iPod Measurements

Introduction

To get a rough idea about the sound quality of a device, one can perform some measurements. Measurements only reflect specific characteristics of a device, and don't define the sound quality of a device. A flat response from 20Hz to 20kHz says something about the static response (sweep measured) or average response (white noise measured) of a device. It doesn't for instance tell you anything about the characteristics of a power supply when it instantly needs to deliver a certain amount of charge because of a sound transient. Measurements also don't tell you if a device sounds "nice" or "nasty", e.g. a distorting tube amplifier can sound much nicer than a flat-response cheapo transistor amplifier.

Measurements can be used to find flaws (if severe deviations from expected measurements are found), or to find some characteristics of a device (e.g. noise shaping).

Frequency Response

The German hifi magazine Stereoplay published in their October 2003 issue the following measurements of the iPod 3G.

The line output looks good. As you can see for the headphones output, the iPod has a significant attenuation of bass when using an output impedance of 25 Ohm. We've also conducted this experiments ourselves (thanks to Guido Tent for using his time and measurement equipment), applying a 33Ohm load. We've measured the same characteristics, as shown by the following numbers:

Freq [Hz] 1 3 10 30 100 300 1000 10000
Output Voltage @ 1MOhm 65 88 90 90 90 90 90 90
Output Voltage @ 33Ohm 2 7 22 51 75 80 80 80

As most mobile headphones are in the range of say 16 to 32 Ohm (including the one delivered with the iPod!!!), this results in a poor bass performance overall, which is quite obvious if you compare the iPod towards other devices.

If you study the specs of the Wolfson Microelectronics WM8731, the output circuit of the iPod, you can see that Apple (or should I say Portalplayer) followed the suggestions of Wolfson Microelectronics to apply a 220uF capacitor at the headphone output stage. I would rather have liked the attenuation to be at least one or 2 decades lower, because an obvious phase shift is induced, which might explain the somewhat "sloppy" bass performance of the iPod.

Applying a higher impedance headphone will give you more bass, but less efficiency, which is a non-issue for EU-based iPods anyway. Apple could easily account for this problem by an equalizer setting that compensates for this phenomenon. Please give them a message on their iPod feedback pages, so that they'll resolve this issue in the next Software Update.

Distortion and jitter

C'T Magazine reports a distortion figure of 0,42% (which is quite some), and 90dBA dynamics. The maximum output voltage is 0,29V (EU version, compared to e.g. 0,86V for the Nomad Jukebox Zen NX...). All measurement concern the headphones output.

Stereophile shows some interesting distortion measurements in their October 2003 issue. These measurements have been performed on the line output, and not on the headphones output.

The first figure (from Stereophile) shows harmonic distortion:

The second figure shows intermodulation distortion:

The last figure shows the jitter performance:

These figures show that the static performance of the digital circuitry is quite good (better than many CD players as reported by Stereophile), and the line output is well defined as well. This shows that it's a pity that the headphones output is not designed with the same care, because intrinsically the iPod shows good potential performance, which might only require one firmware upgrade to reach its headphones output.

Digital Silence measurements

The following two figures show the output signal with a file containing silence. The first one is the headphone output (0-50kHz, 1mV full scale, hence Vout is about 400 uV - 500uV):

The noise increases after 35kHz, indicating some noise shaping techniques (see the digital backgrounds page of our DAC pages for details on this, if you're interested).

The second one is the line output (0=50kHz, 2mV full scale, hence Vout is about 800uV - 1mV):

Notice the difference in spectra, showing that the headphone output seems to perform better on this??? We've also measured the iPod in pause mode (so not playing a file, no backlight), input sensitivity 1mV fullscale.

As you can see, a spike pops up at 26kHz??? The noise however is more towards 400uV, and not raising like in the case digital silence is played (which might be the noise shaping characteristics).

Output impedance

The output impedance of the headphone jack is about 5 Ohm (measured using white noise), and the output impedance of the line output is about 100 Ohm.

Square wave response

No pictures of this one (the scope had some difficulties triggering, and the sample on the iPod was only 5 seconds). The square wave shows ringing, which is logical for a band limited signal. It looks fully symmetrical.

Conclusions

These measurements of the iPod, far for complete, at least show that the iPod has a decent ouput signal, free from big artifacts, although the bass response on the headphones outut is flawed.

© Copyright 2003, Marc Heijligers