User avatar
HermannSW
Posts: 1313
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

[SOLVED] Question on [1-3]×5000lm "-set" exposure times

Tue Feb 05, 2019 7:39 pm

This is a followup of thread "Is 5000lm+5000lm leds = 10000lm with this 50W driver?"
https://www.raspberrypi.org/forums/view ... ?p=1322360

I had ordered two 50W led drivers (identical to the 48V DC one I have) and received long ago.
Today I did add poowr switches to the new two led drivers and connected all three drivers to the three leds mounted nearby:
Image

I did store all images for the 8 experiments here:
test.none.jpg
test.1.jpg test.2.jpg test.3.jpg
test.12.jpg test.13.jpg test.23.jpg
test.123.jpg


Diagram share link:
Image

The node values are exposure times, the leds in photo are numbered 1-3 from left to right.
The exposure values were determined with this command for 2+3 case:

Code: Select all

$ raspistill -t  30000 -o test.23.jpg -set
mmal: Exposure now 3733, analog gain 256/256, digital gain 1059/256
mmal: AWB R=311/256, B=501/256
...
mmal: Exposure now 2996, analog gain 256/256, digital gain 256/256
mmal: AWB R=321/256, B=508/256
$ strings test.12.jpg | grep exp
1st column is for no 5000lm led on.
2nd column is for 1, 2 and 3 led on.
3rd column is for 1+2, 1+3 and 2+3 leds on.
4th column is with all three leds on.

The values in 2nd column are slightly different.
What I don't understand is the value for 1+2, I would have expected it to be more like 1+3 and 2+3.

Here is just one photo as example, with all three 5000lm leds on:
Image


For completeness the complete lines grepped from the images:

Code: Select all

$ strings test.none.jpg | grep exp
ev=-1 mlux=-1 exp=124984 ag=1058 focus=255 gain_r=0.710 gain_b=2.363 greenness=-19 ccm=5446,-1862,518,-670,4328,444,416,-3630,7316,0,0,0 md=0 tg=2844 4001 oth=0 0 b=0 f=2844 4001 fi=0 ISP Build Date: Nov  4 2018, 16:34:19 VC_BUILD_ID_VERSION: ed5baf9520a3c4ca82ba38594b898f0c0446da66 (clean) VC_BUILD_ID_USER: dc4 VC_BUILD_ID_BRANCH: master 
$ 
$ strings test.?.jpg | grep exp
ev=-1 mlux=-1 exp=4509 ag=256 focus=255 gain_r=1.191 gain_b=2.304 greenness=-9 ccm=6022,-2314,394,-936,4728,310,300,-4324,8126,0,0,0 md=0 tg=281 281 oth=0 0 b=0 f=281 281 fi=0 ISP Build Date: Nov  4 2018, 16:34:19 VC_BUILD_ID_VERSION: ed5baf9520a3c4ca82ba38594b898f0c0446da66 (clean) VC_BUILD_ID_USER: dc4 VC_BUILD_ID_BRANCH: master 
ev=-1 mlux=-1 exp=5019 ag=256 focus=255 gain_r=1.347 gain_b=1.625 greenness=-49 ccm=6082,-2028,48,-1212,5468,-156,270,-3890,7722,0,0,0 md=0 tg=266 266 oth=0 0 b=0 f=266 266 fi=0 ISP Build Date: Nov  4 2018, 16:34:19 VC_BUILD_ID_VERSION: ed5baf9520a3c4ca82ba38594b898f0c0446da66 (clean) VC_BUILD_ID_USER: dc4 VC_BUILD_ID_BRANCH: master 
ev=-1 mlux=-1 exp=4452 ag=256 focus=255 gain_r=1.207 gain_b=2.253 greenness=-11 ccm=6022,-2314,394,-936,4728,310,300,-4324,8126,0,0,0 md=0 tg=271 271 oth=0 0 b=0 f=271 271 fi=0 ISP Build Date: Nov  4 2018, 16:34:19 VC_BUILD_ID_VERSION: ed5baf9520a3c4ca82ba38594b898f0c0446da66 (clean) VC_BUILD_ID_USER: dc4 VC_BUILD_ID_BRANCH: master 
$ 
$ strings test.??.jpg | grep exp
ev=-1 mlux=-1 exp=5151 ag=256 focus=255 gain_r=1.355 gain_b=1.621 greenness=-48 ccm=6088,-2004,18,-1236,5530,-194,268,-3854,7688,0,0,0 md=0 tg=271 271 oth=0 0 b=0 f=271 271 fi=0 ISP Build Date: Nov  4 2018, 16:34:19 VC_BUILD_ID_VERSION: ed5baf9520a3c4ca82ba38594b898f0c0446da66 (clean) VC_BUILD_ID_USER: dc4 VC_BUILD_ID_BRANCH: master 
ev=-1 mlux=-1 exp=2656 ag=256 focus=255 gain_r=1.218 gain_b=2.207 greenness=-13 ccm=6022,-2314,394,-936,4728,310,300,-4324,8126,0,0,0 md=0 tg=278 278 oth=0 0 b=0 f=278 278 fi=0 ISP Build Date: Nov  4 2018, 16:34:19 VC_BUILD_ID_VERSION: ed5baf9520a3c4ca82ba38594b898f0c0446da66 (clean) VC_BUILD_ID_USER: dc4 VC_BUILD_ID_BRANCH: master 
ev=-1 mlux=-1 exp=2996 ag=256 focus=255 gain_r=1.253 gain_b=1.984 greenness=-29 ccm=6022,-2314,394,-936,4728,310,300,-4324,8126,0,0,0 md=0 tg=277 277 oth=0 0 b=0 f=277 277 fi=0 ISP Build Date: Nov  4 2018, 16:34:19 VC_BUILD_ID_VERSION: ed5baf9520a3c4ca82ba38594b898f0c0446da66 (clean) VC_BUILD_ID_USER: dc4 VC_BUILD_ID_BRANCH: master 
$ 
$ strings test.123.jpg | grep exp
ev=-1 mlux=-1 exp=2297 ag=256 focus=255 gain_r=1.250 gain_b=2.023 greenness=-26 ccm=6022,-2314,394,-936,4728,310,300,-4324,8126,0,0,0 md=0 tg=287 287 oth=0 0 b=0 f=287 287 fi=0 ISP Build Date: Nov  4 2018, 16:34:19 VC_BUILD_ID_VERSION: ed5baf9520a3c4ca82ba38594b898f0c0446da66 (clean) VC_BUILD_ID_USER: dc4 VC_BUILD_ID_BRANCH: master 
$ 
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW

User avatar
HermannSW
Posts: 1313
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Re: [SOLVED] Question on [1-3]×5000lm "-set" exposure times

Mon Feb 11, 2019 1:11 pm

The effects of the three LEDs to the photographed scene were not similar enough.
I found instruction in the thread I pointed to on top of previous posting:
Imperf3kt wrote:
Wed May 30, 2018 8:04 pm
How are you focusing the light? emitting 5000lm is not the same as shining 5000lm on a surface.

Yesterday I started to arrage the three 5000lm leds symmtrically in a triangle. And I constructed a surrounding triangle tube from cardboard. Finally I did cover the inner walls with aluminum foil. I did all this in trying to "focus" the light at the other end of the tube. This is view from top (v2 camera "looks" through middle of the three leds to scene on ground blue plate):
Image


This is view from the other side, here you can see how far aluminum foil is used:
Image


I did store all images for the 8 experiments here:
test.none.jpg
test.1.jpg test.2.jpg test.3.jpg
test.12.jpg test.13.jpg test.23.jpg
test.123.jpg


Diagram share link:
Image


There is a small difference between led 3 and leds 1/2. But unlike in previous posting the focussing seems to have resolved the issues, now the determined exposure times ~1200µs/~600µs/~400µs are almost reciprocally proportional to lumen values 5000/10000/15000. Interestingly led 3 effect seems to be neglectable for 15000lm with all three leds on.


Find attached the protocol of the commands used for the measurement photos taken for completeness (too long for posting here directly). I did wait quite some time between the measurements to let the not cooled 50W leds cool down before next measurement. And different to provious posting commands I did time raspivid "-set" command for 9s only instead of 30s last time (I did not want to risk cardboard fire).


This is 15000lm lighted photo taken with all three leds on. It does not look that bright (after Automatic White Balancing, awb), but I can assure it was very very bright:
Image
Attachments
cmds.zip
contains cmds.txt protocol
(8.91 KiB) Downloaded 17 times
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW

User avatar
HermannSW
Posts: 1313
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Re: [SOLVED] Question on [1-3]×5000lm "-set" exposure times

Wed Mar 06, 2019 8:11 pm

I wanted to increase the time I can power on a 5000lm led by cooling it.
I did measure with laser infrared thermometer gun that after only 27s the temperature of 5000lm led reached 130°C.
I first did aim with laser into the hottest (top) part of led with left hand (because laser point becomes invisible after 5000lm led is turned on), then with right hand, started stopwatch, then 5000lm led and finally used new welding shield
https://www.raspberrypi.org/forums/view ... 7#p1438457
to be able to read current temperature while saving my eyes from the 5000lm.

I did buy cheap 40 x 40 x 20mm aluminum heat sinks that nearly fit perfectly on my 5000lm leds, not knowing whether they are capable of dealing with 50W. I did use Arctic MX-4 thermal compound paste to glue the heat sink onto the 5000lm led:
https://www.aliexpress.com/item/40-x-40 ... 76983.html

Measuring with laser infrared thermometer gun again, after 27s the temperature of 5000lm led was only 93°C. It took 1:42min to reach 130°C.

Last but not least I used the only working fan I have for experiments, a 5V 0.26A fan from a Raspberry case and blew onto the heat sink. It took more than 4 minutes to reach 130°C, the increase in temperature was very slow:
Image


Summary:
Either better heat sink or better fan is needed to keep temperature constant, best below 90°C.

But compared to using 5000lm leds without any heatsink like I did until today, the added heat sink allows for much slower temperature increase and longer test times for high framerate video capturing. Especially the "raspistill -set" tests I did for 30s do not reach 130°C anymore but keep below 100°C.
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW

User avatar
HermannSW
Posts: 1313
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Re: [SOLVED] Question on [1-3]×5000lm "-set" exposure times

Sat Mar 23, 2019 9:16 am

After successful test with 10000lm led and aluminum heatsink with fan for 200W yesterday, that kept 10000lm led temperature at 68°C even after 0.5h continuous operation,
viewtopic.php?f=43&t=215560&p=1445999#p1445999

I did the temperature measurements again for the 5000lm led with passive aluminum heat sink. The differences are:
  • no heatsink
  • 40mmx40mm heatsink now compared to 80mmx80mm heatsink yesterday
I did stop after roughly 3 minutes of operation when temperature rised above 115°C. I took this photo 2 minutes after powering off the 5000lm led, temperature was still showing 99°C, way above the 80°C (some sources say 60°C) operation temperature for the led:
Image


Something happened to the 5000lm led, perhaps it did fall one times to many. All 50 leds are still working , but they were equal spaced in both directions before, and are not now (photo taken through welding shield safety glass ‒ you can see the 10000lm led photo in the thread pointed to above for absolutely equal spaced leds there):
Image


So it seems there are 5 lines of 10 leds not absolutely fixated in the 5000lm led.
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW

User avatar
HermannSW
Posts: 1313
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Re: [SOLVED] Question on [1-3]×5000lm "-set" exposure times

Sat Mar 23, 2019 6:18 pm

I have three 50W led 38V/1500mA drivers I used with my 5000lm leds and they are fine:
https://www.aliexpress.com/item/High-Po ... 77294.html

I even used two in parallel to drive my 10000lm led successfully.

Today I received another 10000lm led from aliexperess, as well as a 100W led 38V/3000mA driver:
https://www.aliexpress.com/item/Smuxi-P ... 33982.html
Image


First I relatively quickly killed the new 10000lm led because I used it without heatsink and fan. With heatsink and fan running 10000lm led for 30 minutes was no issue and led temparature was 68°C only. Summary: while you can run 5000lm led without heatsink for 30 seconds, that is not the case for 10000lm led. Don't run 100W led without heatsink, even for short time (it can be already too long and kill the led), best with heatsink+fan.


I wanted to compare the brightness of running 10000lm led by two 50W led drivers in parallel, or by one 100W led driver. Since directly looking into the leds is no option, that task is difficult to do when looking through welding helmet safety glass.

So I did measure voltage and current for both scenarios:
100W led driver: 31.2V, 1.22A(!)
1×50W led driver: 32.1V, 1.45A
2×50W led driver: 35.5V, 2.88V

Summary:
While the 100W led driver states to be used with 100W 10000lm led, it does not deliver 100W.
I found a same model 100W led driver I seem to have bought months ago and tested that as well.
Unfortunately with the exact same (bad) measurements, current much below 1.5A, not anywhere near 3A.
Don't trust led drivers ‒ measure and verify instead!


While I can see easily through welding helmet safety glass a slight difference in brightness between 1x50W led driver and 2×50W led drivers connected in parallel to 100W 10000lm led, it is difficult to demonstrate this with android camera. I did take 1-handed photos outside of the welding helmet safety area with android camera. On the left you see photo with 2x50W, on the right side with 1x50W. You can see details on the right that are not on the left, but in genaral it is just very bright.


Image
⇨https://stamm-wilbrandt.de/en/Raspberry_camera.html

https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://gitlab.freedesktop.org/HermannSW/gst-template
https://github.com/Hermann-SW/fork-raspiraw
https://twitter.com/HermannSW

Return to “Camera board”