Does the nano iDSD measure: Good, Bad or Ugly?

Some of you have enquired as to how the nano iDSD measures* – under conditions similar to those by recognised industry bodies/magazines.

The following are our laboratory measurements of the behaviour of the iFi nano iDSD using the Audio Precision System Two.
Audio_Precision_SYS-2322A

Setup
The iFi nano iDSD:
1) Preserved absolute polarity
2) Maximum output level at 1kHz was 1.65V (battery full-charged)
3) The output impedance was around 1 ohms

Based upon the above setup, we made the following four measurements – these follow closely (though can never be identical) the typical industry approach as used by leading audio magazines. One should allow for tolerances in the findings but there should not be any large discrepancies.

Just as an example, take the Total Harmonic Distortion (THD) measurement of the nano iDSD:

Measurements conducted THD Comment
By iFi 0.02%
By audio magazine ‘A’ 0.03% to 0.05% CheckboxYes Perfectly acceptable small variance
By audio magazine ‘B’ 0.15% CheckboxNo Measurement error/different methodology?

 

*NOTES:
i. Measurements though scientific and objective, are subject to errors and honest mistakes.
ii. There is no scientifically proven link between “better” measured performance automatically equating to better sound.
Therefore, please bear these points in mind when you read THIS and ANY other measurements article. As always, listen for yourself and let your ears be the judge. Measurements are just that, measurements.

Measurement 1: Impulse Response

The iFi nano iDSD has two user-selectable digital filters (reconstruction filters):
(i) Standard Phase; and
(ii) Minimum Phase.
With the Standard digital filter, the impulse response with 44.1kHz data (fig.1) is typical for a digital FIR reconstruction filter, with symmetrical “Ringing” either side of the impulse.
02 - iDSD nano Impulse response 44.1KHz standard Filter
Fig. 1 iFi nano iDSD impulse response with Standard Filter (44.1KHz data)

Fig. 1 – Layman explanation:

• This is the standard, de-facto filter that the majority of CD Player/DACs use. The center peak is the input signal; the 8 peaks on each side are “Ringing.”
• Ringing is a digital artifact and is undesirable. So the less Ringing the better as it is closer to the original music.
• Why use an impulse signal?
Because the impulse signal (vs. a sine wave) bears much closer resemblance to a real-world music signal. In order words, the impulse response measurement better mimics a music signal than using the legacy sine wave.

Summary: The nano iDSD exhibits Ringing when the Standard filter is engaged.

With the Minimum Phase filter (fig.2),
(i) the amount of Ringing is much reduced
03 - iDSD nano Impulse response 44.1KHz minimum Filter
Fig. 2 iFi nano iDSD impulse response with Minimum Phase Filter (44.1kHz data)

Fig. 2 – Layman explanation:
• The Minimum Phase filter significantly reduces the Ringing artifacts (from 8 peaks to one peak either side).
• Sonically-speaking, the Minimum Phase filter with less “Ringing” sounds more “life-like” than using the Standard filter.

Summary: We find the nano iDSD more “enjoyable” with the Minimum Phase filter engaged.

Measurement 2: FFT of Noise and Spuriae

FFT of noise and spuriae of dithered 1kHz tone at –90dBFS with 24-bit data (fig.3).
05 - iDSD nano noise & spuriae -90dBFS 24Bit
Fig. 3 iFi nano iDSD spectrum with noise and spuriae of dithered 1kHz tone at –90dBFS with 24-bit data

Fig. 3 – Layman explanation:
• A very quiet 1kHz signal is used and this is the small peak at 1kHz.
• We want to see once the system is running (hence we require a quiet signal), that there is no other unwanted noise produced by the system.
• From 1kHz to 10kHz, there should be no significant (>-110dB) peaks and should be flat.
• The iDSD nano resolves the -90.31dBFS signal correctly, the final FFT noisefloor is around the level of -130dB.

Summary: The nano iDSD is a quiet DAC in its class.

Fig.4 shows the spectrum of the processor’s output while it drove a full-scale 50Hz tone into a difficult 300 ohms. The highest-level harmonic is the second, at a low –92dB (right) and –95dB (left).
07 - iDSD nano 50Hz 0dBFS FFT Headphone out 300R Load
Fig. 4 iFi nano iDSD 50Hz 0dBFS FFT Headphone out with 300R Load

Fig. 4 – Layman explanation:

• A full scale 50Hz signal is used and hence the large peak at 50Hz.
• The peaks (100/150/200/300Hz etc) are the harmonics produced by system. 100Hz is the 2nd harmonic and 150Hz is the 3rd harmonic and so on.
For a good system, these harmonics should be
(i) At a low level, 2nd harmonic better than -90dB (0.003%) and all other harmonics should be better than -100dB (0.001%).
(ii) In a smooth descending scale.
• A well-designed vacuum tube preamplifier has similar harmonics to Fig.4.

Summary: The nano iDSD exhibits ear-friendly harmonics and to an extent, mimics a respectable tube amplifier.

Measurement 3: Intermodulation Distortion

The picture was similar when it came to intermodulation distortion, with a low amount of distortion products (fig.5), all better than -100dB. The wobbly floor between 16kHz and 22.5kHz is due to the software signal generator (WaveGene) that was used.
08 - iDSD nano 19+20KHz @ 0dBFS
Fig. 5 iFi nano iDSD HF intermodulation spectrum, DC–30kHz, 19+20kHz at 0dBFS into 100k ohms

Fig. 5 – Layman explanation:
• For intermodulation distortion measurement, a full scale 19+20kHz signal was applied.
• Low-order harmonics (16/17/18, 21/22/23kHz) should be in a smooth, descending order.
• High-order harmonics (4-15kHz, 24kHz and above) should be low (better than -100dB).
• It is normal that the harmonics rise again at the lower frequency (3kHz and lower). This phenomena is due to the 19+20kHz input signal which has been mirrored back (just normal physics).

Summary: The nano iDSD has good harmonics with minimal distortion phenomena in its class.

Measurement 4: Jitter Measurement

This is a narrowband spectrum of the iFi nano iDSD output while it decoded 16-bit data representing the Miller-Dunn J-Test signal (fig.6). However, no jitter-related sidebands are visible, and the odd-order harmonics of the low-frequency, LSB-level squarewave are all at the correct levels. The slight peak at 9.8kHz could be pick-up from the environment and at -130dB, it is insignificant.
09 - iDSD nano J-Test 44.1K-16Bit
Fig. 6 iFi nano iDSD 09 J-Test with 44.1kHz 16 bit data

Fig. 6 – Layman explanation:

• The input signal is represented by the main peak at 11kHz with periodic side peaks.
• For a low-jitter system,
(i) The skirt at the bottom of the 11kHz signal should be small; and
(ii) Other than the period side peaks, no other significant peak is visible.

Summary: The jitter levels of the nano iDSD are quite impressive in its class.

Conclusion: Going by measurements alone, the nano iDSD does not measure as the best USB DAC in the world. However, in the affordable USB DAC amplifier class, its measurements are comfortably above average and it exhibits no disturbing measurement characteristics.

We hope you found this small exercise reassuring and informative.

If you have any questions, please do not hesitate to open a support ticket and ask one of our technical staff.
http://support.ifi-audio.com/