I try to find a light detector independent of ambient temperature.
Testing an LDR and a photodiode and found both vulnerable to temperature changes.
Speed is not important for my project, and the detected light is in visual spectrum 620nm (red). Not important if other colors are also detected.
Do you have any suggestion?
The good old CDS cell would work. Search digi-key.
Better check out these Adafruit Pro Trinket - 3V 12MHz : ID 2010 : $9.95 : Adafruit Industries, Unique & fun DIY electronics and kits
BH1750, data sheet shows low temperature variation in range -40 +80.
Thanks both for your quick response.
Ray, I visit the link you suggest, however didn't find detailed data for those sensors.
Master, this sensor seems super stable. Is there any version with analog output to connect it directly to an analog input?
No, I've newer seen analog version. There is another type of light sensors on a market - light to frequency.
TSL235R
Stable 150 ppm/°C temperature coefficient
Ray_Beebe:
The good old CDS cell would work. Search digi-key.
Better check out these Search Results for 'Light sensor' on Adafruit Industries
Mmmmm visible LDR are CDS cells?
Ya, noticed that. My bad.
You should be able to make any light sensor, that is resonablely repeatabable, more stable over a temperature range by using it in conjuction with a temperatur sensor. Say the light sensor gives 10k at 10lux at 25c and 11k at 10 lux at 100c. You would apply an offset of -1k at 100c or if the change is relatively linear from 25c to 100c subtract 40 Ohm per degree C. Using analogRead you get 1024 units to work with. If your 40 Ohm change equates to 2 units, you would subtract 2.
Ray_Beebe:
Ya, noticed that. My bad.
Anyway I have to thanks you because to be sure I made a search and discovered that infrared LDR are not made with CDS.
This temperature dependencie of LDR is not written in datasheet I found. How is the amount of it?
Unknown. Only way to know is to test with stable light source and change the temperature to get a good approximation over the temperature and light ranges of concern.
Master: Also this sensor is very stable, however is made for easy installation to counter type circuits.
Ray: the situation get very complicated. My project consist a laser diode, and 2 sensors. Laser is inside a small brass case, and sensors inside plastic.
I place the construction inside a box with a heat source, to control easily the temperature. My result are unpredictable.I take different data when temp rise, and different when fall. Seems some cells take temp quick some other not.
Additionally while laser probably is not stable either over temperature, most of the time laser hit one of those sensors, and I suspect temperature arise in this sensor cause of laser radiation.
Data below show the citation:
(Temperature change is not linear, not stable either)
Celsium Indirect sensor Direct sensor (LDR both)
23 1355KΩ 450Ω
24 1380KΩ 450Ω
25 1411KΩ 447Ω
26 1420KΩ 443Ω
27 1452KΩ 443Ω
28 1478KΩ 441Ω
29 1496KΩ 443Ω
30 1493KΩ 444Ω
31 1465KΩ 443Ω
32 1448KΩ 444Ω
33 1400KΩ 442Ω
34 1344KΩ 446Ω
35 1333KΩ 455Ω
36 1336KΩ 463Ω
I wasn't able to find some useful data about the temperature influence on LDR, I found only that there is an influence but not any amount of it.
But it seems that this influence is not very great.
mailmad, you are talking about temperature rise and fall but... how many degrees?
Do you need a precise light measurement or you need a sort of on/off or a coarse measurement? Are there rapid variations?
LDR are not fast and it seems that they suffer for prolongated light exposition so they take time when going dark.
But I never used LDR with laser, I made small experiments with phototransistors. Maybe your laser doesn't fit well the spectral response of the LDR.
zoomx:
I wasn't able to find some useful data about the temperature influence on LDR, I found only that there is an influence but not any amount of it.
But it seems that this influence is not very great.
http://akizukidenshi.com/download/ds/senba/GL55%20Series%20Photoresistor.pdf
Mine is probably GL5537-1
zoomx:
mailmad, you are talking about temperature rise and fall but... how many degrees?
Do you need a precise light measurement or you need a sort of on/off or a coarse measurement? Are there rapid variations?
Could be more than 30 degrees.
Precise analog reading.
Variations are difficult to predict.
zoomx:
LDR are not fast and it seems that they suffer for prolongated light exposition so they take time when going dark.
According to my expirements, dark make LDR sensors unfortunatly very very sensitive in temperature variations.
zoomx:
But I never used LDR with laser, I made small experiments with phototransistors. Maybe your laser doesn't fit well the spectral response of the LDR.
May phototransistor could be the solution to my problem. Only test will prove that. However, in some articles I read for photodiodes (the closest relative to phototransistor), for every 10 degrees output doubles.
I wondering, why none of the manufacturers doesn't produce a photo detector, with build in temperature compesation sensor, precalibrated from factory.
I plotted your data
In the dark it seems that there is a peak at about 30 degrees. In light it seems opposite.
I use an LDR but never noticed that because I believed that was only a noise in the measurement.
Are you sure that the temperature doesn't affect the other electronics?
There are others sensor, some are digital and use I2C or SPI for communication for example the TSL2561, there is a breakout made by Adafruit. But I am not sure that the temperature will not influence this one also. The datasheeet peak about temperature for soldering or working, not for effects.
You have make great job here!
Yes I see your point.
The experiment condition was those below:
A closed carton box with a size to host the device with arduino light detectors and laser, a small hair dryer as a heat source, and the probe of a remote wired thermometer.
Outside was a laptop to collect data by serial.print, remote thermometer, a power supply unit to supply device, and a dimmer to control hair dryer better.
I decide to collect data during down falling, because was smoother the change of temperature. First I heat up the box few degrees over the highest data point, and hold it there for few minutes, to allow components to heated evenly.
I'm sure temperature effect all component inside device, however after collecting data this weekend, I found indirect sensor to be effected WAY more than anything else inside device.
I will place new data here, in case you like to take a look.
Before I place them, I have to mention that I have make few changes in electronic circuit. Most important of them are:
Instead of pull down resistor for every sensor, I place LM334 adjustable current source each.
Laser have a constant current circuit.
I place a temperature sensor near to indirect sensor (5mm away).
Just take a look in the "Indirect light" column!
Temp Laser voltage Laser current Indirect light Indirect temp Direct light
23 463 203 742 458 458
24 464 202 739 459 449
25 465 202 724 459 442
26 465 203 742 456 439
27 465 202 718 455 431
28 465 207 712 455 431
29 466 206 700 454 439
30 467 206 686 454 439
31 468 206 676 451 433
32 470 210 660 451 433
33 470 209 644 452 432
34 472 209 624 450 433
35 473 210 609 449 432
36 474 211 586 445 436
37 475 212 572 448 437
38 477 212 549 448 430
39 478 213 530 447 415
40 479 213 516 447 412
41 480 214 486 447 408
42 480 215 452 447 407
43 482 215 426 446 391
44 482 215 393 448 393
45 483 214 352 446 394
46 483 215 315 446 390
47 484 217 281 445 384
48 484 218 232 445 385
49 484 218 193 442 383
50 484 220 151 443 387
51 484 218 122 444 392
52 484 218 81 444 386
53 483 218 54 444 402
54 484 220 30 443 402
55 484 221 30 443 399
As you see "Direct light" column doesn't change so dramatically as the inderect one. They have the same light source, so I expected to have relatively the same variation ratio.
Don't overlook light history effect or memory effect for both LDRs and CDS cells which in many cases is much larger than temperature effect. Look how much it can shift your readings:
Green band approximates fluctuations in resistance caused by memory effect at room temperature. Red band approximates effects of thermal drift and changes in optical coupling.
http://www.bilimteknik.tubitak.gov.tr/sites/default/files/gelisim/elektronik/dosyalar/40/LDR_NSL19_M51.pdf
http://edge.cs.drexel.edu/assemblies/tests/assemblies-200209/cds-cell-sensor.xhtml
dlloyd:
Don't overlook light history effect or memory effect for both LDRs and CDS cells which in many cases is much larger than temperature effect. Look how much it can shift your readings:Green band approximates fluctuations in resistance caused by memory effect at room temperature. Red band approximates effects of thermal drift and changes in optical coupling.
Resistive opto-isolator - Wikipedia
http://www.bilimteknik.tubitak.gov.tr/sites/default/files/gelisim/elektronik/dosyalar/40/LDR_NSL19_M51.pdf
http://edge.cs.drexel.edu/assemblies/tests/assemblies-200209/cds-cell-sensor.xhtml
So, I have an additional and greater problem except temperature. How can I deal with it?
I would try a different product, maybe something like these Digital Proximity and Ambient Light Sensors
Peak wavelength is about 620nm and very minimal and linear temperature effect ... should be easy to measure temp and calibrate to get accurate and stable readings. (see datasheet figure 3c).
I'm sure there's even more accurate sensors available, like those uses for light level meters, photography, medical, etc.
Red Enhanced High Performance Silicon Photodiode More...
It seems that there is also a variation of laser voltage and current with temperature
so it seems that there is a variation in laser light emitted.
Sorry for my delayed response.
I experiment until late at night, then I have to go to bed because I work next day, however can't sleep because I'm so excited for the project, and so on! For every day I live, Ι need 2 days more.
Dl After an experiment, I found LDRs to be very stable comparing to photodiode in temperature terms. I guess phototransistors behave the same cause of similar technology.
Very important information this of memory effect, which get worse in the dark, as the chart you post.
Is so important, where I have to make changes to my circuit to deal with it.
I would like to avoid digital sensors, for less code and circuit complexity.
May a solution could be the use of lower resistance LDR.
Zoom You have right. Laser, driving circuit, or voltage reference circuit is drifting.
I see in your chart, voltage and current get flat at top temperature. I wondering why...
News:
I change the angle of indirect LDR, and sensitivity hugely improved! (this minimize a lot the issues we discuss).
I have already order some alternative sensors to test them, however I probably will stick with LDR.
I think after that I approaching the end of project.
Remaining 2 things to solve
a) Contamination from ambient light.
b) LDR memory effect.
I would like to say the help from all of you is huge, and some time in a curious way!
I can't explain it exactly... I see "A", you show me "B" and "C", and through them I found "D".