Power consumption: a590 measurements and waveforms.

[bugsplatter]

Hi there,
I've done a set of measurements with a590-101b.  This work confirms fudgey's observations and also provides screenshots of how various camera functions draw power.  http://bugsplatter.id.au/chdk/a590-power.html -- lots of pictures 

Comments? 

[reyalp]

Excellent!

Some things to try:
AF lock vs manual focus mode
LCD backlight on/off. (set_backlight in script)

If you want to try some hacking, you can try monitoring the various clock speeds http://chdk.setepontos.com/index.php/topic,2139.msg20087.html#msg20087 or even forcing them to a specific value and watching the effect

[fudgey]

Nice! Good of you to go through the trouble to make all them images... I've been meaning to do that for quite a while.

I'm not sure how similar a590 is compared a570 in this respect, but did you try powering up to rec mode directly and see if the draw is any different from powering up to play and then switching to rec?

I ask this because I've found that my camera just plain fails to start up (demands I change batteries) to rec mode using batteries whose resistance is too high (old or empty), but will always start to play mode and then it will often remain powered on when switched to rec from play mode...

After it's finally running, I can take a 1-hour 1000 image timelapse with that fully charged pair of old NiMH cells! So either the camera draws more current when powering up to rec than it does switching from play to rec or the battery condition algorithm is more trigger happy during power-on than later when it has gathered a little bit more information on the current/voltage behavior of the cells.

To be complete, I believe two things are missing from your measurements: effects of set_backlight and video plug. But that's not a big deal, they're just two numbers that don't need plots from a 'scope.

[fe50]

Good stuff !

What also would be interesting: the behaviour after using Canon's "Display off" feature (LCD off, registered to the print/share button) - since this normally cuts off the sensor (and also the AV-out electronics ?) the current should be the same as in record idle mode...

[bugsplatter]

Some things to try:
AF lock vs manual focus mode
LCD backlight on/off. (set_backlight in script)

That be different than pressing DISP button?  Yeah, I'll have a look.

If you want to try some hacking, you can try monitoring the various clock speeds http://chdk.setepontos.com/index.php/topic,2139.msg20087.html#msg20087 or even forcing them to a specific value and watching the effect

Well, I might have seen that indirectly as the cam takes more power when set to 'auto' or 'heart' mode.  This is in a later group of shots I took in the early hours, not on the page yet.

Edit: up now: http://bugsplatter.id.au/chdk/a590-power.html with some on page links near top.  Got AF lock vs MF mode for reyalp, yet to add other fudgey and fe50 suggestions.

[reyalp]

That be different than pressing DISP button?  Yeah, I'll have a look.

Yes, the display and sensor live view still run, so you can do MD etc.

[bugsplatter]

I'm not sure how similar a590 is compared a570 in this respect, but did you try powering up to rec mode directly and see if the draw is any different from powering up to play and then switching to rec?

Done that.

I ask this because I've found that my camera just plain fails to start up (demands I change batteries) to rec mode using batteries whose resistance is too high (old or empty), but will always start to play mode and then it will often remain powered on when switched to rec from play mode...

Now this is where it gets interesting.  Last couple hours I been playing with the cam at low voltage, and it seems the cam will refuse to ''load firmware'' if the battery voltage is too low.  Makes sense when you think about it.  So when I get the cam operating and the lowest reliable voltage, starting up in play mode and loading firmware fails with a ''Change batteries'' message.

As well, I'd made a longer cable for the linear regulator I'm using, and this, combined with the 0R1 current shunt proved an interesting combo -- like a battery with higher internal resistance.

So I paralleled another 0R1 to make a 50-mOhm shunt and switch to a shorter 3V cable -- made a big difference to the measurements I was taking.  Having a break, I'll revisit later.

But I can tell you that at 2.3V the cam draws ~600mA in active record mode, vs ~400mA at 3V.  The cam doesn't like loading CHDK at low battery voltage, probably a design feature to allow successful firmware flash? 

More later.

[bugsplatter]

Good stuff !

What also would be interesting: the behaviour after using Canon's "Display off" feature (LCD off, registered to the print/share button) - since this normally cuts off the sensor (and also the AV-out electronics ?) the current should be the same as in record idle mode...

I'll get there, quite an interesting exploration.  Helps me put off looking at the a580 port and/or fixing some of the missing bits for a590 -- such a lot to play with here

[fudgey]

I'm not sure how similar a590 is compared a570 in this respect, but did you try powering up to rec mode directly and see if the draw is any different from powering up to play and then switching to rec?

Done that.

Thanks. I guess that proves it's not drawing more current.

Now this is where it gets interesting.  Last couple hours I been playing with the cam at low voltage, and it seems the cam will refuse to ''load firmware'' if the battery voltage is too low.  Makes sense when you think about it.  So when I get the cam operating and the lowest reliable voltage, starting up in play mode and loading firmware fails with a ''Change batteries'' message.

Now that is interesting and does make total sense to require a fresher set of batteries for firmware upgrades than for normal everyday operation. I'll have to try experiment with starting the cam without CHDK using old batteries to verify your findings... would do it right now but the camera's out shooting a timelapse. If this is true, it could very well explain why more than a few people think CHDK makes their cameras consume more current (which, now that I've said it, is something you're welcome to prove wrong while you're at it...not that it hasn't been done a few times before ). I can't think of much that could help, except starting via play mode to draw less current during startup.

So I paralleled another 0R1 to make a 50-mOhm shunt and switch to a shorter 3V cable -- made a big difference to the measurements I was taking.  Having a break, I'll revisit later.

Yea I once tried with a current sense resistor too large as well and got the cam to shut down each time it operated motors... didn't have time to tinker more that day.

But I can tell you that at 2.3V the cam draws ~600mA in active record mode, vs ~400mA at 3V.

My camera's is about 2% less efficient (power input wise) at 2.4 V than it is at 3.15 V. You seem to have a way bigger difference there (1.38 W vs. 1.2 W)... are you measuring the current you're supplying to the camera or the camera's input voltage (cable and current sense losses removed)?

[bugsplatter]

My camera's is about 2% less efficient (power input wise) at 2.4 V than it is at 3.15 V. You seem to have a way bigger difference there (1.38 W vs. 1.2 W)... are you measuring the current you're supplying to the camera or the camera's input voltage (cable and current sense losses removed)?

Measuring V and I at power supply regulator end of 25cm or 80cm cable to camera.


That's why I'm taking a break, cam is running a timelapse script test for some hours yet.  Ended up with too many variables and my mental model needs some forgetting  I've got the current sense in series with the cam 3V, so there's 50mV/A to account for there.  Now I have two cam 3V cables, one is 25cm, other is 80cm, so there's a V drop for that.  From memory about 1mV/cm at cam record active mode in chdk, assuming the cam Batt mV readout is accurate.

I now have the DPO calculating output voltage - shunt loss (no differential input), but I need to take a cable resistance measurement as well, tell the cam to show Batt mV in play mode.

While using the lower current shunt resistor and shorter cable stopped the cam shutting down when switching to record mode from play mode, it did not change the cam's refusal to ''load firmware'' at low voltage -- hence my conjecture on the firmware Batt mV safety factor.

So I need to take more care in measurements -- that's why I like taking the screenshots, others can then spot an interpretation boo-boo as long as I accurately describe how the measurements were taken.


Yesterday I also failed to create a bootable SD  I think cardtricks-144 is broken on Win7rc, as I had trouble getting a580 dump recently too.  But there seems more to making SD bootable than writing 'BOOTDISK' to 0x40 in MBR (from linux), as all I got was 'Card Locked!'  Maybe it has to be 'BOOTDISK',0x0?  Yet to try that.  So I was unable to check power up into CHDK mode.  But my recording of power up in play and load firmware showed no great increase in power consumption loading the CHDK and rebooting -- nowhere more than half record active mode power.

And yes, there's no appreciable difference powering up in record mode or switching to record mode from play mode.

could very well explain why more than a few people think CHDK makes their cameras consume more current

CHDK does not obviously draw more current, but there are things people could do to extend battery life.  Using the CHDK menu from play mode will consume about a third of the power compared to being in the menu from record mode.  It's not the LCD or backlight consuming the power, rather, it seems the cam's 'live view' accounts for two-thirds of the power in record mode (Edit: supported by temperature increase in CCD indicating power consumption).  Play mode power is not very different from record mode idle (within 25-30%).  This too is supported by the long play time battery life seen in the camera's specification.

So yes, there will be a followup, maybe later today or tomorrow, I have the DPO on extended loan, lots of time to play  

[bugsplatter]

Okay, here's the next set comparing various schemes for saving power, turning off the backlight is not that useful.


executive summary
Edit: See http://chdk.setepontos.com/index.php/topic,3872.msg36535.html#msg36535 for updated table


Notes:  Script turn off display was about the same as cam's idle mode, my test script's two second cycle time was a little fast for the DPO measurement.  

See the pictures and full story on http://bugsplatter.id.au/chdk/a590-power-2.html

Test script: http://bugsplatter.id.au/chdk/archive/lcdpower-2009-07-08a.bas.txt

Code: [Select]

@title lcdpower

@param a Test Number n 1-9

print "Press 'Shoot' to exit"
sleep 300
print " while display is on."
sleep 300

if a = 2 then goto "test2"
if a = 3 then goto "test3"
if a = 4 then goto "test4"

:test1
print "Test 1) set_backlight"
do
  set_backlight 1
  sleep 2000
  set_backlight 0
  sleep 2000
until 0 = 1

:test2
print "Test 2) Click 'display'"
do
  click "display"
  sleep 2000
until 0 = 1

:test3
:test4

end

:restore
rem 'shoot_full' trap
set_backlight 1
click "shoot_half"
sleep 400
print "Press 'set' to exit..."
wait_click 6000
is_key k "set"
if k > 0 then end
shut_down
sleep 3000
end


[bugsplatter]

Some more numbers - power vs battery voltage

        Record Active     Record  Idle        Record Active
        -------------     ------------        -------------
 V       mA      mW        mA      mW         CHDK Batt mV
----    ----    ----      ----    ----        ----
3.3      374    1234       109     359        3232
3.2      388    1242       112     360          -
3.1       -       -        117     363          -
3.0      414    1242       121     364        2930
2.9      432    1253       126     364          -
2.8      446    1249       131     366          -
2.7      466    1258       136     367        2635
2.6      486    1259       142     368          -
2.5       -       -        147     368        2437
2.4      530    1272       154     371        2339
2.3      556    1279       161     370        2254
2.2      580    1276       169     371        2140
2.1       -       -         -       -         2047

Note: these are raw numbers, no spreadsheet here to do the corrections for shunt + wiring V drop, I'll adjust the web page after download/install openoffice and adjust / interpolate the figures to fill the table. 

[fudgey]

I have now tried my a570is with and without CHDK on an old pair of 2700 mAh NiMH cells (fully charged a few weeks ago). The test was quick, only a couple dozen power cycles in total, but I think it was enough to get results.

The camera always starts to play mode and works perfectly while in play mode, with or without CHDK.

CHDK autoloading activated, camera only once powered on to REC mode without demanding to replace batteries and even then it immediately powred off when I half shooted. Same result from switching from PLAY to REC while CHDK is active (no matter if it was autoloaded or manually started); mostly it shut down right after switching, but once it let me press half shoot and then shut down. It always shuts down cleanly with the change batteries message.

CHDK disabled, the camera always starts to REC mode and also switches properly to REC from PLAY, and allows me to shoot successfully. It does blink the battery icon, but all in all it works.


Tampering with the battery test part of the firmware has been discussed before... many have doubts about safety, but IMO it now seems that there are two modes of operation and we're ending up with the wrong, trigger happy version meant for protecting firmware updates only meaning switching to the regular one can't be harmful if a way to do that is found.

[fudgey]

Oh and since you're being delightfully thorough, there's one more thing missing unless your camera is different from mine: play mode input power with lens out differs from input power with lens in on my a570 (0.654 W vs. 0.526 W @ 3.15 V input), this is of course with backlight and display enabled, without av plug.

[bugsplatter]

Oh and since you're being delightfully thorough, there's one more thing missing unless your camera is different from mine: play mode input power with lens out differs from input power with lens in on my a570 (0.654 W vs. 0.526 W @ 3.15 V input), this is of course with backlight and display enabled, without av plug.

Thanks   I'm taking requests while I have the DPO on loan, so ask for anything that will clarify / extend our collective knowledge for CHDK now.  I'm trying to provide hard data because there's much confusion in some of the long interval threads about best power saving methods. 

I think I've noticed what your're saying here, but not followed it up.  I'm forgetful, and easily distracted so I miss stuff or forget to go back and check anomalies.

Other thing is I'm including mW because you do, but battery life is a tad more complex as batteries are rated in AH, not WH.  And AH depends on load (SLAs rated at their 20 hour rate, for example), temp and stuff. 

Couple more little observations I wrote down but didn't put up here:

a) CHDK Batt mV displayed value seems to have a constant slope integrating filter the takes about 10 seconds to reach a step-changed battery value.  Dunno if that filter is in CHDK or the cam, but I noticed the camera operates on actual battery voltage for low-going event triggers (for example on the last picture of second measurement set when camera activated lens park and power down as soon as Batt V went below 2.0V ), not the displayed mV value.

b) Once low battery state is detected by the camera, it stays in lobatt state (flashing red icon) until power is cycled even when the voltage is increased again.

c) Going too low on battery voltage, the cam will shutdown without pulling lens, seems if it decides there's not enough power to complete the operation it's safer to leave the lens out.

d) Taking the battery voltage down very slowly, I could keep the camera active down to 1.9V, it seems the cam also uses a derivative (rate of change) value for determining when batteries near end of life.  This one makes sense too if you consider that as battery capacity nears end, voltage drop and camera's current demand increases, thus pulling battery voltage even lower, faster, near end of battery life.