Clip v1 vs. Clip+ - RMAA and Spectrum Analysis

[dfkt]

EDIT: Here are the latest frequency response graphs with Rockbox, all is fine: http://www.anythingbutipod.com/forum...&postcount=745

Here's the RMAA test, both Clip v1 and Clip+, loaded and unloaded: http://rmaa.elektrokrishna.com/Compa...20Loads%29.htm

Same as before - the Clip and Clip+ are very nicely performing players. The new Clip+ shows a tiny bit better results under load - at least on paper, probably not to anyone's ears.


Frequency response, perfect as ever:




Crosstalk, the weakest point of the Clip - and most other players under load:

+++++++++++++++++++++++++++++++++++++++++++

And here's the Sound Forge 8.0 spectrum analysis results about the often mentioned "Clip and Clip+ playing too fast" issue. I've used a Blackman-Harris windowing function with a FFT size of 64k (best frequency resolution, least time resolution). All tests have been made with WAV, FLAC, Vorbis, and LAME. All codecs produced the same results. The soundcard used for recording the 1kHz sine wave from the players was an Echo AudioFire4.

Dark blue graph is the Cowon S9, playing perfectly at 1000Hz. Green is the Clip+ (FW 1.01.05P), playing at 1003Hz. Turquoise is the Clip v1 (FW 1.01.32A), playing at 1007Hz:




Here Rockbox on the v1 Clip is turquoise (999Hz), stock firmware x.32 is green (1007Hz):




Here's the same for 500Hz playback. Orange is Cowon S9 (exactly 500Hz), Green is the Clip+ (501Hz), Turquoise is the Clip v1 (503Hz):



Everyone should draw their own conclusions, but for me the whole "issue" appears to be very much negligible.

[d_headshot]

So for the people complaining...

[Mikerman]

Thanks for the testing--fascinating and interesting stuff.

In the end, I still can understand the issue for people whose ears are sensitive, and better than mine; and especially for musicians. And even for the rest of us, one can't help but wonder what we might end up sensing as a whole, without noting specifics.

And then, there's the question: why couldn't they just have "gotten it right"/closer to begin with, and have issued a correction/fix? It's been interesting reading the comments of engineers who have said that at their own workplaces, they would not have been allowed to issue out firmware with that amount of error; and that a fix would be relatively easy (certainly, Rockbox found a way).

[dfkt]

To put that in perspective, here's some files to download:

1000 Hz sine wave
1007 Hz sine wave

Music recorded from Cowon S9 output
Music recorded from Clip+ output
Music recorded from Clip v1 output

Download the three music tracks and play them looped... can you hear the difference? I can't.

While the difference between a 1000 and 1007 Hz sine wave is blatantly obvious, it should be almost impossible for "normal" ears to detect the difference between the two music tracks. The difference is less than 1/10th of a half tone.

Here's some significant differences between the three music tracks zoomed in (S9 is orange, Clip+ is blue and a bit quieter, Clip v1 is green):



Seems my ears aren't as good as Sound Forge's FFT analysis, despite being a musician and audio engineer.

[saratoga]

Good to see someone else could duplicate my (private) testing of Rockbox's pitch. I was beginning to wonder if I did something wrong, since pitch in Rockbox on the clip looked nearly perfect to me (using matlab for frequency analysis).

[dfkt]

I wonder what the problem is... a bug in the AMS chips? I mean, of course it doesn't matter in reality that Rockbox plays a 1000Hz sine wave at 999Hz, and the stock firmware at 1007Hz - but it is curious indeed.

[saratoga]

Its just that the easiest clock speeds for the chip aren't readily divided by 44100 so theres some pitch error. The datasheet recommends some clock speeds to use that give exactly 44100, but they're pretty high and Sandisk doesn't use them. I guess we copied some of the Sandisk settings and so are a slightly off (but not nearly so bad as them).

I'm not sure if theres some technical reason for this, messing with the AMS clock speeds seems to be a nightmare, so I've left it well enough alone. Probably why Sandisk hasn't fixed it on their older players.

[defaultluser]

Originally Posted by saratoga

Its just that the easiest clock speeds for the chip aren't readily divided by 44100 so theres some pitch error. The datasheet recommends some clock speeds to use that give exactly 44100, but they're pretty high and Sandisk doesn't use them. I guess we copied some of the Sandisk settings and so are a slightly off (but not nearly so bad as them).

I'm not sure if theres some technical reason for this, messing with the AMS clock speeds seems to be a nightmare, so I've left it well enough alone. Probably why Sandisk hasn't fixed it on their older players.

Yeah, that would make sense, a higher source clock is going to burn more power, in the source as well as the clock divider hardware.

The Clip / Clip+ is designed to be an ultra-low power platform - the OLED display is color-on-black to absolutely minimize power consumption. The audio power output is much lower than other Sansa players (including my old e280). Since there is no need for a beefy processor with video playback capability, it makes sense that Sandisk would push the reference clock as low as possible.

In the end, power is everything, because they just couldn't increase the size of that tiny little patch of a battery (290 mAh!). They managed to eck-out 15 hours playback, which is really impressive.

It looks like the results above confirm the earlier tests, which show the Clip+ to be off in pitch by %0.25. That's good enough for me, especially when you consider the unbelievably flat frequency response!

[Elias Hiebert]

Originally Posted by dfkt

To put that in perspective, here's some files to download:

1000 Hz sine wave
1007 Hz sine wave

Hey, that hurts!

[mattbatt]

Of course it hurts 1K is the center of the human hearing range. We are most sensative to it and it is the most common in speaking.
As an audio Engineer I got all worried when I heard about this problem. Now I am relieved.
Fun trick play both sounds at the same time and you get some interesting phasing. I made some mp3s to demonstrate but I can't upload them.

[titansausage]

Don't you think, that the Clip+ sounds very warm? Too warm?
I bought one, because the Review on ABI told, that the Clip+ sounds very good. But i think he does not Sound neutral. The high frequencies are bad and muffling. Also the Soundstage and Resolution are bad.
My Sony Player sounds much better, clearer, has more details. I'm disappointed.
BTW i tested the Clip+ with the Neutral SF3.

Can you tell me why the RMMA Test shows such good results?
Because my hearing says something different and my hearing is very very good.
Thanks for answering.

Regards from Germany and sorry for my bad English

I forgot to say, that i bought a Fuze for my brother and he sounds clearer than the Clip+

[saratoga]

Originally Posted by titansausage

Don't you think, that the Clip+ sounds very warm? Too warm?
I bought one, because the Review on ABI told, that the Clip+ sounds very good. But i think he does not Sound neutral. The high frequencies are bad and muffling.

You should review this thread. The sound is shown to be essentially perfectly neutral, high frequencies included.

If it sounds otherwise, you've probably just got a pair of headphones that strongly attenuates high frequencies, have high frequency hearing loss, or some other condition that sensitizes you to bass. In this case you may prefer MP3 players with powerful EQ so that you can compensate for this effect in your equipment or hearing.

Originally Posted by titansausage

Can you tell me why the RMMA Test shows such good results?
Because my hearing says something different and my hearing is very very good.
Thanks for answering.

Regards from Germany and sorry for my bad English

I forgot to say, that i bought a Fuze for my brother and he sounds clearer than the Clip+

Perhaps your brother has more neutral equipment or hearing then you.

[titansausage]

I got lots of Headphones. SR60, NuForce NE-7M, Porta Pros, Denon AH P372 and the NEUTRAL SuperFi 3. Understand?
The SuperFi 3 are Neutral. That means: If the Clip+ would play Neutral they would play Neutral too, because they play exactly what the DAP tolds them.
And i testet the Fuze with my own hearing and with my own Headphones and he sounds clearer.

[Enigmatic]

Originally Posted by titansausage

I got lots of Headphones. SR60, NuForce NE-7M, Porta Pros, Denon AH P372 and the NEUTRAL SuperFi 3. Understand?
The SuperFi 3 are Neutral. That means: If the Clip+ would play Neutral they would play Neutral too, because they play exactly what the DAP tolds them.
And i testet the Fuze with my own hearing and with my own Headphones and he sounds clearer.

You are probably experiencing the placebo effect. A neutral-sounding MP3 player with neutral-sounding headphones is probably going to sound—wait for it—neutral.

[titansausage]

I hope you are right

[mattbatt]

I am assuming that you are doing all this "testing" using the same flac file on all these players. Mp3 screw up the high end even at 320 kbs. There is this song I like to use to test subs and get an over all feel for the sound system it's "#1 Crush" by Garbage off of the Romeo + Juilet soundtrack. There is this great bass line in the intro that really tests the clarity of subs. It also has a very present hi-hat at the beginning it's a constant tapping. All lossless encoders screw this up and make it sound swishy. I normally can't tell Mp3 from CD but there a handfull of songs that I have heard as cd's enough that I can tell.
BTW i don't feel that the Clip+ is too "warm" but that is just my opinion.

[Orum]

I don't know about you, but I can ABX the Clip+ vs S9 100% of the time (after normalization, and PortaPro headphones). I suppose they aren't lossless recordings, so if we had some FLAC recordings it might produce more scientific result. But I still feel like the difference is obvious enough to my ear that the Clip+ wouldn't be a worthwhile purchase.

This is disappointing, as the Clip+ has almost all the features I want aside from the playback quality! I'm glad I did some digging into the issue before buying it, unlike many other customers who weren't so lucky.

[saratoga]

Originally Posted by Orum

I don't know about you, but I can ABX the Clip+ vs S9 100% of the time (after normalization, and PortaPro headphones). I suppose they aren't lossless recordings, so if we had some FLAC recordings it might produce more scientific result.

Thats probably due to the S9. Its frequency response isn't so flat as the clip, so its probably not very hard to ABX if you pick bass heavy music:

http://rmaa.elektrokrishna.com/Compa...l%20Volume.htm

[Orum]

Ah, darn, I wish I had the source audio clip to compare the recorded output to.

[mrsmr2]

Forgive the thread resurrection but I've just received my Clip+ and wanted to comment.

I can hear a difference between my V1 4GB Clip and Clip+ 8GB. Now, my terms are probably not correct but it seems that the + has lost some of the airy feel of the original. The layers of music that the original had seem a little more congested and it feels slightly less bright.

Overall, I like the +. The folder browsing is a godsend, it has a better UI, and less hiss than my V1 Clip and D2.

Just a tad disappointed as I wanted a Clip with more storage to replace my D2.