Back in the days, when phototransistors were hard to get (especially in some parts of the world like the communist eastern block), one could brew their own by just filing/sawing off the top of a TO 18 NPN or PNP to expose its junction. The the collector-emitter pair would serve as a photosensitive device.

Since I'm just an amateur, I'd like to know whether it's possible to obtain something out of an unjunction transistor via the same treatment (e.g. the universal 2n2646 is an ideal, cheap and common candidate). The goal would be to obtain a light controlled saw-tooth/relaxation oscillator (probably an on-off modulation might be the most intuitive behavior).

Later edit: I managed to build a simple audio saw-tooth oscillator with a home-brewed unijunction phototransistor. To test it, I used both visible range and IR light from a remote control, and, as in this question's hypothesis, the transistor seemed to shift its curve slightly, the only testimony being the modulation of the audio frequency by the strength of the light shone onto the exposed junction. Anybody interested can try it out using such a circuit.

UPDATE A theoretical back-up of the observations/hypothesis of this question is provided here.

  • \$\begingroup\$ Yes I remember scratching off the black paint from red spot transistors OC71's, OC44's and the like. (Germanium transistors for the younger engineers here). Not sure it would work with a unijunction. For a transistor the light generates a base - collector current that is amplified by the transistor. As the unijunction is a PNPN structure, light falling on it may stop it functioning all together. \$\endgroup\$ Commented Jun 5, 2013 at 17:57
  • \$\begingroup\$ @JImDearden: No, he's talking about a true unijunction transistor, not a so-called "programmable unijunction" transistor. \$\endgroup\$
    – Dave Tweed
    Commented Jun 5, 2013 at 19:56
  • \$\begingroup\$ Yes indeed, a single PN junction one (it should be like a resistor-diode-resistor star structure roughly - but not quite as I understood). The other options are tetrodes like BRY39 or the 2n6027 PUT (which can't be filed to expose the junctions since it's enclosed in a TO92 casing usually) or even an SCS like it was suggested in a deleted answer here. About those germanium transistors: you usually had to put them in some kind of a solvent to get rid of the thermoconducting paste they were stuffed with :) \$\endgroup\$
    – teodron
    Commented Jun 6, 2013 at 6:55
  • \$\begingroup\$ You can answer your own question and mark it as accepted with the results of your test. \$\endgroup\$
    – Phil Frost
    Commented Jul 2, 2013 at 11:41

1 Answer 1


You can get the effect of a phototransistor by putting a photodiode on the base of a BJT, and if you don't have a photodiode, any diode in an optically transparent case will do. All PN junctions exhibit a photocurrent effect to some degree. LEDs work quite well:


simulate this circuit – Schematic created using CircuitLab

Would this work with a UJT? I suppose if you file off the case you would get some photocurrent. I have no idea if it would be enough to get you over the peak region so you can make an oscillator, at least with practical light sources. If you try it, post your results as an answer!

  • \$\begingroup\$ Thank you for the proposed alternative to a phototransistor (that should've come like a natural solution once an optically exposed PN junction was available). But what could happen to the junction present in an UJT when exposed to light? It should exhibit a different characteristic curve (perhaps a translation in the intensity axis?).. Like I said, I am not schooled in Electronics so I might fail badly in my logic :).. \$\endgroup\$
    – teodron
    Commented Jun 5, 2013 at 17:55
  • 1
    \$\begingroup\$ @teodron I haven't a clue. If I had to guess, I'd say that the photocurrent wouldn't be sufficient to get you over the peak region, so you wouldn't be able to do anything interesting without a very bright light. Try it and let us know? :) \$\endgroup\$
    – Phil Frost
    Commented Jun 5, 2013 at 17:55
  • \$\begingroup\$ thanks for the encouragement, I'll do it over the weekend and post back the results for those curious enough :) \$\endgroup\$
    – teodron
    Commented Jun 5, 2013 at 17:57
  • \$\begingroup\$ @teodron Yes - try it out and let us know. \$\endgroup\$ Commented Jun 5, 2013 at 18:00
  • \$\begingroup\$ If your intent is to have light hitting the LED cause the transistor to turn on, then you have the LED backwards. \$\endgroup\$ Commented Jul 2, 2013 at 11:02

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