Reading diodes to create a thermal imaging system

[Udo Klein] was working with some 1N4148 transistors and was interested in the specs relating to their performance at different temperatures. The forward voltage actually changes quite a bit depending on temperature and wondered if this could be reliably measured. He hacked his own LED shield for the Arduino to use as a 1×20 thermal imaging system.

The screenshot above is mapping the voltage measurements from a row of diodes (see the video after the break to get the full picture). He’s holding an ice pack over the row of diodes and observing the change. The on-screen display is facilitated by a Python script which is pulling data from the Arduino. Since there aren’t enough analog inputs to read all twenty diodes separately they have been multiplexed. Four I/O pins each enable five of the diodes, readings are taken with five analog inputs before moving on to the next set.

What can this be used for? That is precisely the wrong question… sometimes you’ve just got to go where your curiosity takes you.

35 thoughts on “Reading diodes to create a thermal imaging system

  1. Absolutely cool idea. I thought about a similar thing a while ago and the best in home-made-bolometer I could come up with is a matrix of 0402 tiny thermistors. They show even more signal change, a homogeneous absorbant surface and can be mounted closely spaced in 2 dimensions forming 10×10 sensors for example. Plus, they are cheap.
    Combined with a cheapo old thermal imaging lens off ebay it makes a full fledged getto tech FLIR system!

    1. > Combined with a cheapo old thermal imaging
      > lens off ebay it makes a full fledged getto tech
      > FLIR system!

      I’m looking for those lenses, how should I search for them?? regular “FIR lens” or “infrared lens” keywords don’t show much…

      (moderator please delete my duplicated post below, I forgot HaD supports threaded replies for quite some time..)


      1. 1N4148s are diodes, not transistors. NE555 is analog timer, not microprocessor – or did you know that and are only being snarky?

        Transistors (NPN or PNP) can make “ideal diodes” by connecting B to C. Linear temp response with constant current.

    1. Come on, everyone that should be considered to be worthy on posting on HackADay should not think that such a common part as a 1N4148 is a transistor…

  2. Diodes like 1N4148 are OK, but taking an NPN transistor like a 2N3904 and connecting B to C creates a “super diode” with properties much closer to ideal. Voltage is linear with temperature when fed a constant current – available from the C terminal of another transistor (use PNP 2N3906 for example with E connected to V through a resistor and constant voltage from low V band-gap shunt regulator between B and V – Ireg = (Vreg – Vbe)/R ). Approximately 2mv per degree C if you keep current source at constant temp. You can use another PNP as super diode to compensate – B,C connected to B of regulator transistor and resistor to 0V, E to – side of shunt regulator with side to V. Now Vbe is compensated out so Ireg = Vreg / Re.

    Each temp sensing super diode will have its own offset voltage (I know – I measured a hundred from different mfg lots) so accurate measurements will require calibration. This Voffset will be fixed, however.

    Best way to assure same current through diodes is to connect them in series, but then you need differential analog measurement. Besides, each diode takes 0.7V and that adds up, especially when V is low.

    Good luck!

    1. You are completely right. However the whole point of my blog is to work with simple parts / simple circuits. It can be argued that an Arduino does not qualify as a simple part but then again this is what I need to make it accessible to a larget audience.

    2. Lee, Your’s is the most substantive comment in this thread so-far at my post time. What about connecting many diodes or transistors (as diodes) in parallel (ideally from a same manufacturing batch)? Does that help reduce calibration requirements? Diodes like 1N4148/914 are cheap like popcorn, slightly less for common transistors. Many of us have cut-tape bands of these things. What about Schottky diodes, better or worse than the likes of 1N4148 (smaller band-gap with Schottky)? I must review the physics…

  3. Ok, n00b question here.

    Does this set up mean it can be turned into an actual Thermal Imaging Camera like the cores we see mounted in bumpers for $2000.00? Or is this a considerably closer range device like an infrared thermometer that doesn’t require physical contact?

  4. @steve
    > Combined with a cheapo old thermal imaging
    > lens off ebay it makes a full fledged getto tech
    > FLIR system!

    I’m looking for those lenses, how should I search for them?? regular “FIR lens” or “infrared lens” keywords don’t show much…


    1. The reaction to IR would be so miniscule I expect that you at best might see a small spike if held right in front of a thermal lance, or maybe with a lens looking at the sun.
      Sunspot detector maybe :)

  5. true thermal imaging works like a camera so i doubt you can use this for ghost hunting work but it may be good for a thermal gradient say for example.

    tape a grid of diodes on a large heat sink and see the distribution of heat across the heat sink

  6. I have seen 1n4148 used as cheap temp sensors in many things, bread makers come to mind, to tell the uC when it is over a certain temp.

  7. Polypropylene fresnel lenses are just a few dollars. Typical germanium lenses are ridiculously expensive on ebay.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s