I have relay that I'm using to switch between two batteries, if Battery 1 is on charge then it uses that, if not Battery 2. After 15 months it's stopped working. May I ask for suggestions on how to prevent the replacement breaking too please.

Some more information. The relay has a 12v coil. It's rated at 10A. Battery 1 is on the normally open contacts, Battery 2 is on the normally closed. When Battery 1 voltage is above roughly 13v a voltage sensitive relay energises the coil and connects Battery 1 to the relay common. It's a DPDT relay, because the Battery 2 is connected to a solar charge controller that uses positive common and so both positive and 0v must be switched. The load is a 4A peltier cooler.

The relay still clicks but no connection gets made to the normally open contacts, I suspect arcing has killed the contacts. The relay has always stuck on for a few seconds (maybe just natural as the battery voltage decays post-charge), and occasionally bounces.

I wonder if a flyback diode or snubber might improve reliability?

enter image description here

  • 1
    \$\begingroup\$ A flyback diode or snubber network is only effective for inductive loads. I presume that your peltier cooler is purely resistive. What investigations have you done? Is it possible to inspect the relay contacts (even if it means partially destroying the case)? If so. try cleaning them as a temporary fix to prove the theory that they are at fault. The only contacts switching an inductive load are the VSR ones switching the main relay. How much hysteresis does the VSR have? Can it be adjusted? Can you quote the make and models of the relays? How often do the relays switch per day? \$\endgroup\$ Commented Apr 17, 2022 at 21:36
  • 1
    \$\begingroup\$ Thinking about it there's a small fan on the peltier so not entirely resistive. But there's also 3ft of figure-8 cable so quite a lot of capacitance too. \$\endgroup\$
    – Adrian
    Commented Apr 17, 2022 at 22:22
  • \$\begingroup\$ Replacement relay arrives Tuesday, so I can take the old one apart then. VSR does have threshold adjustment (not hysteresis), not sure what that would show (and hard to test)? Relay is mostly switched to solar/battery 2, very rarely on battery 1, maybe 100 times over a year. \$\endgroup\$
    – Adrian
    Commented Apr 17, 2022 at 22:29
  • \$\begingroup\$ I was just wondering how often the relay operated. If it was switching back and forth several times a day then it might have exceeded the contact life if it was a badly made one. However the answer below looks to me to be a good start to fixing your problem. You could try a snubber network across the load (fan motor) as well \$\endgroup\$ Commented Apr 18, 2022 at 10:01

2 Answers 2


There are two problems with your circuit:

  1. The lack of flyback diodes across the relay coils.

  2. The 'voltage-sensitive relay' contacts unnecessarily carrying the load current.

Arcing could have caused the 'voltage-sensitive' relay contacts to deteriorate. The load current through those contacts could have caused a drop in the 'main' relay coil voltage, leading to damage to it's contacts also.

Here's the correct schematic.

enter image description here


In small armature relays and all reed relays, NC contacts are less robust, since they are held down with less pressure on the contacts than the NO contacts. The problem you have seems to be that the NC contacts overheated and stuck together.

If the relay is from a mainstream manufacturer and has safety ratings and/or is a recognized component, then the datasheet should indicate the ratings for NO and NC contacts. They may not be the same.

If you’re using such a relay within its ratings, then you’re over stressing it on inrush currents and need to derate it. Use a 20A or 30A relay then.

When contacts are inadvertently overstressed, the NC contacts are usually at an inherent disadvantage in relays, and it’s typical for them to fail shorted. A contractor would be much better in that case, since NO and NC contacts have the same contact pressure, mechanically stabilized and highly independent of contact temperature. Small armature relays like you’re likely using rapidly lose NC contact pressure as the contacts get hot, and those contacts suffer from a potential thermal runaway then. Bad news.

If the relay is a from a “what’s cheap today” source, then don’t use NC contacts for carrying power: use a second relay, actuated when the first one is open. There needs to be a crossover isolation delay added into the circuit unless the HVSR already has a pair of alternating outputs. Inadvertently connecting the two batteries together on switchover causes inrush currents that overloads the contacts, so that’s bad thing.

Cheap relays, or relays that are worn out, also have a problem with the mismatch in the speed in which the contacts in the pair switch. It’s entirely possible then that one of the contacts has fully connected to the second battery, while the first one is still conducting to the first battery, even if just through an arc.

That’s yet another reason why using two relays’ NO contacts is advantageous: the crossover delay can be made long enough to ensure that in all conditions one of the relays has fully opened before the other one closes.

The HVSR also needs a “dead period” or “hold off delay” when it’s quiescent after each shiftover, ie. it needs a limit on the switching rate. In your application this limit would be one event per several minutes.

If the HVSR doesn’t offer this feature, or the period can’t be set long enough, you’ll need a simple PLC between the HVSR and the load-switching relays. This will make it very easy to implement the switchover delay as well as the rate limit/inhibit, and will let you use NO relay contacts to maximize the contact life. The simplest PLC you can get will do the job usually.


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.