MOVs help on AC relay coils?
Phil Pilgrim (PhiPi)
Posts: 23,514
Someone came to me yesterday with a problem relating to magnetic reed-switch contacts failing short. The float-coupled NC reed switch is connected to a 24VAC relay coil. It's used in an alarm system at a local aquarium for making sure their saltwater exhibit tanks do not overflow during off-hours. The question was whether an MOV across the relay could would help to prevent the reed-switch contacts from welding shut due to arcing. My gut reaction is yes, but I'm not an AC guru. I also wonder if the failing short could be due to phenomena other than arcing since, presumably, the switch does not open all that often (i.e. only when a tank overflows or when it's being mechanically tested).
Any ideas?
Thanks,
-Phil
Any ideas?
Thanks,
-Phil
Comments
In that case, it may be sticking.
Contact welding is usually inrush current related - how big is this 24V AC relay & how long is the cable ?
A MOV of bidirectional TVS would reduce open-time arcing and that can only help life cycles.
I used to use mechanical relays with a couple of drying ovens and I had several fail shorted. These relays were being abused by being triggered every 10 seconds (PWM at 0.1Hz). I switched to a solid state relay and haven't had a problem since (I probably made the change three years ago). I kind of miss the clicking but my wife loves the silent operation.
I'm much less an AC guru than any of you guys but I thought I'd pass along my experience.
The doctrine on snubbers on AC and DC relay connections has been well established for a long time as relays were very much a key element in electrical control before solid-state took over.
The short story is this...
A. One should always include a snubber circuit to extended the contact points life. (This seems to have been entirely overlooked by our electronic tutorial world.) We tend to only focus on a coil snubber circuit to reduce noise that might upset the microcontroller.
B. Solutions are specific and different for either AC or DC contacts.
C. The type and size of the load may indicate a need to use large device. For instance, inductive motor use recommendations tend to derate the relay capacity to 40%. (That means a 10amp relay will only survive well with a 4 amp or less motor).
D. Some relays have better contacts than others. People pay extra for longer durablity.
++++++++++
In particular, I don't think MOVs alone are a good solution.
From what I understand is they are really one-shot protection devices that self-destruct after that one incident unless another circuit protects them from abuse. What you want is something that will quickly and repeatedly knock down transient voltage spikes that errode the contacts.
Aligent Application Note 1399 is a good introduction to how to resolve these issues.
cp.literature.agilent.com/litweb/pdf/5988-6917EN.pdf
IDEC has some useful info as well. See attached PDF.
Try Googling 'relay contact protection' for many excellent white papers written for industrial relay users.
=========
My personal conclusion is that any relay solution is not complete without investigating and adding a snubber circuit to both sides (coil - for noise suppression, contacts - for arce suppression).
There is the possiblity that some back-to-back diodes will work well.
And a lot of people will recommend a solid-state relay just because they never mastered fitting a snubber to the contact side of a mechanical relay. It just easier to buy a product that happens to internally have a snubber built-in.
But it does seem that they don't extend contact life by very much. Other devices and strategies are necessary. Some diode and capacitor snubbers claim 5 to 10 time the life as without.
And of course, the quality of the relay is central to long life. It may be necessary to upgrade to a more durible relay.
Adding a zero crossing optoisolator between the reed relay and the relay coil would eliminate the problem if it was from welding the contacts.
Using a hall effect or opto interrupter and zero crossing optoisolator would take care of both problems.
However, as many have already indicated, going solid state is the long term answer irrespective of some oscillation. Basically, the reed is not rated for inductive switching, nor even more than a few milliamps really. Either, A, control a SSR with the reed or, B, change the reed to a hall based sensor. Option B will be much easier to fill by also changing to a DC supply.
PS: Most industrial designs these days feed such signals directly to a PLC input. This is always solid state so things like reed switches work just fine as a magnetic sensor attachment. Stand-alone relays are pretty rare now. Mostly relays are either part of the PLC or contactors are used for real power delivery.
Some random thoughts from a non-spurt. Seems to me the switch may not be up to switching the current through the 24VAC coil.
As the contacts of a switch move apart an arc is created that can melt a small surface area and maintain conduction. A back EMF from the 24V coil at that time could increase the energy into the contacts and sustain more arcing. Quickly reversing the switch position, closing the contacts, may cause the contacts to weld. I've looked at welded contacts in NC signal relays under a microscope and seen contact pitting and metal buildup on the opposite contact. Our solution to this ongoing problem was to go to a SS relay drop in replacement with the same form, fit and function. We got lucky.
I'm wondering if a 1uf ceramic cap across the NC reed switch would help. It might tend to keep the potential across the contacts low when they first open.
As to how often it cycles, it might be lots if there's no physical hysteresis when it does switch. A tiny wave action may switch it many times per second.
I am reluctant to use a reed switch on anything. The contacts are notorious for bounce. And I see hobbist grabbing them up in salvage sale, but not getting much use out of them.
I prefer to 12VDC and 24VDC automotive relays that are designed for extremes of weather and temperature, have an excellent supply chain, and can be pig-tail wired into many installations.
I realize the specifical here is for AC contacts, but the majoriity of relays I look at are dual rated 24VDC at so many amps or 120/240VAC at the same. Apparently, the big advantage of AC is the metal wear travels in both directions on the contacts, so your pitting in one direct is repaired by pitting in the opposite.
DC snubbers are certainly simpler. A 1N2007 diode can at 5-10 times contact life in a DC contact. It seems that AC contacts require more exotic use of capacitor and resistor networks.
This explains why they're rated at "off" voltages. For example use an MOV across coils and contacts that operate on 120VAC, etc. They work well when used as intended.
I'd never bothered to look it up before ... Wikipedia is saying the metal oxide powder forms thousands of diodic junctions. That helps explain the unclear nature of the clamping voltage. As compared with an actual manufactured zenor where the breakdown voltage is precise. High current (TVS) manufactured zenors, both AC and DC, exist separately. I often use the term tranzorb, after one of the brands, be clear I'm not referring to a MOV.
For the record I've never used an MOV for any type of voltage regulation nor have I ever seen them used in that way. I've always used them as snubbers across the coils of contacts switching AC or inductive loads like relay coils, selnoids, etc. Since I've always considered them snubbers the exact voltage was never important to me. When putting one across a 120VAC contactor does it really matter if the exact voltage it regulates to is 128vac or 135vac?
A snubber can actual serve one of several purposes and it seems that the MOV is more useful for RFI noise suppression. What the original poster is concerned with is extending the useful life of the relay contacts.
So be careful in researching snubbers... there are a variety of solutions for different purposes clumped together.
Tranzorbs are heavy duty zenor diodes. Please don't go calling them a MOV.
I don't think that a MOV or snubber circuit will completely solve this one. You have mentioned that the reed switches are normally closed unless an overflow event occurs. So what that means is that the only transient events that will occur are at power up of the system when the coil is energized through the series of reed switches, at power off when the field collapses in the relay, and in the event of overflow when it happens and if it corrects itself.
What this sounds like is the current through the reed switches may be to high at 24V and the contact points of the reed switches are heating, becoming resistive, heating more, and then sticking. What is the relay pull-in current? Does this exceed the current rating of the reed switch? The better solution (to me anyway) would be to replace the 24V relay with a rectifier, filter, and a transistor driver for a DC relay so that only a minimum current flows in the reed switch chain. And an MOV or RC snubber across the relay coil to protect the transistor. Perhaps a diode from C-E as a backswing clipper to protect the transistor.
Solid-state relays have gotten smaller, better, and cooler. But mechanical relays still seem able to swing more raw power when needed.
With DC contacts, contact life extension is very quick and cheap -- just a properly installed power rectifier diode across the contacts. I really have yet to try contact life extension with AC contacts, but I do think it can be relatively cheap as well, just more thought about the right capactitor and resistor sizes.
+++++++
People seem to have a lot of confused ideas about MOVs and a host of other devices that are realitively new technologies. I suspect sales material has tried to create a demand for such even though they may not be necessary.
MOVs and Transorbs for lightning protect is a whole different subject as often it is impossilbe to design anything that won't be sacraficed in direct hit. The good designs allow those parts to burn up and the be easily replaced. This isn't that kind of situation.
-Phil
The up-grade path to an intermediate SSR versus an intermediate mechanical relay solution is just a matter of personal preference and experience. Often we never hear back whether the solution actually works or fails.
A long time ago when I was in my last year or so working as an EE in the field of industrial electronics there were hybrid SSRs. A couple of advantages of the SSR is that there are no contacts to burn out and if triggered properly it was trivial to implement "zero-cross" triggering which meant that the trigger occurred when the AC voltage is at zero volts. That puts less electrical noise on the line and reduces the phase shift between voltage and current which helps with the electric bill! A contactor is capable of neither! The problem with an SSR is that triacs even when in the "on" state drop voltage and dissipate power while "on". Not a problem when switching an amp or two but when switching 20-30-100+ amps it adds up to a lot of power and heat. Another problem with contactors is the arc when the contacts open and close. What they were starting to do back in the day was a hybrid SSR. This is a "standard" SSR with zero-cross triggering with two output pins for a contactor coil. The idea is that when turning on the ssr would make initially turn on and a few milliseconds later would supply power to the contactor. the output of the ssr and contactor were wired in parallel and when the contactor closed the primary current flows through the contactor. When turning off the power to the contactor coil was dropped and a few milliseconds later the ssr broke the connection. A great system and I used a number of them.
The problem with putting a diode across the contacts of a contactor switching DC is that when passing a fair amount of current an arc will start as the contacts are opened. With enough voltage and current even when that contacts are completely seperated the arc between the contacts will still pass a significant amount of current. That's where the 'snubber' comes into play. They're purpose is to "snub" the arc.
Again these are memories from ~20 years ago and I would be surprised if there haven't been significant changes in technology (hybrid SSRs as they were called them are an example) I would also be surprised if I haven't butchered the language of the technology a bit.
As far as I can tell, the general industrial approach is just use a heavier contactor, they can hack a huge amount of abuse. It probably depends on just how big the inrush current being dealt with is. However, pure solid state is also very common now, ie: DSP type handling has become cheap. Variable Speed Drives are everywhere now. Older motor installs are being converted to VSDs. Zero-crossing for MOSFETs and IGBTs is a piece of cake in the form of a "soft-starter".
The nice thing with these solutions is the sparky doesn't need to deal with confusing supply wiring. It's just mains in and load out.
EDIT: Now I think about it, the modern soft-starters may be exactly what you are talking about.
Typo in the above... sorry to be so slow to get this corrected. I have been down with a summer cold.
It should have read....
Adding MOVs or other snubber schemes CANNOT overcome that.
++++++++
I was trying to make the point that one First has to select an adequate switch, or it gets more expensive to work around the problem. But the truth is that I didn't fully understand the original propblem. And these days, we have a lot of tiny switches and push buttons used to trigger swinging huge loads successfully...if the design is not expecting switches to handle excessive currents.
I generally avoid reed switches completely. They take a lot of specific knowhow to avoid problems as they bounce a lot, and are fragile. In 'the olden days', a float switch would use a mercury switch, but they have been outlawed.
With printed circuits, we now have a lot of switches that will only handle a fraction of an amp and they won't work well without an immediate boost off some sort - either an SSR or an opto-isolator.
In this case, we have a magnetic reed switch at one end that has failing contacts and the coil from a 'contactor' being controlled.
Both these devices might be generically considered relays, and now we add a third solid-state relay inbetween to fix it all. It works, it is a good fix. But explaining the 'all and everything' behind this fix to a new user is a bit of a challenge.
The solution simply isolates the existing reed switch from the heavy inductive load of the contactor. And it likey means that all the other stuff - the physical location and configuration of the reed switch, the float, and the the contactor don't have to require a big rebuild. One just pops a tiny SSR in line, replaces the defective reed switch, and life is good.
+++++++
I am still a hold out for mechanical relays and putting the right mechanical relay in the same place as the SSR with the right snubbing for contact arc could be just as good. But that is all a bit pedantic. MOV insertion is a distraction.