Power Consumption - some Measurements - General Discussion and Assistance - CHDK Forum
« previous next »
Pages: [1] 2 3 4   Go Down

Power Consumption - some Measurements

  • 33 Replies
  • 12548 Views
*

Offline Hacki

  • ****
  • 358
  • SX100
Power Consumption - some Measurements
« on: 30 / September / 2008, 09:55:57 »
Advertisements
I hooked my SX100 up my laboratory power supply, and made some measurements regarding the power consumption.
All measurements were done with the screen brightness set to minimal.
Setting the screen brightness to average adds an additional power consumption of 40mA. Setting it to maximum adds
70mA. Disabling the screen by sticking something in the A/V out saves 70 mA (70mA less than minimum brightness)

SX100 Power consumption @ 2.5V

Without CHDKWith CHDK
Playback mode190 mA190 mA
"P", idle530 mA~535 mA
"P", half press670 mA~675 mA
"P", idle (MF)1670 mA680 mA
"P", half press (MF)670 mA680 mA
"Video" mode, idle   660 mA670 mA
Recording video680-780mA680-780 mA
Charging flash810 mA (2 sec)810 mA (2 sec)
Zebra, hist. idle-540-550 mA
Zebra, hist. half press - 670-690 mA
Auto DR curve processing-630 mA
Raw sum/average   (Play)2-230 mA
Raw sum/average   (Rec)-570 mA

1) Indeed, enabling manual focus causes the camera to consume 140mA more! Even when i disable the "MF Point zoom" Magnification thing, the additional power consumption stays.

2) Apparently those CPU intensive tasks dont need that mach power after all. 40mA additionally only. Custom Curve processing on the other hand needs 100mA more.


As it seems, running CHDK actually causes a slightly increased power consumption, but obviously not mote than 5-10mA. Unfortunately my power supply isnt that accurate, it only shows steps of ten (0.44 for 440 mA e.g.), so if the power consumption is fluctuating between 530 and 540, i can only guess.. (that when i rounded to ~535 e.g.).

Enabling zebra & histogram needs some a small amount of additional power too..  Then again, those increases are so minimal, it's not worth to worry about.

The manual focus thingy is probably interesting for time lapse photographers.. Running manual focus causes the camera to use around 140 mA more power constantly, using a normal focus method needs up to 960mA with every half press - but only for a short time, around half a second. So, when the camera is set to shoot only every 15 seconds for example, it would need less power to use autofocus...

BTW: The SX100 runs fine with voltages from 2,3V (Batter indicator starts blinking below 2,25V) up to 3,70V.
I didn't want to raise the voltage more, as i'm afraid of frying my camera, but i thought that it might interest someone that the cam runs with a range of voltages.. I guess there's no real need to throw out 50 bucks for a canon power supply..

Edit: I totally forgot about the power saving mode! When my cam goes into this mode, it only needs 140mA. I guess it would be quite nice if we had an ubasic or lua command to activate this mode, so the cam can be sent to sleep when its doing nothing.. (Again, useful for time lapse photography)
« Last Edit: 30 / September / 2008, 10:09:36 by Hacki »
Logged
*

Offline wontolla

  • ****
  • 413
  • S3 & G9 & A720
Re: Power Consumption - some Measurements
« Reply #1 on: 30 / September / 2008, 14:17:38 »
Very interesting experiment Hacki!

Now we have numbers to backup the theory of CHDK not causing much battery burn.

The manual focus thing is of course going to take some power because it involves the mechanical parts of the camera.

I always thought that the display was responsible for much of the energy consumption, so it surprises me that the histogram and zebra don't use much of the energy, considering the constant update of the screen.

It looks like those DIGIC processor are really low power. And most of the CHDK features concern the processor only.
Logged
*

Offline Hacki

  • ****
  • 358
  • SX100
Re: Power Consumption - some Measurements
« Reply #2 on: 30 / September / 2008, 14:31:11 »
Code: [Select]
The manual focus thing is of course going to take some power because it involves the mechanical parts of the camera.
Well yes, when the focus is actually doing something, its completely logical that it burns some energy. But i meant that the camera needs more power if i set it to manual focus and leave it idle - it does nothing, just sits there, waiting for input, and still - it burns more energy when the focus method is set to "MF" instead of "Macro" or "Normal".

Logged
*

Offline reyalp

  • ******
  • 13290
Re: Power Consumption - some Measurements
« Reply #3 on: 30 / September / 2008, 14:58:14 »
Very interesting. I too find the MF thing quite surprising.

Might be interesting to monitor the clockspeeds while you do this.
Logged
Don't forget what the H stands for.
*

Offline NewbieToobie

  • ***
  • 102
Re: Power Consumption - some Measurements
« Reply #4 on: 30 / September / 2008, 17:18:54 »
Neat info!

Manual focus: Here's my guess. The lens elements move freely. They have to to move quickly. There is no high-viscosity lubricant on the mechanics as in old manual focus lenses. Which kept the focus and zoom positions set no matter how you carried the camera. But even then some larger lenses would shift focus or zoom with the lens held vertically. In these cameras the lens must be magnetically "locked" during manual focus to ensure it stays where you set it no matter how you orient the camera. A constant "apply voltage to keep the lens there" situation.

Thanks for the info. I use MF mode a lot. Now I'll be sure to turn it off when not really needed.

More thinking: I bet this is why Canon didn't want to let you set MF as a default power-up option to save for Custom settings. They were trying to extend battery life, for the consumer as well as for advertising. Makes sense. Now I don't hate them so much. :)
« Last Edit: 30 / September / 2008, 17:25:31 by NewbieToobie »
Logged
*

Offline Hacki

  • ****
  • 358
  • SX100
Re: Power Consumption - some Measurements
« Reply #5 on: 30 / September / 2008, 17:26:47 »
Nice guess - i'm sorry to disappoint you, but youre wrong. ;)

For some reason, the LCLK increases to 72 Mhz in MF mode. I checked it with the modifications from the link reyalP gave me. Could be related to the MF "magnification glass" thing, i guess.

reyalP pointed out in IRC that the clock speed can be set somehow.. so i assume it should be possible to reduce the LCLK speed again when MF is enabled.. Dont know if that had any drawbacks. I assume in the worst case, the camera would crash.

Then again, its possible that this isnt a feature, but rather a bug. Maybe some developer at canons made a mistake, it could be possible that this behaviour only occurs on my camera. I'd appreciate it if someone else would check the LCLK stuff on a camera with MF mode.

Logged
*

Offline reyalp

  • ******
  • 13290
Re: Power Consumption - some Measurements
« Reply #6 on: 30 / September / 2008, 18:12:11 »
Careful. Although it seems likely that the higher lclk is responsible for some of the power drain, it's not clear that it is the only cause.

If you could control it, you could try just forcing it to 72 in another mode, and seeing if there is a similar increase in power draw. ISTR there were other situations where lclk changed around, so you might be able to get some just by looking for a correlation between lclk and power draw in various situations.

I would be surprised if the camera is using extra power to actively hold the lens hardware in position when you are in MF mode. It seems to me there would be better ways to achieve the same result.

On my camera (a540), it appears to stay at 1 (=36mhz) most of the time in shooting mode, whether in MF or not. It drops to 0 (=72mhz) when brightness changes. Normally in playback, it sits at 3, although in one test I did, it seemed stuck at 0 ???
Logged
Don't forget what the H stands for.
*

Offline fudgey

  • *****
  • 1705
  • a570is
Re: Power Consumption - some Measurements
« Reply #7 on: 01 / October / 2008, 14:34:07 »
I did a quick test on my a570is hooked up to a lab psu at 3,15 V (Canon's nominal DC input). Input current increases in manual focus on this camera too, from about 420 mA to 490 mA (note that decreasing the voltage will of course increase input current...it was about 445 mA and 520 mA at 3,00 V. The two AA batteries work at around 2,4 V most of their life, I didn't try a DC input that low since I feared it might make the batteries kick in. I guess they don't from what Hacki measured and the fact that my camera shut down during high-current zooming when I first tried with a 0R5 current meas resistor instead of the PSU's output reading.

There's significant noise in the current measurement results when done without a digital scope and RMS calculation as the input current constantly switches between two states (10 mA or so apart if my memory serves) in a semi-random digital fashion. Didn't bother going 4-lead for this one either so the voltage at the cam may have been slightly lower than mentioned.
Logged
*

Offline Hacki

  • ****
  • 358
  • SX100
Re: Power Consumption - some Measurements
« Reply #8 on: 02 / October / 2008, 15:13:47 »
Just tested what happens when i change the clock speeds manually.

Setting the LCLK speed to 72 Mhz in another focus mode causes the cam to only consume 10-15mA more. Therefore, i guess the additional power consumption in MF mode is caused by something else.

Then, i tried setting it to 18 Mhz while in record mode. The camera retracted its lens and showed a "E16" in the lower left corner. The second vers.req-page now shows:

Code: [Select]
Adj Ver.009.005
CcdFifo
2008.10.02 20:44:52


And, although i dont even know what those different clocks are, i also tried setting the MCLK and HCLK.

The hclk cant be changed, as it seems. Nothing happens when i call one of the function, the value shown in the debug-window doesnt change either.

Setting the MCLK crashes the cam instantly. Some tilted stripes are appearing on the screen, then after 10 seconds the watchdog kicks in and shuts the cam down.

Besides the E16 in the log, my cam still works fine.

BTW: MF works fine when the LCLK gets set to 36 Mhz.. Doesnt save much power though

In case someone wonders, there are functions for setting the clock speed in the firmware.

Here the addresses for the SX100 1.00c

Code: [Select]
NHSTUB(SetHClk18MHz                            ,0xFFC053E0)
NHSTUB(SetHClk36MHz                            ,0xFFC05418)
NHSTUB(SetHClk72MHz                            ,0xFFC0542C)

NHSTUB(SetLClk18MHz                            ,0xFFC05464)
NHSTUB(SetLClk36MHz                            ,0xFFC05474)
NHSTUB(SetLClk72MHz                            ,0xFFC05484)

NHSTUB(SetMClk18MHz                            ,0xFFC0535C)
NHSTUB(SetMClk36MHz                            ,0xFFC05394)
NHSTUB(SetMClk72MHz                            ,0xFFC053A8)



Logged
*

Offline RaduP

  • *****
  • 908
Re: Power Consumption - some Measurements
« Reply #9 on: 04 / November / 2009, 14:26:21 »
I did some tests on my SD980, but unfortunately I can't connect it to the lab power supply, because it has a non standard connector.

Anyway, I wrote some code that is supposed to reduce the power usage of the camera. The code looks like this:

Code: [Select]
void wait_for_intrerrupt()
{
asm volatile (
"MCR p15, 0, R0, c7, c0, 4"
);
}

void deep_sleep(int duration)
{
int i;
int g;

for(i=0;i<duration;i++)
{
*((volatile int *) 0xc0220130) = 0x46;
for (g=0; g<1; g++) // Wait a while
{
    wait_for_intrerrupt(0);
}
*((volatile int *) 0xc0220130) = 0x44; // Turn off LED
for (g=0; g<1000; g++) // Wait a while
{
    wait_for_intrerrupt(0);
}
msleep(10);//let the camera do some other stuff if needed
}

}

void camera_hibernate(int duration)
{

int ret;
*(int*)0x29EC=0;//trick the camera to think the lens is retracted
_StopWDT();
_Rec2PB();
msleep(3000);
_turn_off_display();
*(int*)0xC0220018=0x44;//turn off SD power, after switching to play mode, because the camera might need to read the last image from the SD card
deep_sleep(100);
_StartWDT();
_turn_on_display();
*(int*)0xC0220018=0x46;//turn on SD power
_PB2Rec();
*(int*)0x29EC=1;
}

camera_hibernate() is called from a basic script:

Code: [Select]
@title Hibernate test
:loop
hibernate 5
print "next.."
sleep 2000
shoot
goto "loop"
end

So I charged the battery, and fired it up.. The camera survived for slightly less than 8 hours, taking a photo every 15 minutes or so.
I am not impressed at all with the power saving, I was hoping it would last at least 12 hours. The camera was eating something like 130 mA/h, @ 3.7v. That's almost half a watt in standby!

Anyway
Logged
Pages: [1] 2 3 4   Go Up
« previous next »
 

Related Topics