look too bad (fig. X.). Upon closer look however it is revealed that the ASIO driver resamples when playing a stream it hasn´t been configured for (fig. XI.). In this case, a 96 kHz signal is resampled to 44.1 kHz. Suspiciously, the resampled frequency response looks exactly like a response caused by the bad resampler built into the Windows audio engine which is why I assume FiiO's driver not to be a 'real' ASIO driver at all.

Fig. XII. FiiO 'Andes', IMD, ASIO driver
configured to 44.1 kHz, playback at 96 kHz

Fig. XIII. FiiO 'Andes', THD + noise, ASIO driver
configured to 44.1 kHz, playback at 96 kHz

Figs. XII. & XIII.: atrocious resampling causes audio to distort severely. Distortions, no matter if THD or IMD, are just below -70 dB, making them very likely audible. This is not how an ASIO driver should perform. BTW, the imbalanced noisefloor is caused by using the Soundblaster X-Fi HD USB as a recording device; its left channel doubles as the microphone input on the front whose amp is always active, emanating noise.

Fig. XIV. FiiO 'Andes', THD + noise, FiiO driver,
WASAPI, configured to 44.1 kHz, playback at 96 kHz

Fig. XV. FiiO 'Andes', THD + noise, generic MS driver,
WASAPI, configured to 44.1 kHz, playback at 96 kHz

Just for comparison I´ve added two charts showing the FiiO driver (fig. XIV.) and the generic MS driver (fig. XV.) playing back 96 kHz WASAPI streams. In both cases WASAPI was instructed to resample everything to 44.1 kHz, which can be circumvented however by programs using the exclusive mode WASAPI offers. It enables direct access to the driver through the WASAPI software / hardware interface. There isn´t any difference between them, yet this was one of the few instances where the FiiO driver actually worked using WASAPI, it really was a rare incident. I´d like to use these two examples to make it clear to you that you can safely use the generic driver Microsoft automatically installs when you connect the Andes for the first time.

Line-in

Fig. XVI. FiiO 'Andes', RMAA chart, line-in, 4300 Ohm load

Fig. XVII. FiiO 'Andes', THD + noise, line-in, 4300 Ohm load

These measurements were made in January 2013, shortly after receiving the Andes. They serve to show that those distortions visible in the graphs for the line-out are absent, I assume that those are indeed related to the USB connection being active (I´d still love an explanation though). THD + noise construe that the highest volume setting of 60 should be avoided (fig. XVII.); while distortions stay below -90 dB I´m quite sure that RMAA is unable to measure this properly. Because I don´t know how to measure everything it´s entirely possible that some other procedure would disclose stronger distortions. I´ve opted to present the THD + noise graph only, everything else (IMD, frequency response, dynamic range, etc.) looks just like the line-out using USB.

Jitter

Fig. XVIII. FiiO 'Andes', jitter, 44.1 kHz, line-out

Fig. XIX. FiiO 'Andes', jitter, 96 kHz, line-out

German audio magazine AUDIO measured in their tests of the Avinity USB DAC Mobile a jitter performance of not-very-good 2000 ps. I really wonder if something peculiar happened there, because on my unit jitter is absent. Figs. XVIII. & XIX. demonstrate that the E07K has some amount of random low frequency jitter (the spreaded base around the 11.025 Hz &12 kHz sines), always below -110 dB. Spikes of high frequency jitter are at an even lower -125 dB. I don´t know very much about audibility of jitter, yet I´m 100% confident that these results are completely insignificant. To me this is very interesting because the FiiO transfers data using the Isochronous Transfer with Adaptive synchronization, the USB Descriptor Dumper from Thesycon confirms this. According to countless people, this mode is supposedly extremely susceptible to jitter. From all you can read online one gains the impression that it´s detrimental to general sonic performance, unsuited for perfect audio playback, yadda, yadda, yadda. Now listen to me very carefully: IT IS NOT TRUE. At least not with every interface. The Isochronous Transfer with Asynchronous Synchronization would in theory be superior, my X-Fi HD USB uses it, my E-MU 0202 USB used it. Both interfaces do or did not jitter, exactly like the FiiO. A fact that serves to exhibit that not the mode is important but how it is implemented. Numerous cases using adaptive synchronization have occasionally created horrible amounts of jitter on other interfaces, yet the only conclusion you´re allowed to draw is that those interfaces have been constructed sloppily... and not that the mode itself is faulty.

Impulse

Fig. XX. FiiO 'Andes', impulse response, 44.1 kHz

You know, I really like seeing those perfect, symmetrical impulse responses (fig. XX.), anything else wouldn´t have felt this trustworthy. I refer to impulses lacking any pre-ringing, in other peoples' ears they might sound 'analogue' but to me they sound awkward (remember, I can experiment with them using iZotope RX) Furthermore, they introduce phase distortions and horrible post-ringing so it´s good that FiiO avoided them.

Last update: 18.09.2013