Bloody light sensor

This is a report on an assignment for a course in process measurement technology at Novia UAS. The task is to design, simulate, construct, test and calibrate an analog illuminance sensor based on an LDR.

(Downloads are throttled, go make a cup of coffee)

Sneak peek at the schematic

(Picture of schematic)

LDR characteristic curve

It's surprisingly difficult to find any information about resistance/illuminance characteristics for photoresistors/LDRs. At best you get a straight line on a log-log graph which looks too good to be true considering that datasheets specify at which illuminance the gamma value is measured.

R = R10*(10/E)gamma

R is the resistance, R10 is the resistance at 10 lux, E is the illuminance and gamma is a "constant"

Datasheets tend to define gamma = log10(R10/R100)

Adafruit is the only source I've found with more detailed information about typical characteristics of LDRs: slightly curvy traces on a log-log plot. I meant to add this as a reference in part 2, but unfortunately I forgot about and I don't feel like making any more late ammendments.

Own "research"

The sensor is designed to translate conductance (the inverse of resistance) directly into a voltage: U = k/R

Approximate model: log10(U) = a*log10(E)2 + b*log10(E) + c

A lot more details are in the second PDF

The inverse formula is a bit more funny:

Sprechen Sie LaTeX?
        $$ E = 10^(\frac{0.789-\sqrt{0.488-0.124 log_{10}(U)}}{0.0621}) $$
        E is the illuminance in lux, U is the voltage in volts. $E \pm 9.4\%$

I like to call this "unevenly engineered". If it was over-engineered it wouldn't use a sucky LDR and have a sucky 10% accuracy.


I won't bother cluttering the reports any further with explicit permissions unless someone actually wants permission to use them. If you do please send me a message and I'll do something about it.

I also may have to do something about one of the images, I'd rather make an equivalent myself than trusting fair use if I were to open source this project.