Bimetallic switches reliability

I am looking for a thermostat (e.g.:

formatting link
icreset.html) for basic on-off temperature control. I noticed that specifi ed service life never exceeds 100k cycles. I wonder whether it is due to co ntacts burnout or bimetallic strip fatigue. Does anybody have any insight? Thanks. Michael

Reply to
Michael
Loading thread data ...

aticreset.html) for basic on-off temperature control. I noticed that speci fied service life never exceeds 100k cycles. I wonder whether it is due to contacts burnout or bimetallic strip fatigue. Does anybody have any insight ?

Furnaces are best switched infrequently, are rather inefficient if you only operate them for a few minutes. So, a thermostat with one cycle per hour will take 1

1 years (or if heating season is only three months per year, 45 years) to reach that l ife.

It may be that no one cares to wait for longer testing.

Reply to
whit3rd

On Mon, 10 Aug 2015 14:26:24 -0700 (PDT), Michael Gave us:

Usually based on contact degradation. Bimetal works forever unless it rusts away.

Reply to
DecadentLinuxUserNumeroUno

On Mon, 10 Aug 2015 16:07:12 -0700 (PDT), whit3rd Gave us:

It is based on already known or expected service life of the contacts. No need to perform some lifelong test, which would not provide accurate MTBF numbers anyway.

Reply to
DecadentLinuxUserNumeroUno

Thank you for not bothering to specify what you are controlling with the thermostat. That makes answering your question much easier since I can more easily produce vague generalities instead of something you might find useful.

The snap disc thermostats in your URL are not designed for a large number of cycles. They are over-temperature thermostats, designed to cycle when something goes awry, gets too hot, and requires a shut down before the heat producer burns down the building. Hopefully, such events are rare.

Also note that the loads are AC and resistive only. That means you're not going to successfully switch a relay or DC. I found that out the hard way when I designed a similar thermostat into an RF power amplifier and found that the thermostat would last about 10 cycles before welding the contacts together. Of course, nobody noticed until the PA almost melted. Oops(tm).

If your unspecified application really needs more than 100K cycles, then I suggest you look at what the typical HVAC thermostat is doing for temperature control using no moving parts: and avoid the mechanical flavor.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

maticreset.html) for basic on-off temperature control. I noticed that spec ified service life never exceeds 100k cycles. I wonder whether it is due to contacts burnout or bimetallic strip fatigue. Does anybody have any insigh t?

ng ...

I have some time to elaborate. The goal is to keep temperature in an enclosure within "narrow" range. 20 a nd 30 degC is considered barely acceptable. The power dissipated inside is. .. er... probably between 40 and 60W, the box is used outdoors. The box siz e - ?? My guess that surface area is ~1.5m^2. Outside temperature is -10...

+50degC. I wish I knew whether there will be any wind or sun load. It is acceptable to thermally insulate the box, I do not think I can ask fo r more than, say, 10mm. I have ~110V and =24V as a power source. Can anybody recommend turnkey solution vendor? I would be the best. I conta cted ... forgot the name ... few days ago, haven't heard from them yet.

The plan is to have one (demo) system running in one month ("Nothing gets b uilt on time or within budget"). NO TIME for anything elegant, hence seemin gly dumb solution (thermal switch for heating, another one for cooling). Ov erlapping should be OK, I think. Heating will be needed more often, but it easier (7A,=24V), requires a relay. If thermal switch contacts define longevity, I can use them. If it is bimet al disk/strip, I need to keep looking. Thank you Michael

Reply to
Michael

I'm waiting for Windoze to update, so I also have some time.

Sounds like a chicken egg incubator.

You'll need a refrigerator. If the outside temp is 50C, you won't be able to lower it to 20 - 30C with a heater.

Any chance you can supply dimensions (LWH) instead of surface area? With those, one can calculate volume and surface area. Going the other way doesn't work.

All that can be calculated if could also produce the mass of the contents and the box. Also what's the nature of the box or contents: Solar head load and wind cooling will complicate matters, but it's still possible. Basically, what you're trying to calculate is whether

40 to 60 watts is sufficient to stabilize the temperature. I've already demonstrated that you'll need a refrigerator when the box gets hot, so the short answer is no, it won't work.
10 mm of what? Vacuum? Fiberglass? Foam board? 10 mm is not very much in the way of insulation.

At how many amps or watts? Electric heating is 100% efficient so your sources should be able to produce 40 to 60 watts.

Chicken egg incubator. Plenty of kits and instructions on the web.

Alien contact is highly over rated.

I never make plans. That way, I'm never disappointed. Unfortunately, other people tend to make plans for my time, so that doesn't really work.

The box seems small enough that CO2 cooling should work. I used to work with various environmental test chambers for temperature testing radios. -40C to +75C was typical. It think they were made by Tenny. CO2 was a cheap way to avoid refrigeration. This looks close:

Controlling the temperature is one of the easiest parts of your project. There are off the shelf controllers which will probably do a better job than a bimetalic thermostat, which incidentally has lots of hysteresis. How does $3.00 sound?

Incidentally, if you decide to buy an environmental oven, be sure that it fits a pizza platter.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

You wrote: > Sounds like a chicken egg incubator. Same idea, different contents. You wrote: You'll need a refrigerator. If the outside temp is 50C, you won't be able to lower it to 20 - 30C with a heater. Very insightful You wrote: > 10 mm of what? Vacuum? Fiberglass? Foam board? Probably some kind of foam. Ideally need ~1.5-2" of foam for the internal power to be enough to keep the box warm in winter, much more if it is windy. The box needs to be reasonably sized - proper insulation is out of question. You wrote: I never make plans. That way, I'm never disappointed. I am with you 100%. You also said something about other people making plans for you... c'est la vie

$3 controller from ebay sounds very tempting. I will keep it in mind when/if I decide to build incubator at home. For this application something with manufacturer name/part# and datasheet is required ... I need to do some "googling".

Reply to
Michael

Your local Home Depot might have a bimetallic thermostat as was once used in every household. Have you tried that?

Reply to
John S

aticreset.html) for basic on-off temperature control. I noticed that speci fied service life never exceeds 100k cycles. I wonder whether it is due to contacts burnout or bimetallic strip fatigue. Does anybody have any insight ?

I've got a few thermostat's with a Mercury switch attached to a bimetalic c oil. Maybe 20 years old... I think they should be good for another 20 years?

George H.

Reply to
George Herold

On Tue, 11 Aug 2015 07:01:36 -0700 (PDT), George Herold Gave us:

snip

More proof that the MTBF determining factor is the contacts.

Reply to
DecadentLinuxUserNumeroUno

Hysteresis is almost eliminated by compensation. Uncompensated bimetal stats are a curse.

You can increase mtbf with a snubber, and increase it again with a series parallel set of 4 stats, but how far those will get you I don't know. Its also increased by keeping switching current low and avoiding inductance.

NT

Reply to
tabbypurr

How does one compensate for mechanical hysteresis? To the best of my limited understanding, this thermostat requires the snap action of the bimetallic strip, which needs intentional hysteresis in order to obtain the snap action. No hysteresis -> No snap action. Where the contacts to slowly drift together or apart, the result would be arcing, a noisy connection, and generally lousy reliability. The best my imagination can conjure for a switch where the snap action is eliminated, is a bimetallic strip moving a variable resistor, which eventually trips a comparator (presumably without the benefits of hystersis. At that level of complexity, one might as well use a thermistor or thermocouple and do it all electronically and eliminate the bimetallic question. (Apologies to Sherlock Holmes).

Do you have an example of a compensated thermostat without any or very little hysteresis?

With a snap action thermostat, one can reduce turn on and turn off times sufficiently that any arcing is minimized. Much depends on the contact material, if there's a DC component, current level, whether it's switched under load, and environmental issues. In other words, you can't optimize the life of relay, switch, or thermostat contacts unless you know the operating conditions. RC snubbers can help, but only if they're tuned to the specific operating conditions. Done wrong, they make electrical contact metal migration worse.

Incidentally, bimetallic switches also tend to drift as the two metal strips tend to delaminate after many thousands of operations. It's part of "If it moves, it breaks" which means that if you want it to be reliable, don't design in any moving parts.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

On Thu, 13 Aug 2015 18:23:40 -0700, Jeff Liebermann Gave us:

An encapsulated mercury switch on the end of a bimetal coil reaches a tipping point and the contacts close or open in a vacuum as the mercury ball falls to one end or the other of the tube.

There are also steel ball switches which impinge on hard contact reeds in a similar manner within a vacuum 'tube' assembly. They rely on gravity and balance.

Reply to
DecadentLinuxUserNumeroUno

Such mercury thermostats have a huge hysteresis. I was discussing what happens if you take the hysteresis out of the thermostat. Therefore, mercury switches don't qualify and are also banned for ecological reasons. On the other foot:

I haven't seen any of those. However, I suspect that like any similar type of "tipping" thermostat, it would have a rather large hysteresis. I could probably get rid of much of the hysteresis by making the tube assembly circular and roughly concentric with the arc followed by the bimetallic spring arm. However, as the spring slooooooowly creeps in the direction need to close or open the contacts, there will be plenty of time for the gap to be the proper size to produce an impressive arc. If a low voltage thermostat, plenty of contact noise instead.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

Not necessarily the ONLY factor, though; peak current, arcing, dirt sensitivity, contamination, oxidation... there's a lot of ways to kill a switch. Mercury switches don't always have a MTBF rating, but mercury relays do.

Reply to
whit3rd

A little heater resistor comes on, raising the bimetal by about the same temp rise as the hysteresis. The result is a fraction of a degree hysteresis.

every British central heating stat that hasn't been replaced with an electronic one, eg

formatting link
includes three

how? if you want low hysteresis you have to live with low contact pressure, and that means slow switching. Arcing goes with the territory. You could add a relay to relieve current, but the coil is inductive. You could add electronics and sense either the short or capacitance to enable even a well oxidised contact to work, but then you've got the unreliability of the electronics.

Kind of. In practice electronic goods seem less reliable than electromechanical.

NT

Reply to
tabbypurr

On Fri, 14 Aug 2015 11:42:06 -0700 (PDT), whit3rd Gave us:

Silly child. Those are all CONTACT failure modes.

Everything has an MTBF rating otherwise nothing would. It's the weakest link thing.

That is why America would be absolutely stupid to put an idiot like Trump in office, much less Clinton.

Reply to
DecadentLinuxUserNumeroUno

On this side of the planet, we don't have those. Mostly, we have setback thermostats like these:

I don't see how the little heater resistor trick would work. What turns on the little heater before the main contacts close? When going from off to on, the contacts haven't closed yet, so there's nothing to trigger the little heater. When going from on -> off the little heater is presumably on, so the little heater resistor would need some way to predict that the contacts are about to open and turn off early.

Those are all electronic thermostats that use thermistors or thermocouples. What I would like to see is the guts of a mechanical bimetallic thermostat with the little heater resistor trick. I tried Googling for such a device, but couldn't find anything, probably because I'm missing a UK specific buzzword. What are they called?

Easy. A strong spring will close quicker than a weak spring. If you want to reduce arcing duration, just switch faster. That's one way big electrical contactors prevent contact meltdown from arcing. Lots more under arc suppression, but none which really applies to a fairly low current thermostat. My only point is that if the thermostat contacts open slowly, the arc duration is longer, and therefore the damage more extensive. It may be a tiny arc, but after 100,000 cycles, the damage can accumulate.

Yep. Zero contact pressure will give you nearly zero hysteresis. However, if you want to minimize the damage caused by even minimal arcing, faster switching will be needed, which means stronger spring tension, which implies more hysteresis.

I guess I should mention that even the electronic thermostats have designed in hystersis. It would not do to have the HVAC system cycling ever few seconds or every time the room door is opened or closed. Hysteresis keeps the machinery from getting pounded to death by too many starts and stops.

If you're going to add the level of complexity, you might as well go all electronic.

Not exactly. The components used in electronic solutions are generally more reliable than a mechanical solution. The problem is that electronic solutions tend to have far more components than electronic solutions. In other words, electronic solutions are less reliable simply because there are more things that can fail. There is also a tendency to design electronics so that it only lasts 1 day more than the warranty period, which offers additional opportunities for failure. While this can also be done with mechanical solutions, I haven't seen it done as often. However, it's not the reliability of bimetal thermostats that make them attractive. It's the low price.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

Mercury wetted relays are rated in cycles. I learned this the hard way while helping troubleshoot a fixture to test 16 bit DAC's for linearity and missing codes. We used relays that were rated for 2 million cycles. The LSB relay would thrash 65,000 times during one test cycle. Therefore, the relay was good for: 2*10^6 / 65*10^3 = 31 DACs Every 31 DACs, we had to replace the LSB relay, and every 62 DAC's, we also had to replace the next bit, and so on. That rapidly became very tedious but at the time (mid 1970's) we didn't have a better way.

Incidentally, I came up with a method of rejuvenate the relays. I would ultrasonically shake the relays which caused the breakup of whatever slag was covering the contacts. It wasn't reliable, but lasted long enough until the next batch of replacement relays arrived.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.