Battery Test vs Bitrate Results

Being curious how long my Clip’s battery lasts while playing certain bitrate songs, I started conducting tests.   Testing details are at the end of this post.

The results:    I will update as I complete more tests.  Test Results in Hours Minutes Seconds  HH:MM:SS

 

Battery Recharging Test Summery: - Charge times from a completely dead battery.

 

Charging Silver Clip 4GR1:


   Charging from Desktop PC USB:  Test1:  03:39:00    Test2:  03:45:00
   Charging from AC adapter:      Test1:  02:20:00    Test2:

 

Charging Black Clip 8GR2:


   Charging from Desktop PC USB:  Test1:  02:40:00   Test2:  02:30:00  * charged faster than Silver same USB port*
   Charging from AC adapter:      Test1:  02:16:00   Test2:  02:24:00

 

Do headphones make a significant impact on battery life?

 

  Black 8GR2 playing Flac Level8 encoded album at 1/2 Volume

 

  Standard Sansa Earphones       xx:xx:xx
  Skull Candy FMJ 9mm in-ear     07:46:00
  Sennheiser HD-595              07:40:00
  No headphone - Zero 0 Vol      07:45:00

 

Playing Time - Comparing the Silver 4GR1 vs… Black 8GR2:  1/4 Volume

 

  Silver Flac Lvl8            : 09:02:00  * Silver wins lasting an hour longer than the Black 8GR2 *
  Black  Flac Lvl8            : 07:46:00    I don’t understand why???  I’m going to retest again

 

  Silver Flac Lvl0            : 08:59:00  * Silver wins lasting an hour longer than the Black 8GR2 *
  Black  Flac Lvl0            :~07:50:00    Died between two checks - this is an approximate guess +/- 10 min

 

  Silver Ogg q8.5 ~256k       :           * Going to retest - Silver died while I was away *
  Black  Ogg q8.5 ~256k       : 08:05:00  * Black was still playing so it wins but not sure by what *

 

  Silver Ogg q5 ~151k         : 10:15:00
  Black  Ogg q5 ~151k         : 12:12:00  * Black 8GR2 lasted 2 hours longer than the Silver 4GR1*

 

  Silver MP3 v0 ~252k VBR     : 15:07:00  * Appears ARM clocking to be same as that of v2       
  Black  MP3 v0 ~252k VBR     : 17:18:00  
Black 8GR2 lasted 2 hours longer than the Silver 4GR1*

 

  Silver MP3 v2 ~190k VBR     : 15:09:00  * Appears ARM clocking to be same as that of v0       
  Black  MP3 v2 ~190k VBR     : 17:22:00  
Black 8GR2 lasted 2 hours longer than the Silver 4GR1*

 

  Silver MP3 v6 ~128k VBR     : 17:40:00
  Black  MP3 v6 ~128k VBR     : 20:47:00  * Amazing.  lasted 3 hours longer than the Silver 4GR1*
 

 

Playing Time - Comparing Various Encoding Options:

 

Silver 4GR1: Standard Sansa Earphones

 

   Flac Level8        :  1/2 Vol = 09:02:00    1/4 Vol = 09:15:00

   Ogg q5 ~151k VBR   :  1/2 Vol = 10:03:00    1/4 Vol = 10:15:00
   MP3 128k CBR       :  1/2 Vol = 17:20:00

   MP3 v6 ~128k VBR   :  1/4 Vol = 17:40:00

 

Black  8GR2:  Skull Candy FMJ’s

 

   Flac Level8        :  1/2 Vol = 07:46:00

 

Black  8GR2:  Standard Sansa Earphones

 

   Ogg q5 ~151k VBR   :  1/4 Vol = 12:12:00

   MP3 v6 ~128k VBR   :  1/4 Vol = 20:47:00

   MP3 v2 ~190k VBR   :  1/4 Vol = 17:18:00

 

Playing time vs… Volume level?

 

  Black 8GR2 - Flac Level 8 encode

 

  1/2 Vol - Sennheiser HD-595  07:40:00
    0 Vol - No headphones      07:45:00
  1/4 Vol - Standard phones

 

 

***********************************************************************

What effects the battery life of the Sansa Clip?

 

  Processor:   Arm CPU clock speed


  Amplifier:   Volume level?


  Display:     OLED display with adjustable Brightness - not a factor in these tests -
                 display not on frequent enough to be measurable.


  Music:       Codec and Encoding Bitrate affects Arm CPU clock speed,
                 Type of? Easy listening vs… HardCore? Power required for amp to reproduce the waveform?


  Headphones: Impedance of headphones?

 

Hardware: Both clips were formatted, reset to factory settings and loaded with the same test files.  EQ on Normal

 

  Silver 4 gig Revision 1 Clip  Firmware version: 01.01.30a
  Black  8 gig Revision 2 Clip  Firmware version: 02.01.16a

 

Headphones used:
  Standard Sansa Earphones
  Skull Candy FMJ 9mm in-ear
  Sennheiser HD-595 - Full sized headphones

 

Music used for test:
  Album used:  Kings of Leon - Only By The Night - Length 00:42:39
  Ripped from CD using dBpoweramp - AccurateRip 1411 Kbps Flac 44.1K/s 16 bit Stereo
  Ripped as a single Flac file, 42 minutes 39 seconds long   00:42:39

 

Codecs used for encoding:
  MP3 lame 3.98
  Ogg aoTuV b5
  Flac 1.2.1

 

Test Details:

The album was played endlessly on repeat until the battery ran out of juice and the player died (battery low).  Test Results in Hours Minutes Seconds  HH:MM:SS. Seconds were ignored, not significant.  In any individual data sampling test there is uncertainty.  Don’t be surprised at the variation.  You can’t calculate to the minute exactly how long your battery will last.

 

Observation on battery power percentage measurement:
Power percentages through windows property and through the Clips menu vary.  I’ve seen the clip at 0% but windows reports it at 11%.  Use the Power% like a gas gauge.   It isn’t perfectly linear and sometimes it seem to stick.  Turning the Clip off and on can sometimes display a different Power%. 

 

The Power% measures in increments.  0,7,14,21,28,35,42,50,57,65,71,78,85,92,100

Volume Control Settings used for this test:
  Clips volume control appears to have approximately 40 clicks from zero volume to Max volume
  0 clicks = zero Vol,  10 clicks = 1/4 Vol,  20 clicks = 1/2 Vol,  30 clicks = 3/4 Vol

Edit :  cleaned up post - added results

Message Edited by niko_sama on 02-01-2009 02:41 PM

Edit: added results

Message Edited by niko_sama on 02-02-2009 02:10 PM

Message Edited by niko_sama on 02-03-2009 03:42 PM

Message Edited by niko_sama on 02-04-2009 08:16 AM

Message Edited by niko_sama on 02-05-2009 12:34 PM

Add mp3 v2 compare

Message Edited by niko_sama on 02-07-2009 07:46 AM

20 hours of play time out of my black 8gig Rev 2,  vs   17 hours out of my Silver 4gig Rev 1.     Makes me wonder if its just the variation between two clips.  Or if its changes in design and components between the two Revisions.  Hmmm?     SansaFix can you  say?

There are variations in battery capacity and hardware power consumption that could affect the results.

great work, thank you! :slight_smile:

Thank you

update*  added mp3 v0 vbr comparison

I’m curious why ogg appears to use so much more battery power. 

I’m going to test another ogg bitrate and see how it comes out.

I also don’t understand why   Flac8 on the Rev 1 lasted longer than the Rev 2 clip.    All the other tests show the Rev 2 beating out the Rev 1 by a few hours.   Strange…   I may have to test Flac Lvl0 and Ogg Q0 see how the higher bit rates perform.

Next Testing Ogg Q ?    Un decided.   probably one around 252k/s avg.   to match the mp3 test.

Added Flac 0 comparison


Comparing the Revision 1 to Revision 2 units , I’m wondering if Rev2 is clocked higher for Flac than  Rev 1 firmware.    Can someone else who has a Rev 2 and a Rev 1 clip,  reproduce a similar test and compare them.    Compare the two clips  at a high MP3 rate and then a flac file.   the mp3 can be a baseline and the flac can be the study and see if it looks as if Rev 2 Flac is clocking the Arm processor higher and getting shorter battery time than Rev 1.  

If this information could be passed to the Sansa software design team I’d appreciate it.   

Its possible that this is something that could extend Rev 2’s battery power playing Flac files.

Rev1 has 14% longer battery life playing Flac.       Where usually Rev2 has about a 12% longer battery life on mp3 and ogg.   That’s a change of 26%…   

I contribute the 12% to be a variation between my rev2 to my Rev 1…  

But the flac testing is off by 26% from the variation I expected.

Ogg vorbis uses more battery power to accomplish its decoding (a known fact–nothing is wrong here).

@miikerman wrote:
Ogg vorbis uses more battery power to accomplish its decoding (a known fact–nothing is wrong here).

Who said anything about anything being wrong with Ogg other than it takes up 1/3 more power?     Are you responding to my post about Flac?  If so there is perceivably something different between rev 2 and rev 1 .  

If you are replying to my Question in the Fuze forum,   SansaFix already mentioned  Ogg uses more power due to a higher clock.

Just curious what you were refering too :slight_smile:

@miikerman wrote:
Ogg vorbis uses more battery power to accomplish its decoding (a known fact–nothing is wrong here).

Who said anything about anything being wrong with Ogg other than it takes up 1/3 more power?     Are you responding to my post about Flac?  If so there is perceivably something different between rev 2 and rev 1 .  

If you are replying to my Question in the Fuze forum,   SansaFix already mentioned  Ogg uses more power due to a higher clock.

Just curious what you were referring too :slight_smile:

I was referring to your comment a few above: 

Thank you

 

update*  added mp3 v0 vbr comparison

 

I’m curious why ogg appears to use so much more battery power. 

And I simply added the statement as to all being fine with Ogg vorbis so that no one would think that there was any issue with it–it simply is the nature of Ogg vorbis to use more power to decode.  Ciao.

Cool thanks - ^_^  I was lost for a moment  :slight_smile:

<p><strong><font size=“3”>Some conclusions</font></strong> so far,  these are my own opinions, always prone to misinterpretations so feel free to draw your own conclusions and share.</p><p> </p><p>The variation was consistent between my Rev 1 clip and my Rev 2 clip.  except in the FLAC test, where I feel the Rev 1 clip is being clocked slower than the Rev 2.   I’m not sure if this is true,  <strong>I’d like Sansa to check see if this is true</strong> and see if Rev 2 firmware can be changed to clock at the same rate as Rev 1.   I’d also like to know if Sansa could make a chart that shows what Arm Clock speeds are used for various formats / bitrates.</p><p> </p><p><strong><font size=“3”>FLAC:</font></strong>  There is ~ 50% power consumption penalty using FLAC over MP3 VBR v0/v2.  Your battery duration will last one half as long 8 hours vs. 15 hours.  And file sizes much larger than MP3 VBR v0 -  4 to 6 times larger.</p><p> </p><p><strong><font size=“3”>Ogg:</font></strong>  There is ~ 33% power consumption penalty for using Vorbis Ogg q7 vs. MP3 VBR v0/v2.  Your battery duration will last 1/3 less. 10 hours vs. 15 hours.  Ogg file sizes will be 30% to 40% smaller than for compatible MP3 quality files.</p><p> </p><p><strong><font size=“3”>MP3:</font></strong>  You will get a decent quality for average file size.   Lame MP3 v0/v2 offer excellent quality, while still getting 15 hours of music from the Sansa Clip.  File sizes are 30% to 40% larger than Ogg for similar quality.</p><p>   </p><p><strong><font size=“3”>The Sansa Clip estimate of 15 hours of useable battery is indeed possible</font></strong> with MP3 VBR v0/v2,  and actually you can get more the lower your bitrate.  Example: MP3 VBR v6 lasted up to 20 hrs of play time.</p><p> </p><p><font size=“3”><strong>Battery power consumption is mostly consumed by the Arm CPU’s clock speed.</strong></font>    A small amount maybe used by volume or type of headphones, but not nearly as much as the effect that the music’s format and bitrate have to play.</p><p> </p><p><strong><font size=“4”>Personal Comment:  MP3 VBR will give you greater quality than CBR.   Don’t encode using CBR, its not efficient.</font></strong></p><p> </p><p> </p><p><strong><font size=“3”>Next tests:</font></strong>   I will perform a test using <strong><font size=“3”>various types of music</font></strong>, to see if music genre effects battery consumption.  Category’s to test will include top 100 billboard (mix of pop, rock, country, r&b), heavy metal (no ballads), and Classical.   These will be encoded to MP3 v2 from CD or FLAC sources.</p><p> </p><p> </p>

Some conclusions so far,  these are my own opinions, always prone to misinterpretations so feel free to draw your own conclusions and share.

The variation was consistent between my Rev 1 clip and my Rev 2 clip.  except in the FLAC test, where I feel the Rev 1 clip is being clocked slower than the Rev 2.   I’m not sure if this is true,  I’d like Sansa to check see if this is true and see if Rev 2 firmware can be changed to clock at the same rate as Rev 1.   I’d also like to know if Sansa could make a chart that shows what Arm Clock speeds are used for various formats / bitrates.

FLAC:   There is ~ 50% power consumption penalty using FLAC over MP3 VBR v0/v2.  Your battery duration will last one half as long 8 hours vs. 15 hours.  And file sizes much larger than MP3 VBR v0 -  4 to 6 times larger.

Ogg:   There is ~ 33% power consumption penalty for using Vorbis Ogg q7 vs. MP3 VBR v0/v2.  Your battery duration will last 1/3 less. 10 hours vs. 15 hours.  Ogg file sizes will be 30% to 40% smaller than for compatible MP3 quality files.

MP3:   You will get a decent quality for average file size.   Lame MP3 v0/v2 offer excellent quality, while still getting 15 hours of music from the Sansa Clip.  File sizes are 30% to 40% larger than Ogg for similar quality.

  

The Sansa Clip estimate of 15 hours of useable battery is indeed possible with MP3 VBR v0/v2,  and actually you can get more the lower your bitrate.  Example: MP3 VBR v6 lasted up to 20 hrs of play time.

Battery power consumption is mostly consumed by the Arm CPU’s clock speed.    A small amount maybe used by volume or type of headphones, but not nearly as much as the effect that the music’s format and bitrate have to play.

Personal Comment:  MP3 VBR will give you greater quality than CBR.   Don’t encode using CBR, its not efficient.

Next tests:    I will perform a test using various types of music , to see if music genre effects battery consumption.  Category’s to test will include top 100 billboard (mix of pop, rock, country, r&b), heavy metal (no ballads), and Classical.   These will be encoded to MP3 v2 from CD or FLAC sources.

I was curious about the performance of various codecs on ARM CPUs, so I did an even bigger comparison in Rockbox on the Clip v2 :

128k (give or take for AC3) lossy files:

flac_8.flac                     7.67MHz
a52_stereo_192.ac3     17.29MHz
mpc_128.mpc               18.01MHz
wma_128.wma             19.26MHz
vorbis_128.ogg            20.16MHz
nero_128.m4a              24.26MHz
lame_128.mp3             26.35MHz

 

 

(need to update that with WMA Pro) 

Lots more testing and info: 

http://www.hydrogenaudio.org/forums/index.php?showtopic=82125 

FWIW it seems like most of the rockbox decoders are quite a bit more efficient then the Sandisk ones, probably because I worked on them :wink:   It seems like the Sandisk one has some issues with Vorbis and FLAC, so if you use those formats much, you’ll probably be a lot better off with Rockbox.   For Mp3 its a toss up, the Sandisk decoder seems good even compared to Rockbox , maybe even better though I’m not 100% sure which is faster.  I guess MP3 is probably their number 1 priority since its what most people use.