My Ipod Classic 160gb measurements - and thoughts

[Hales]

Setup:

Sinewave sweeps from 20hz to 20khz, White Noise, 1khz Impulse Response in wav format.

Sampled at 96khz in Adobe Audition CS3.

Recorded to internal sound card on a Dell Precision M90(@96/24). Recorded to external Creative Labs USB sound card(@96/24) for verification.

Files recorded from Ipod Classic 160gb headphone jack and the line out using the ipod dock that came with my 4g. Files also burned to CD and recorded from a Marantz 8260 cd/sacd player.

Files were level matched using a 1khz and 3khz test tones that preceded the sweeps.

Processing:

All wave forms were displayed at at 1920x1200 and arranged so that screenshots could be taken. The screenshots were then imported into Photoshop.

Each of the screenshots were overlaid with varying opacity so that any changes between the waveforms could be seen.

Findings:

The headphone out and line out on the ipod classic have almost exactly the same frequency response up to 20khz.

The headphone out appears to have a brick-wall type filter to stop any output over 20khz.

The line out appears to have no filtering on content above 20khz.

Compared to the reference Marantz CD player, the ipod classic has identical frequency response from 20hz to 3.5khz.

From 3.5khz to 19khz the ipod increasingly deviates with a shelved up treble response. The deviation steadily rises from .1 db at 4khz to .5db from the headphone out at 19khz.

If you set the eq in the ipod classic to "flat" it yeilds exactly the same response curve as "eq off".

Impulse responses from the Marantz were completely symmetrical. Impulse responses from the ipod classic were similar to those displayed here with a ringing after the impulse.

My interpretation of the results:

There is a shelved up frequency response on my ipod classic 160gb with firmware 1.0. I can hear it, and I could hear it before I was told it was there. If it were ONLY a .1db bump at 19khz, I agree that it would be inaudible. But its not, its a steadily increasing treble response from around 3.5khz up. This kind of broad shelving will be audible on many headphones. You will either love the sound or hate it.

Since the response is the same from the headphone and line out, I am hopeful that the issue can be corrected through a firmware update. This update would either remove the shelving if it was put there intentionally, or provide us with a revised EQ curve that would allow for the correction of the anomalies.

 

[Quaddy]

Quote:

Originally Posted by Hales /img/forum/go_quote.gif Setup: Sinewave sweeps from 20hz to 20khz, White Noise, 1khz Impulse Response in wav format. Sampled at 96khz in Adobe Audition CS3. Recorded to internal sound card on a Dell Precision M90(@96/24). Recorded to external Creative Labs USB sound card(@96/24) for verification. Files recorded from Ipod Classic 160gb headphone jack and the line out using the ipod dock that came with my 4g. Files also burned to CD and recorded from a Marantz 8260 cd/sacd player. Files were level matched using a 1khz and 3khz test tones that preceded the sweeps. Processing: All wave forms were displayed at at 1920x1200 and arranged so that screenshots could be taken. The screenshots were then imported into Photoshop. Each of the screenshots were overlaid with varying opacity so that any changes between the waveforms could be seen. Findings: The headphone out and line out on the ipod classic have almost exactly the same frequency response up to 20khz. The headphone out appears to have a brick-wall type filter to stop any output over 20khz. The line out appears to have no filtering on content above 20khz. Compared to the reference Marantz CD player, the ipod classic has identical frequency response from 20hz to 3.5khz. From 3.5khz to 19khz the ipod increasingly deviates with a shelved up treble response. The deviation steadily rises from .1 db at 4khz to .5db from the headphone out at 19khz. If you set the eq in the ipod classic to "flat" it yeilds exactly the same response curve as "eq off". Impulse responses from the Marantz were completely symmetrical. Impulse responses from the ipod classic were similar to those displayed here with a ringing after the impulse. My interpretation of the results: There is a shelved up frequency response on my ipod classic 160gb with firmware 1.0. I can hear it, and I could hear it before I was told it was there. If it were ONLY a .1db bump at 19khz, I agree that it would be inaudible. But its not, its a steadily increasing treble response from around 3.5khz up. This kind of broad shelving will be audible on many headphones. You will either love the sound or hate it. Since the response is the same from the headphone and line out, I am hopeful that the issue can be corrected through a firmware update. This update would either remove the shelving if it was put there intentionally, or provide us with a revised EQ curve that would allow for the correction of the anomalies.

nice work sherlock

does the importing of the charts at a set graphical resolution skew the indices at all, or was that merely to have a like for like?

 

[Hales]

They were all level matched with 1khz and 3khz tones to make sure that the wave forms were the same going in.

To make sure the visual representation of those wave forms were the same coming out I had to set the view only show the 5 seconds of tone sweep, then set the view to full screen. This allowed the wave forms to be captured at the same size.

I then overlaid the files in photoshop and varied the color and opacity so that I could see the difference in the wave file vs the reference from the Marantz CD player.

 

[bigshot]

Quote:

Originally Posted by Hales /img/forum/go_quote.gif From 3.5khz to 19khz the ipod increasingly deviates with a shelved up treble response. The deviation steadily rises from .1 db at 4khz to .5db from the headphone out at 19khz.

That is as flat as dammit.

A frequency response shift that tiny is just as likely to be a deviation unique to your particular player as it is the design of the player itself. Either way, it's inaudible. Get up into the 1 to 3 dB range in the core frequencies (40Hz to 10kHz) and it might start to matter.

See ya
Steve

 

[Hales]

I disagree - flat would be the color 4g - which I just tested.

The response of the 4g is exactly the same as the reference Marantz from 45hz on up. I mean exactly the same.

 

[zip22]

are you testing loaded or unloaded? would that matter?

 

[Hales]

Ok - I need to figure out something. The .1 to .5 shelving was not db - it was dbfs. So it was .1 to .5db derivation from full scale digital.

I don't know how to convert this to regular db as it would apply to speakers. In speakers and headphones, +3db measured difference is twice the volume level. But the stuff I am talking about is all wave input levels at 96/24.

 

[Hales]

I am testing the headphone out unloaded.

If there were a difference in the response curve of the headphone out vs the line out I would retest with a 32ohm load and a 600ohm load.

Since the response curves of the headphone out and line out were identical, I did not test loaded.

 

[Hales]

I just emailed a jpg of the results to Vinnie and Redwine. I am wanting to get his opinion on the results before posting the visual findings here.

 

[mrarroyo]

0.1 to 0.5 dB will not be heard! Specially if you are walking on a busy city street or in a bus/train/plane. Lets get real!

 

[Vinnie R.]

Hi Hales,

Thanks for posting your findings!

I think it will be worthwhile to try your experiment under load. I would use 32-ohms for the headphone out, and maybe 5k or 10k for the line-out (I don't think many headphone amps have an input lower than 5 or 10k).

Quote:

Originally Posted by Hales From 3.5khz to 19khz the ipod increasingly deviates with a shelved up treble response. The deviation steadily rises from .1 db at 4khz to .5db from the headphone out at 19khz.

This isn't much of a rise and I can see why some would argue that this is not something one can hear.

Quote:

Originally Posted by Hales Impulse responses from the Marantz were completely symmetrical. Impulse responses from the ipod classic were similar to those displayed here with a ringing after the impulse.

Now this is where things are more interesting and I think this is what is audible.

Quote:

I just emailed a jpg of the results to Vinnie and Redwine. I am wanting to get his opinion on the results before posting the visual findings here.

Thanks for the plot. It is a little tricky to read with everything on the same page, and it would be helpful to see Frequency on the X-axis. If you can change the scaling on the Y-axis, that would make it easier to view as well.

Thanks again for your post,

Vinnie