OPT101 Photodiode

EricHayward

New Member
Attached is the normalized spectral response curve from the OPT101 photodiode datasheet. I attach the output from the photodiode to a 08M2 pin and issue a READADC command. Now suppose I shine a blue LED, with a wavelength of 500nm, on to the photodiode and the READADC reading is 21. With reference to the responsivity curve,is the value of 21 only 40% of the true value? Should the true value be 21/0.4 = 51?
 

Attachments

AllyCat

Senior Member
Hi Eric,

Welcome to the forum. In principle yes, but that curve is for the "short-circuit" current and the PICaxe ADC can measure only voltage.

So the photodiode must be loaded by a resistor and for a linear photoelectric response the resulting voltage (from a single silicon photodiode) should not exceed about 200 mV. A READADC (as opposed to a READADC10) will need at least 300 mV to indicate a value of 21, assuming the supply rail (i.e. the default ADC reference voltage) is 4 volts.

Cheers, Alan.
 

AllyCat

Senior Member
Hi again,

Hmm, 90 views and nobody (including the OP) appears to have picked up that the OPT101 is not simply a "Photodiode" but includes a transimpedance amplifier. So my warning above might well be redundant.

However, photometry is still a complex subject and we really ought to be told what you are attempting to achieve and to see a circuit diagram of your hardware and a listing of your software, to give any definitive answer. And what is meant by "true value". ;)

Cheers, Alan.
 

techElder

Well-known member
... value of 21 only 40% of the true value? Should the true value be 21/0.4 = 51?
How would anyone know how to answer such a specific question without developing the same hardware. Some posters don't give many of us a chance.
 

EricHayward

New Member
My apologizes for not mentioning that the OPT101 photodiode comes with its own amplifier.

I'm considering constructing a scanning spectrophotometer, similar to the one in the picaxe gallery submitted by nickcinquino. Basically a spectrophotometer splits polychromatic light into its constituent colours. A sensor scans through the colours to reveal which are absorbed by the test sample and which pass through.

My concern was that if the sensor does not have the same responsivity in all wavelengths, then the results will not be a true reflection on the extent to which each colour is absorbed.

I understand that many DIY spectrophotometers use light dependent resistors instead photodiodes. However, I have found that LDRs also have a variable response depending upon the wavelength of light hitting the sensor.

Eric
 

erco

Senior Member
I have found that LDRs also have a variable response depending upon the wavelength of light hitting the sensor.

Eric
Yes, traditional cadmium sulphide photocells (LDRs) are most responsive to yellow-green light. Nonetheless, I see a lot of of recent projects using them to detect light from red lasers. Brute force still works in this marvelous digital age.
 

AllyCat

Senior Member
Hi,

Yes, most sensors will have a non-constant response, CdS LDRs are used because they correlate moderately well with the human eye (which also has a very non-constant response! ). Crystalline Silicon cells extend (and peak) in the Infa-Red range, but some are "blue-enhanced" to better cover the visible spectrum.

However, I believe your major problem may be with the light source. The majority of "modern" light sources (fluorescent, LED, etc.) emit "spikes" of quite narrow bandwidth (colour) which are chosen to "fool" the human eye into seeing "white". You probably need to use a suitable incandescent lamp, with an appropriate calibration curve for the exact voltage being used. But beware that like the Silicon Cells, its response will peak in the Infra Red.

Cheers, Alan.
 

erco

Senior Member
Lots of specialized light sensors to pick from. Search Google or Ebay for "color sensor".
 

eggdweather

Senior Member
So this device will output 40% of it's maximum output, regardless of scale set, when it is illuminated with blue light. So assuming the ADC used has sufficient resolution, it will be quite easy to differentiate between colours.
40% or 0.40 of Vdd = Blue
50% or 0.50 of Vdd = Green
60% or 0.60 of Vdd = Yellow
75% or 0.75 of Vdd = Red
Probably using READADC10 would give the best results and give a sufficient signal-to-noise ratio, the standard ADC would vary by ~ 6 counts between colours.
Differentiating between a mix of colours might be difficult, for example say the light source has equal parts of Blue and Green, then how would it respond?
 

fernando_g

Senior Member
There is a reason that calibrated colorimeters cost a fortune. Calibrated light sources are also quite expensive.

But in principle, you can take TI's datasheet plotted response. Bear in mind that this is a typical response, though.
 

eggdweather

Senior Member
http://www.ebay.co.uk/itm/TCS3200-Color-Sensor-Recognition-Module-Detector-Module-3-5V-for-Arduino/201450492007?_trksid=p2047675.c100009.m1982&_trkparms=aid=222007&algo=SIC.MBE&ao=1&asc=20140328180637&meid=03325fa846bc4839a65f33bfd8aa1bd9&pid=100009&rk=2&rkt=10&sd=371484061803

So these TSC3200 devices just need power, then you read (from S0, S1, S2 and S3) the frequency that is directly proportional to the RGB colour intensity with a clear filtered S output for colour cast correction. Looks useful for this sort of application and the PICAXE should be able to do this, perhaps with a 4-way data selector (e.g. 74HC153) controlled by two PICAXE pins so it can use 1-frequency measurement input.
 
Last edited:
Top