I was surprised to learn that a mechanical switch (SPST) bounces when it is opened. What causes the contact to temporarily return to touching?

Does this phenomenon only happen with certain types of switches (e.g., with sliding switches but not with most pushbuttons)?


3 Answers 3


Basically no substance is perfectly smooth, or inelastic. When two objects come together with any force, there is kinetic energy that needs to die down upon contact. So you get some "ringing", plus since the surface isn't smooth the contact makes/breaks a bit randomly until settling (e.g. even with a sliding switch the contact resistance will vary upon movement)

It happens with all switches; sliding, pushbutton, relay, etc, though sliding or wetted contacts (e.g. mercury) are better. It also happens upon opening the switch, not just on closing. Some are better than others - you can get some awful ones that will bounce for tens of ms, and some good ones that settle within <5ms.

To add more on the opening of the switch, it's due to the irregular surfaces opening in a not perfectly perpendicular manner. The widely varying resistance this creates appears similar to the bounce during closing. When pressure is released, one part of the contact may disengage, but another part may briefly obtain a lower resistance than it had previously, the more irregular the surface and the more of an angular force between the contacts, the more pronounced this effect is likely to be.

Although it looks smooth to our eye, the copper surface looks rather different under a microscope:

Copper Surface

  • 1
    \$\begingroup\$ I understand why there's bouncing when two contacts come together. I'm interested in why there is bouncing when two contacts are pulled apart. \$\endgroup\$ Jan 2, 2013 at 18:33
  • \$\begingroup\$ It's more to do with the contact smoothness here rather than bounce, creating varying resistance as the contacts separate. Basically all of the contact does not separate at the same time (the direction of travel will not be perfectly perpendicular to the other contact, so you get the wiping action mentioned by Olin and David) \$\endgroup\$
    – Oli Glaser
    Jan 2, 2013 at 18:40

Think about the reverse. What if you were tasked with designing a switch that did not bounce on opening. How would you guarantee at the microscopic level that the two conductors always moved away monotonically without any wiping action (sliding, which can cause momentary opening and closing)? You have to consider that everything is a spring at that level, no matter how rigid it may seem at a macroscopic level. Even if it were somehow possible to not have any wiping action as the contacts separate, how do you guarantee there will not be any oscillation of the contacts as they move away? You have to consider the velocity starts out at 0, so even a little backlash or ringing due to breaking of the static friction can be significant.

Think about it and describe us a dry mechanical switch that won't bounce if you actually think that is possible.

  • 3
    \$\begingroup\$ Olin is right, but I'd like to add a little. By "wiping action", Olin means that the two contacts slide against each other. One contact does not simply lift off the other, but it also slides. You can see this by putting your elbows against your hips and pressing your hands together in a praying pose. Think of your hands as the contacts and your arms are springs. Now press one hand harder against the other so that your hands are not centered on your body. You're hands will slide. Any reasonable switch will always have some spring action on the contacts and thus some sliding and bouncing. \$\endgroup\$
    – user3624
    Jan 2, 2013 at 17:16
  • \$\begingroup\$ What about relays? Do their contacts disengage without wiping action? \$\endgroup\$
    – dext0rb
    Jan 2, 2013 at 17:18
  • 1
    \$\begingroup\$ @DavidKessner: So if Olin is right, why not +1? :) \$\endgroup\$
    – user8459
    Jan 2, 2013 at 17:23
  • \$\begingroup\$ All mechanical contacts will do this; including relays. \$\endgroup\$
    – gbarry
    Jan 2, 2013 at 17:53
  • 8
    \$\begingroup\$ I edited the first sentence, there is not a need to write in such a way that clearly states you think the user is at fault for not realizing the answer. If you feel a question is below you dont answer, dont insult the user or you will not be welcome to continue on this site. \$\endgroup\$
    – Kortuk
    Jan 3, 2013 at 2:50

In the 1950s and early 1960s I worked as an electrical engineer in a team designing high current (>100 amps) regulated dc power supplies. We had been using magnetic amplifiers for control, but then what we called SCRs (Silicon Controlled Rectifiers, now called Thyristors) began to appear. They were very susceptible to overloads, and would blow faster than the fuses available, so one way to protect them was a "crowbar", that is a switch that short circuited the transformer secondary winding, so that the supply was cut off from the SCRs, and fuses in the supply line would blow -- the hefty transformer and its cabling could take the shock safely. The trouble was that there was an awful lot of dangerous, and frighteningly noisy, sparking as the switch inevitably bounced. The clever engineers in the research lab did design a "no bounce" switch, but it needed very delicate adjustment so was no good for production equipment. In the end improved fuse design and more robust SCRs made it unnecessary.

As I recall from a high speed film of the switch, an upper contact descended on to a lower one, and the combination was sprung so that they stayed together as they recoiled and swung up and down. When we tested one there were indeed no sparks from the switch, but a loose connection to the transformer went off with an almighty bang. (before the t dotters and eye crossers complain, yes I know it should be "there was no spark", but what I wrote sounds more like reality)


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.