1
\$\begingroup\$

I'm wondering from the perspective of the durability of a project, how can it be detected that a DPDT switch does actually switch.

For example, if I use two switches to switch between two power sources having different voltages, how can I know that none of the two switches is broken and it did not switch? If for example one of the switches (the positive voltage) did switch, but the other did not (the ground), then that will mean that the project is supplied with the positive voltage from one of the sources and the ground from the other.

The timing is ignored. Meaning, lets imagine that both switches switch instantly, if they are not broken.

The same example can be given and with MOSFETs I guess, and any other IC that have two internal switches regardless of what kind, so it is a general question, of how to detect a faulty component in the described example and prevent damages.

The only thing that comes to my mind is that I have to measure the expected change, but what if one of the power sources is not connected at all, there will be no current to measure.

If the switch was transparent, it could be visually detected I guess (that the switching actually took place), but not any switch is transparent and have components that are visually measurable.

\$\endgroup\$
13
  • \$\begingroup\$ There are devices known as safety relays that use mechanical interlocks to make sure that either all contacts switch, or no contacts switch. \$\endgroup\$
    – Hearth
    Oct 2, 2022 at 15:28
  • \$\begingroup\$ Thank you, I see, they look quite large, is there something smaller that could be used? Or maybe some circuit that can be made? \$\endgroup\$
    – verbessern
    Oct 2, 2022 at 15:39
  • 3
    \$\begingroup\$ Right, but if something goes wrong, is the danger that the load will be destroyed, or that the user will be injured? That changes the safety margins you want to use. \$\endgroup\$
    – Hearth
    Oct 2, 2022 at 15:48
  • 1
    \$\begingroup\$ Consider the use of "established reliablity" relays, where the reliability has been extensively tested. Also, if it's just equipment being damaged, ask how reliable are these relays in actual fact and what is expensive in this actual circumstance (downtime, parts replacement, what?) Then your problem becomes tractable. This article has some very interesting graphs for relay failure rates and causes \$\endgroup\$
    – jonathanjo
    Oct 2, 2022 at 15:53
  • 1
    \$\begingroup\$ A rather good MSc these on relay reliability: Ballasteros Hernando, 2017; although it's focussed on automotive, most of the general principles will apply to any method of switching. \$\endgroup\$
    – jonathanjo
    Oct 2, 2022 at 16:13

3 Answers 3

2
\$\begingroup\$

How can it be certain that a switch is actually switching?

You can't know that, unless you monitor the switch with some external circuit.

There exist specialized components such as safety relays, which use forcibly guided contacts. Meaning that if one of the contacts weld/get stuck, the other will not move either. This allows us to use one pole for supervision. If you don't supervise it, then there's still no telling, forcibly guided or not.

Safety functions are correctly designed if the designer assumes that errors will always happen and detect them when they do, then handle them in a safe manner.

They are incorrectly designed when some quack only stares at Minimum Time To Failure (MTTF) and make an assumption "this is oh-so reliable, it will last for 30 years (and then catastrophically fail)". Because in the real world, such assumptions only last until the first oxidation, the first cold solder joint, the first power failure, the first manufacturing error etc etc. You simply can't design safety-related systems based on MTTF alone.

However, safety is also just one big probability game - it's all about reducing probabilities for failure. Safety functions are therefore often designed with 2 supervised safety relays ("redundancy") with the signals placed in series. It's very unlikely that 2 relays fail at the same time, and in the dangerous position at that.


they look quite large

What does? Safety relays come in all manner of shapes and sizes. Depending on how many poles you need, NO/NC, currents, clearances (if applicable) and so on. There's plenty of PCB mounted ones.

\$\endgroup\$
1
\$\begingroup\$

Minimizing risk of fire and other catastrophic consequences is best accomplished by using proper fusing. Cross-connection of grounds should not cause a dangerous condition unless there is significant difference in potential and sufficient current capability.

If there is cause for concern, you might use a 4PDT switch in series such that a failure of one pole will not cause a problem.

\$\endgroup\$
4
  • \$\begingroup\$ I have bought exactly 2 of such for this task, but was hoping that there can be some other "more digital" solution. \$\endgroup\$
    – verbessern
    Oct 3, 2022 at 12:40
  • \$\begingroup\$ Regardless of that the problem still stays: what if one of the connections in this 4PDP switch breaks and stalls in place? There is the problem, one might get positive from one power supply and ground from the other. \$\endgroup\$
    – verbessern
    Oct 4, 2022 at 11:29
  • \$\begingroup\$ This answer assumes some very specific system. A low current control system won't benefit the slightest from fuses. \$\endgroup\$
    – Lundin
    Oct 4, 2022 at 13:10
  • \$\begingroup\$ To be more clear, the poles of the 4PDT switch would be wired in series, so if any one contact welds, the other will still open, and the fault will not occur. If a contact remains open, there is still no fault. As for fuses, they should be sufficient to prevent fire due to overloaded wires and components. Low current systems don't have enough energy to pose a danger. And the two circuits should be designed to self-protect against faults. \$\endgroup\$
    – PStechPaul
    Oct 5, 2022 at 4:05
1
\$\begingroup\$

A double-make / double-break contactor is a must.

enter image description here

Edit # 1

Or else, a DPDT knife switch may be used for manual change-over.

enter image description here

Image credit: Carolina.com

Edit # 2

Dual redundancy has been incorporated to satisfy your requirement.

enter image description here

\$\endgroup\$
4
  • \$\begingroup\$ That is the thing, what happens of one of the switches breaks internally? One might get positive from one power supply and ground from the other. \$\endgroup\$
    – verbessern
    Oct 4, 2022 at 11:23
  • \$\begingroup\$ My answer has been edited. \$\endgroup\$
    – vu2nan
    Oct 4, 2022 at 12:55
  • \$\begingroup\$ This solution requires the user to "see" that by flipping the switch one of sides connected to the handle broke and remained connected, during the switching. The protection here is that the user has to observe. I hoped that the question implies that the user does not have to care for the protection, but some kind of circuit cares for this. I apologize if it was not clear. \$\endgroup\$
    – verbessern
    Oct 4, 2022 at 13:27
  • 1
    \$\begingroup\$ Got it. My answer has been further edited. \$\endgroup\$
    – vu2nan
    Oct 4, 2022 at 18:42

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.