I would like to design a circuit which has outputs LOW when the voltage it is connected to is below a threshold, like 3.3V.

There are some ic's you can buy to do this, but I would like to make my own so I can tweak the detection voltage for each specific application. From looking at the datasheets for these chips, it looks like they create a reference voltage and then use a comparator to compare the input to the reference, then that drives a transistor. Does that seem correct? Sorry if this is very simple, I am a noob :)


Example IC by microchip: http://ww1.microchip.com/downloads/en/DeviceDoc/20001434K.pdf

Another example by ST: https://docs-emea.rs-online.com/webdocs/135f/0900766b8135f981.pdf


I don't need a very accurate device, the threshold voltage can vary quite a lot for my application. I found a voltage reference which is quite cheap and might be suitable, but I guess the reference voltage needs to be lower than the input voltage will ever be.


  • \$\begingroup\$ What do you need in terms of accuracy, precision, repeatability between circuits, etc? Since you are looking for a discrete design, you either require reference ICs or else you go to some significant design (even then, likely unable to beat an IC without custom tweaks and calibration and re-calibration from time to time.) If you can accept some variation one circuit to another as well as over operating temperature ranges, then discrete and cheap and tweakable thresholds with hysteresis are all possible. \$\endgroup\$
    – jonk
    Nov 18, 2018 at 3:01
  • \$\begingroup\$ @jonk I dont need very high accuracy, precision, or repeatability. For the 3.3V example, a max threshold voltage anywhere from 3.45V to 3.15V would be fine. I will definitely use a reference IC. I found a very cheap one here: lcsc.com/product-detail/Voltage-References_TL431_C181103.html \$\endgroup\$ Nov 18, 2018 at 3:17
  • \$\begingroup\$ Yeah. I have buckets of TL431's. The variability of BJTs from the same family might be \$\pm 20\:\text{mV}\$ for their \$V_\text{BE}\$ (at the same temp.) Without compensation (not hard to add some) account for another \$-1.8\text{ to }-2.4\:\frac{\text{mV}}{^\circ\text{C}}\$ for operating temperature differences. But a crafty design can do a lot to minimize temperature variation. (It will look like a parabolic curve, which you can work to "flatten out" pretty well with a good design.) Your range is probably achievable with the TL431 without too much added work. Without it? Trickier. \$\endgroup\$
    – jonk
    Nov 18, 2018 at 3:40
  • \$\begingroup\$ Oh. Two TL431's and you are more than golden. If you are willing to spend two on a simple circuit. See this TI paper for some ideas. \$\endgroup\$
    – jonk
    Nov 18, 2018 at 3:43
  • 1
    \$\begingroup\$ Or you can look here, as well: schmitt trigger. \$\endgroup\$
    – jonk
    Nov 18, 2018 at 3:50

1 Answer 1


How about this?


simulate this circuit – Schematic created using CircuitLab

Some of the resistors should not be 100 ohms.

  • 1
    \$\begingroup\$ it's much easier to use a TL431 and a few resistors. \$\endgroup\$ Nov 18, 2018 at 5:40
  • \$\begingroup\$ How well do circuits like this work without matched BJTs like the 2N3906? \$\endgroup\$
    – DKNguyen
    Sep 29, 2019 at 22:12

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.