I am wanting to create a system that supplies approximately 24V and 1A to a load (essentially a resistor) for up to 1 second. The process will not be repeated for any one load and there would be no continuing current flow after about 1 second. The positive wire will be switched by a momentary SPST switch but I also want to use an arduino output to have the option of solid state control (Arduino) or manual control (SPST switch). I wish to use 16 SPST switches to control 16 separate loads but I will also need to use switched grounds as I want to use an 8-core conductor from the circuits to the loads to control all 16 of them independently (i.e. the loads will be "matrixed").

My question is can I use NPN transistors as in my below diagram to allow switching of the POS and GND to a particular load via the use of a single SPST switch (i.e. mimicking a DPST switch through the use of transistors)?

Circuit Diagram

If so, I will be needing a total of 192 transistors (aiming for 6 "banks" of 16 loads) so which transistor would you guess would be the cheapest and could handle this sort of current for the brief period intended (1 second)? I was thinking TIP120 / TIP 125.

Lastly how would I calculate the correct resistance of the resistor so that only enough current flowed from the SPST switch to the base of the 2nd NPN transistor to activate it while maintaining maximum current to the load.

Thanks for your help Aaron. So you suggest this arrangement?

Redo circuit http://imageshack.com/a/img674/7408/7QRCHi.jpg

So the resistor between the SPST switch / drain of the P-FET to the gate of the load-negative N-FET isn't needed? (as in no current would flow through the gate as it would all go from the drain to source of that N-FET)

  • \$\begingroup\$ Can you elaborate on the need for the switched grounds? I'm a little confused by that part. And what is the collector of the bottom transistor connected to? \$\endgroup\$
    – Dan Laks
    Dec 9, 2014 at 21:11
  • \$\begingroup\$ Basically I need switched grounds as the outputs / loads will be matrixed so as to minimise the number of conductors needed in the cable that will connect the switch panel (picture above) to the outputs / loads (which will be at a distant enclosure). I'm trying to get 16 independent outputs but only using 8 conductors in the cable. i.e. Load 1 = Pos 1 & Gnd 1; Load 5 = Pos 1 & Gnd 2; Load 16 = Pos 4 Gnd 4 etc. However DPST momentary switches are too expensive and don't allow the possibility of Arduino control, hence the need for SPST switches and transistors instead. \$\endgroup\$
    – joeka101
    Dec 9, 2014 at 23:55
  • \$\begingroup\$ The collector of the bottom transistor is connected to the negative side of the load / output (i.e. it is the ground switch) \$\endgroup\$
    – joeka101
    Dec 9, 2014 at 23:58
  • \$\begingroup\$ Yeah, like that! :-) Keep an eye on Rds_on, Vds_max, Vgs_th, Vgs_max, and max power for each FET, and you should be fine. They will likely be different for all three. \$\endgroup\$
    – AaronD
    Dec 10, 2014 at 21:40
  • \$\begingroup\$ As for the gate resistor to the load N-FET, people usually include them to limit the inrush current because of some parasitic capacitance between the gate and the channel. Typically more to protect the thing that drives it than for the FET itself. And you might need a pulldown on that FET also because the same gate capacitance might keep it high with no other influence. \$\endgroup\$
    – AaronD
    Dec 10, 2014 at 21:44

1 Answer 1


The arduino interface won't work as-is, but the rest looks good.

When the switch is on, the arduino's transistor will probably not like the emitter and collector both connected to +24v while the base is at +5v or whatever the arduino gives you. That's -19v, which is probably outside of the spec.

When the switch is off, the arduino will only be able to supply about +4.3v to the load (5v - Vbe) because the transistor is then operating as an emitter-follower.

The concept should work, but you'll need a slightly more complex interface from the arduino. Probably use either a PNP or a P-channel FET across the switch with a pullup resistor to +24v, then pull that down with either an NPN or an N-channel FET that is driven by the arduino.

As for your other questions, go to mouser.com or digikey.com or any other electronics warehouse and plug your requirements into their parametric search engine, one parameter at a time, most important first, and select as many options within each parameter as will technically work for you. When you run out of requirements, sort by price at your required quantity or round up to the next common price break.

Sometimes, the specs that you care about are not listed in the parametric search. In that case, specify what's there, sort by price, and start reading datasheets. For example, the base resistor for your ground connection depends on the maximum base current and the transistor's gain. You want enough base current to saturate the transistor (make more current available to the load than it will use), but don't exceed the maximum.

  • \$\begingroup\$ I should have said that the SPST switch will never be used at the same time as the Arduino output. Maybe a diode at the emitter side of the "arduino transistor" to prevent 24V supply to C & E simultaneously being a problem. If they were never used simultaneously would an arduino be able to saturate the "arduino transistor" and so supply the full 24V to the load? or would pull up & pull down as you say be better (my concern is that the above circuit will be repeated 96 times and I'm trying to minimise components / costs). Many thanks. \$\endgroup\$
    – joeka101
    Dec 10, 2014 at 13:44
  • \$\begingroup\$ @joeka101: It doesn't matter if they're used at the same time or not. My analysis of the emitter-follower was with the switch off anyway. That topology simply cannot supply more volts than it receives. It's only a current amplifier. I know you want to keep the parts to an absolute minimum, but I think you'll have to double-buffer this one like I described in my answer. And then it won't matter if they're used at the same time or not; it'll be just like two switches in parallel. \$\endgroup\$
    – AaronD
    Dec 10, 2014 at 15:32
  • \$\begingroup\$ Actually, I would seriously consider using FET's all around instead of BJT's. They don't require any current to operate, and they behave as low-value resistors when they're saturated. A BJT of either polarity will reserve some voltage for itself even when it's saturated, which feeds into the P=V*I equation for power dissipation. Use an N-channel for the load negative, a P-channel for the load positive with a pullup resistor to +24v, and another N-channel to pull that down as controlled by the arduino... \$\endgroup\$
    – AaronD
    Dec 10, 2014 at 15:38
  • \$\begingroup\$ ... The bare minimum to make that work in total is two N-channels (one for power and one for signal), one P-channel, and one pullup resistor for the P-channel. The power N-channel doesn't technically need one because there's no current flow into the control terminal (gate), but it wouldn't hurt to put one there anyway. \$\endgroup\$
    – AaronD
    Dec 10, 2014 at 15:40

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.