Rockbox Technical Forums

Support and General Use => Hardware => Topic started by: Buschel on May 22, 2007, 02:53:17 AM

Title: Experiments on power consumption (iPOD 5G, 30G)
Post by: Buschel on May 22, 2007, 02:53:17 AM
Hello, after some discussion yesterday I tried to use the spirit of patch #7036 (Sansa E200, ATA/LCD off) for lowering power consumption of te iPOD 5G. I tested 2 things first: Disabling (imho) unused devices of PP502x (OPTO, ATA, PIEZO, FIREWIRE) for whole runtime and disabling IDE0 if disk is not used (HAVE_ATA_POWEROFF-define).
Result ist -> no effect at all.
Does anyone know, if the registers defined in PP0520.h (DEV_EN, DEV_RS etc.) and the bitmasks (DEV_OPTO, DEV_IDE0 etc.) work?

Next try will be the "LCD-off"-stuff. I am not quite sure the defined bitmask will have effect...
Title: Re: Experiments on power consumption (iPOD 5G, 30G)
Post by: Buschel on May 22, 2007, 02:48:46 PM
Also no results on LCD-stuff. Seems like the defines / addresses / bitmasks which are contained in PP5020.h do not match the iPOD-reality  :-\
Title: Re: Experiments on power consumption (iPOD 5G, 30G)
Post by: saratoga on May 23, 2007, 05:20:55 PM
Have you found anything?  If not, I suggest logging onto IRC sometime when Europeans are likely to be awake, and asking there.  That or the mailing list are often a better way to get answers to development questions, since the forums are largely support.
Title: Re: Experiments on power consumption (iPOD 5G, 30G)
Post by: patwa on May 23, 2007, 07:14:38 PM
Hi, just saw this topic and wanted to post my benchmark for the 30GB Video.  I'm getting about 20 mins of playing time for 10% battery usage, so that's 200 minutes on average for a full battery.

this is playing at max volume, max bass and treble, hardware and software equalizer set, crossfade and crossfeed, replay gain, etc.  I can post my settings file later if it helps.

Given that the normal iPod is supposed to last 5/6? hours, 3 hours 10 mins isn't that bad when you consider the superior sound quality (and features), but I wonder if the power consumption issue was improve on the 80GB port?  Even if it wasn't, taking the same ratio one should still expect 9/10 hours from it.

Interesting throughts about disabling processes and hardware functions, I wonder if it's possible to disable the LCD altogether? I know this won't suit most people, but I use the speech only so don't really need it. How much would this add to the battery life?

H.
Title: Re: Experiments on power consumption (iPOD 5G, 30G)
Post by: Llorean on May 23, 2007, 07:24:04 PM
Basing it on 10% use in 20 minutes is a bad assumption. Run a full battery bench, as that 10% isn't strictly reliable.
Title: Re: Experiments on power consumption (iPOD 5G, 30G)
Post by: Buschel on May 24, 2007, 12:36:54 AM
@Saratoga: No, I did not find anything yet.

@Patwa: The battery lifetime depends on your settings and the file formats you are playing. E.g. EQ costs additional CPU (and therefore battery lifetime), .mpc will consume less power than .mp3-playback. Also: the higher the bitrates of files, the higher the power consumption (through more often HDD-activities as well as higher CPU-load). Good idea is to run a battery bench and post all your settings as well as the file format, bitrate and overall size of the folder.

@All:
From iPOD-site -> iPOD Video runtime = 14h with 400mAh-battery -> 29mA consumption
Rockbox (from battery benchs) -> 7.5 - 8h with 400mAh -> 50-53mA consumption
Sansa E200 can save 4mA through disabling ATA if not used
Sansa E200 can save 12mA through disabling LCD if backlight off
-> If Sansa E200-patches could be done on iPOD Video the runtime would be able to go up to 400 / (53-4-12) = 10.8h

Another question is the HDD itself. From Toshiba-site: 30GB HDD in sleep mode takes 20mA. Is it possible to switch off/on the whole HDD power consumption and do re-inits on demand?


Title: Re: Experiments on power consumption (iPOD 5G, 30G)
Post by: DefineByte on May 24, 2007, 06:53:36 AM
Could the method for turning off the LCD be brute-forced or is there not enough to go on?
Title: Re: Experiments on power consumption (iPOD 5G, 30G)
Post by: Buschel on May 25, 2007, 06:49:52 AM
From my side I cannot go on as the information about how to disable components like LCD is not available.
Title: Re: Experiments on power consumption (iPOD 5G, 30G)
Post by: Llorean on May 25, 2007, 12:53:18 PM
As a note, the Apple Firmware very clearly doesn't turn off the LCD, so that alone at least isn't one of the major causes of concern.

While the Sansa is unreadable without the backlight on (or very nearly so), iPod LCDs are very readable without the backlight on and turning them off is probably a usability handicap (though might be acceptable as an option.)

I do think we already turn the iPod HDs off (though I'm not certain, I just know we do it on some other players).
Title: Re: Experiments on power consumption (iPOD 5G, 30G)
Post by: linuxstb on May 25, 2007, 12:59:53 PM
I do think we already turn the iPod HDs off (though I'm not certain, I just know we do it on some other players).

Rockbox does power-down the hard disk on most players, but not ipods (beyond spinning the disk down and putting it in sleep mode).
Title: Re: Experiments on power consumption (iPOD 5G, 30G)
Post by: Buschel on May 25, 2007, 01:26:47 PM
@Llorean: Nevertheless turning off LCD (at least configurable for own build) could be a good point for additional power savings.

@linuxstb: Yes, putting the disk in sleep mode is what I understood from the source. Do you know how to completely power down the disk?
Title: Re: Experiments on power consumption (iPOD 5G, 30G)
Post by: linuxstb on May 25, 2007, 01:40:11 PM
@linuxstb: Yes, putting the disk in sleep mode is what I understood from the source. Do you know how to completely power down the disk?

If I (or anyone else) knew how to do that, it would be in the code in a heartbeat.

AFAIK, no-one has done any serious investigation (i.e. disassembling the original firmware and/or experimenting with the aid of a multimeter to investigate how different hardware settings affect power consumption) of the ipod hardware - which is what's required.
Title: Re: Experiments on power consumption (iPOD 5G, 30G)
Post by: mnhnhyouh on May 25, 2007, 08:03:53 PM
Just a quick side note, and possibly a helpful hint, the power drain on the iRiver H320 was discovered, IIRC, by a thermal camera. This showed extra power consumption by the USBOTG chip.

Worth a try for the iPod?

h
Title: Re: Experiments on power consumption (iPOD 5G, 30G)
Post by: Llorean on May 25, 2007, 08:16:16 PM
Not likely to be. This has been suggested a few times, but the general feeling is that it's the PortalPlayer chip that's the problem, since the problem persists on all PP targets. Using an IR thermometer would just tell us what we already suspect: It's the PP chip and we don't know enough about it to know how to resolve it.
Title: Re: Experiments on power consumption (iPOD 5G, 30G)
Post by: robin0800 on May 26, 2007, 09:26:42 AM
Just wonder if using electronic test equipment Logic Analyzers Ociliscopes etc the PortalPlayer chip, when running original firmware could give up its secrets, power supplies, data lines, signal input lines and outputs.
Just a thought.