Technically, every relay or transistor could be replaced by an ideal switch if there was such a thing.

However, I learned that those parts are satisfactory in most cases. Thus, I'm looking for a scenario where either:

  • a certain device can't be built because it would require such an ideal switch
  • the ideal switch would improve the performance (cost, efficiency, ...) of a device to such a degree that the alternative (relay/ transistor/ other switch) is no longer a meaningful option (assuming that the ideal switch costs the same as a relay/ transistor suitable for the same application)

Ideally, I'm looking for power-switching applications since I learned that RF-MEMS come close to the idea of an ideal switch. Thus HF/ RF applications are less suitable because the next best alternative (RF-MEMS) is already good.

I have already researched the following solution that could serve as an inspiration:

Electrical motor controllers use three-phase AC. They could technically be switched on by a TRIAC. However, the heat emission would be too much for most casings. On the other hand, a relay can't handle the current peaks that occur when the motor starts. Thus, they use the TRIAC to switch on the motor (to handle the current peak) and then short-circuit the TRIAC with a relay to avoid heat emissions. An ideal switch could replace the hybrid construction with one switch. (This should serve as an example - saving one component is too little impact to qualify as a solution, unfortunately. It has only a minor impact on the overall cost of the device).

Edit: An ideal switch has the following properties:

  1. if open, it is an open loop (infinite resistance)
  2. if closed, it has zero resistance (short-circuit)
  3. switching speed is (close to) infinite
  4. does not require energy to be operated (or very, very little)

It's size would be 10% of the next best alternative.

  • \$\begingroup\$ First define "ideal switch", then we can discuss it's applications. \$\endgroup\$
    – Eugene Sh.
    Jan 11, 2023 at 21:46
  • \$\begingroup\$ Ideal switches don't exist except in simulators or in theory. What non-idealities are you likely to accept? \$\endgroup\$
    – Andy aka
    Jan 11, 2023 at 21:52
  • 1
    \$\begingroup\$ Every power conversion application would benefit from an "ideal" or more ideal switch, some would benefit greatly. Let us know when you have samples available! \$\endgroup\$
    – John D
    Jan 11, 2023 at 21:56
  • 2
    \$\begingroup\$ “every relay or transistor could be replaced by a . . . switch” No, in general a transistor used in an analog circuit could not be replaced with a switch. \$\endgroup\$ Jan 11, 2023 at 22:31
  • 1
    \$\begingroup\$ The possibility of achieving a Platonic ideal of a switch here in the physical plane presupposes the existence of a magic wand, and a magician to wave it. Such a person could just wave their magic wand and make the world so ideal that switches were no longer necessary. Therefor, the question is moot. \$\endgroup\$
    – TimWescott
    Jan 11, 2023 at 22:37

2 Answers 2


A possible way to make an "ideal" switch is to enormously improve a real switch. This idea is implemented in the circuit of the so-called "precision diode rectifier" where the undesired forward voltage drop VF across a real diode switch is compensated by an equivalent voltage VF produced by the op-amp. As a result, an "ideal diode switch" with zero voltage drop across it is obtained.

See for example this recently asked question where, by the help of the op-amp A1, the real diodes D1 and D2 act as "ideal diodes".

In another answer of mine, you can see how an "ideal" transistor switch is made.


In most practical use cases you don't actually want an ideal switch.

Imagine that you have your ideal switch and some parasitic capacitance around it. What happens with the current when the switch is closed?

  • \$\begingroup\$ Not to mention, what happens in a parasitic inductance, when an ideal switch opens in no time. \$\endgroup\$
    – Justme
    Jan 12, 2023 at 0:20

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.