Screen off when hold on in WPS

[darkslategrey]

On the newer iPod models(3g nano, classic, and later), the screen will for the most part shut off if left idle while playing audio except for a time and battery readout. Would it be possible to implement a similar feature in Rockbox?

[saratoga]

http://www.rockbox.org/tracker/task/8523

[dreamlayers]

I was thinking about a similar idea recently.  There could be a configuration option: power off screen when hold is on.  When I turn on hold, I'm generally putting the device where I can't look at it.  When I'm looking at the screen, I don't need hold to avoid unwanted button presses.

[psycho_maniac]

what player do you have dreamplayers? i have had a gigabeat and ipod 5.5g and they both had a "backlight on hold" feature for a long time.  or maybe i read the post wrong. you want the whole screen off while hold is on?

[dreamlayers]

I'm thinking of my Video iPod.  I also have the option to turn off the backlight when hold is on.  I think it would be nice to also have an option to turn off the screen.  When the screen is off, the BCM (DSP chip interfacing to the screen) could also be turned off or put to sleep (if a way to do this can be figured out).

[karashata]

I'm thinking of my Video iPod.  I also have the option to turn off the backlight when hold is on.  I think it would be nice to also have an option to turn off the screen.  When the screen is off, the BCM (DSP chip interfacing to the screen) could also be turned off or put to sleep (if a way to do this can be figured out).

I don't think the BCM chip is even used in Rockbox because there's no information available for it, and I believe said chip is only used to decode video in the OF, rather than all processing all of the information sent to the screen...

Someone correct me if I'm wrong...

[dreamlayers]

I don't think the BCM chip is even used in Rockbox because there's no information available for it, and I believe said chip is only used to decode video in the OF, rather than all processing all of the information sent to the screen...

Someone correct me if I'm wrong...

Rockbox is forced to use the BCM chip because the LCD is connected to it.  See firmware/target/arm/ipod/video/lcd-video.c.

What's going on there is Rockbox is repeatedly issuing command 5 to update a region of the screen.  (Commands are encoded as (~cmd << 16) | cmd.)  The command's arguments are placed starting at 0xE0000.  Rockbox initially sets the region to the whole screen, and each update only sends data which needs to be updated.

Before Rockbox starts, firmware in flash bootstraps the BCM chip, which includes loading the vmcs part of the flash into it.

Update:

I just made some current measurements on my 30GB 5th generation iPod using FS#9728.  That patch needs to be tested, but anyways, here's what I got:

One part of the shutdown sequence saves 4.5 mA by shutting down the LCD.  The next LCD update command turns it on, so implementing this would be trivial.

A later part of the shutdown sequence saves 3.5-3.6 mA, probably by cutting power to the BCM.   After this, the BCM probably needs to be bootstrapped before it can be used.  This is a multi-step process which takes a bit of time.

Total savings is just over 8 mA.  Data is in a Google Docs spreadsheet.

Update2:  Measurements of FireWire input current seem to confirm those results.  See the last page of the PDF attachment at http://www.rockbox.org/tracker/task/9728#comment27942

[dreamlayers]

Sorry about the double post, but it's been a while since the last post was made, and people may not have seen the updates.

I've written code to initialize the BCM and load firmware from the vmcs part of the 5G flash.  This means it is possible to shut down the BCM along with the LCD, for a total savings of at 7 to 8 mA.  Is this feature wanted and how should it be integrated? 

I'm thinking the "Backlight on Hold" setting should be changed to "When on Hold", with the following options: "Normal", "Backlight Off", "LCD Off", "LCD On" and "LCD and Backlight On".  I'm not sure what's the point of the "On" options but I guess I should have them because there currently is an "On" option.

I think there should probably also be an "LCD Timeout" setting in "LCD Settings".  Because there is a voice interface, there should probably be an "Off" option, even though that allows the user to turn off the LCD and not be able to see the interface to turn it back on.  Maybe I'd have to make sure that either the voice interface or the LCD are enabled?

The main limitations are that initializing the BCM takes about a second, and that when the LCD is off there is no way to see that the iPod is turned on.

BTW.  BCM Initialization can also be used to disable TV out and return the BCM to its normal level of power consumption.

[Llorean]

Maybe name the option "LCD on Hold" with choices "Normal", "Off", "Backlight off" and "Backlight On." I'm not sure how "LCD On" would be any different from "Backlight Off" and "Backlight On" implies the LCD is on, I think.

Another option is to ignore all this have have a simple "Yes/No" option for "LCD off with backlight" that simply turns the LCD off any time the backlight is off, which would mean no other options are necessary.

[dreamlayers]

Maybe name the option "LCD on Hold" with choices "Normal", "Off", "Backlight off" and "Backlight On." I'm not sure how "LCD On" would be any different from "Backlight Off" and "Backlight On" implies the LCD is on, I think.

Yes, "LCD on Hold" seems like a better choice.

I was thinking "Backlight Off" meant the backlight is off, and the LCD is subject to timeout, and "LCD On" meant the backlight is subject to timeout and the LCD is always on.  But yeah, this may be too many options.

Another option is to ignore all this have have a simple "Yes/No" option for "LCD off with backlight" that simply turns the LCD off any time the backlight is off, which would mean no other options are necessary.

The 5G LCD can be usable just with ambient light, and the backlight uses much more power than the LCD.  Because of this, I think they should be controlled separately.

Right now I'm thinking I'll first do a simplified "LCD on Hold" menu, and then I can add a separate LCD timeout if people want that.

[soap]

Not to take this too far off topic - but if you can save 7-8mA with the BCM off (That's, what,  ~20% of playback current?)
...What does that reduce the total power consumption to when everything we know how to turn off is off and the cpu is set to its lowest speed?

[dreamlayers]

Not to take this too far off topic - but if you can save 7-8mA with the BCM off (That's, what,  ~20% of playback current?)

Yeah, about 20%

...What does that reduce the total power consumption to when everything we know how to turn off is off and the cpu is set to its lowest speed?

That's not a simple measurement which can be taken now.  Additional code would have to be written to shut off some things.  For example, I don't think the codec is ever put to sleep and, other than the accessory power supply, no PCF power supplies are ever shut down.

BTW.  According to firmware/drivers/pcf50605.c, PCF5060X_D3REGC1 is the LCD voltage supply.  Shutting that off might save additional power when the LCD is off, but am I be sure that is safe?  I guess I need to do some more research.

Update: I probably should have looked at the code a bit more before asking about what to implement.  I just found out that Rockbox already has support for the hardware-independent aspects of LCD shutdown.  The following defines seem relevant:

Code: [Select]

#define HAVE_LCD_ENABLE
#define HAVE_LCD_SLEEP
#define HAVE_LCD_SLEEP_SETTING
#define HAVE_LCD_SHUTDOWN

I'm going to use these now.  LCD shutdown options relating to hold aren't in Rockbox.  Those can be added later as a separate patch.

Update2: Here's the patch: http://www.rockbox.org/tracker/task/9890