I'm currently completing my final year project and am currently stuck going round in circles as how to complete it.

To summarise: I need to design a circuit which is capable of IDENTIFYING a light source (primarily biomass flame light) from other light sources using photodiodes.

My circuit will (hopefully) be in 3 parts. Part 1 - photodiode sensor circuit; part 2 - microchip to analyse part 1 and prepare results; part 3 - output results on display. Part 1 is causing me my problems as no one has done anything like this before from my research into this.

I have made a very simple Transimpedance Amplifier using a photodiode in photovoltaic mode feeding into an Op-Amp - basically a current-to-voltage converter. Howvever through exhaustive testing I have found that the circuit responds to any and all light and does not allow me to distinguish as to what the light source is therefore I stopped with this circuit.

My next port of call was to use the chip "TCS3200" as a standalone chip which has filters in front of the photodiodes. Connecting S0/S1/OE to 5V and monitor Pin6, I would measure light levels using the filters (Red, Green, Blue and Clear/No filter) to allow me to make an educated guess as to what the light source is based on the readings. I have yet to do this but will update this as soon as I have.

However my question is - is there ANY way that this could be done using photodiodes? Being able to not only detect light but more importantly to IDENTIFY it as well. Is there a way to do this? Have I overlooked something or am I missing something? Any help would be appreciated. Thanks.

  • \$\begingroup\$ If color the only identifier you're working with or could you use other things like 'flickeryness' or whatever ... ? \$\endgroup\$
    – brhans
    Jan 14, 2016 at 3:17
  • \$\begingroup\$ Well I'm going with colour for now as that seems to be the best way to go - the human eye uses colour filters so I thought why not my circuit. I have no idea what you're on about regarding flickerness - please would you mind explaining it a bit more? \$\endgroup\$
    – U5M4N8R
    Jan 14, 2016 at 3:20
  • \$\begingroup\$ Have you tried using a prism to separate the light for you prior to analyzing it? \$\endgroup\$
    – Dave
    Jan 14, 2016 at 3:55
  • \$\begingroup\$ Yes, but I've decided not to for the following reason - I don't have much time left (got to be done in the next 3 months). Trying to get the angle right so that the light goes straight into the prism, and then the split light would need photodiodes in place exactly where the light could hit them is too much of a mechanical issue for me at this late stage - it'd just become another problem to add to the list. \$\endgroup\$
    – U5M4N8R
    Jan 14, 2016 at 3:59
  • \$\begingroup\$ TCS3200 seems like a good choice for measuring wavelength and intensity of incident light. Note the -OE pin is active-low, so connect it to GND to enable normal operation. Datasheet \$\endgroup\$
    – MarkU
    Jan 14, 2016 at 4:20

1 Answer 1


I'm afraid the simple answer (photodiode(s) + a few filters) is no. Identifying flames requires a spectrometer, and even then not all flames can be identified. Google "flame spectroscopy". Some flames are characterized by narrow emission lines, but detecting these will require much finer wavelength discrimination than a few cheap filters will provide. The instrument you are looking for is a spectrometer.

If you can procure (borrow) one of these, you'll need to examine the particular biomass flames you're interested in to see if they have any distinguishing features, but be forewarned - some flames just look like blackbody emitters. On the other hand, neither sunlight not fluorescent tubes look exactly like blackbodies, so there is some hope for you.

  • \$\begingroup\$ Your 100% right on spectroscopy. If OP understands what spectrum he is looking for, it is possible he could use biased colored LEDs with additional selective light filters in front of them to absorb specific spectrum with something like ±20nm 3dB specificity. (Poor mans spectrometer) since he does not care about absolute signal accuracy. \$\endgroup\$
    – MadHatter
    Jan 4, 2020 at 20:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.