Bimetallic switches reliability

Assume it's wired parallel with the contacts

Thanks, that should work. With the contacts open, there's some current going through the heater resistor, which increases the bimetal strip temperature to something less than the hysteresis. When the contacts close, the resistor is shorted, eliminating that source of heat. Assuming the ambient temperature doesn't change, the bimetal strip now drops in temperature to again something less than the hysteresis. It sits there forever, or until the device temperature drops the remaining difference. The bimetal strip then trips, opens the contacts, and the cycle repeats.

Looks like it needs more than just a resistor: Looks like the missing buzzword might be "heat anticipation thermostat".

I can see some problems adjusting such an arrangement, but with a consistent enough bimetal strip, it should work.

However, I still question if it's necessary or desirable. A zero hysteresis thermostat will be constantly cycling with every passing breeze. That would beat up the HVAC machinery or whatever it controls. A time delay could be added to either the thermostat or the machinery, but that puts us back to an all electronic solution. Do the heaters in the UK have built in on/off delays?

On Fri, 14 Aug 2015 15:31:33 -0700 (PDT), Phil Hobbs wrote:

Bad planning and a lazy engineer (me). Rotating the relays is what everyone wanted and suggested. The hourly labor hired to run the test turned out to be incapable of replacing the relays because the fixture had not been designed with relay sockets[1] and ease of access. Since it was a one of kind, no effort was spent making it easily repairable. At about 15 mins per full test per device (or 15 msec/bit), 31 tests would be about 8 hrs before one or more relays required replacement. There were two shifts, so I was sentenced to replacing the necessary relays when I arrived and again when I left work. Since the relays had to be desoldered, I was not particularly thrilled with spending an extra hour or so moving all the relays towards the LSB. So, I made a chart of which relays had to be replaced, and performed only the minimum replacement, which usually took about 10 mins. There was also the problem of the nearby precision resistors, which tended to drift when exposed to soldering iron heat. Fortunately, the typical lot run was about 100 DAC's quarterly, so the most I had to deal with this was: 100 / 31 = 3.2 days of swearing and snarling every 90 days. Manufacturing declared that it was easier to tolerate my complaining than to rebuild the fixture[2]. I vaguely recall performing this ordeal for about 3 years until the product finally died a merciful and well deserved death. The worst part of this was that it wasn't my project or even my division. The company had two divisions (modular products and marine radios) and I worked for marine radios. I was "helping out".

Lesson learned: Never make a suggestion where I'm the only person available or qualified to implement it.

With bimetals the heating is just switched off at night. Houses only drop a couple of degrees overnight on a very cold day.

ctronic one, eg

Google isn't showing compensated ones. I'll see if I can get a pic.

but that's not much use when you want low hysteresis

there are other options

hanical.

another reason is simply that typically with (inevitably complex) electroni cs, some failure mode was not predicted. Goods can be very well designed in almost every respect, it only takes one little oversight to render them de ad. And its tough to avoid that.

both are attractive. I've replaced electronic with bimetal for reliability. The old dream of replacing everything mechanical with electronics to solve the reliability issues hasn't really worked out. In some cases electronic is better, in some not.

NT

its just a resistor

if you want to find patents maybe.

The point is not to zero hysteresis, its to reduce it to the level desired, typically about 0.5C. Electronic room stats tend to shrink this further to trim energy use, down to 0.25C.

NT

Ok. Got it, maybe:

What surprised me is this line from the above URL: "The anticipator also creates hysteresis, so the furnace does not cycle too often." You suggest that it reduces hysteresis which also what I would expect. So, is the web page in error?

In addition, it mentions: "QP means by hysterisis that we introduce a lag or delay between the change in room temperature and the response of the thermostat. However the heat anticipator is only operating to turn the thermostat "off" a little "early". It does not introduce a delay in turning the thermostat back on. - Ed." I thought it worked in both directions?

I still have a problem with reducing the hysteresis. You mention

Full disclosure: I don't know much about HVAC and missed the step in growing up where I was suppose to tear apart the wall thermostat to see how it works. I've also played manager in my fathers apartment building and rental landlord, where I've replaced such thermostats and never needed or bothered to adjust any.

Ok, so you do have an external timing delay to prevent "thrashing". I guess 12 to 15 mins cycle time is tolerable. At this point, I'm not sure which is worse, a fan that cycles every 12 mins, or (for example) a 5 degree variation in temperature over a longer period.

I didn't ask, possibly because my parents considered my messing with the house thermostat a capital crime. Other than clean out the dust, I don't think we ever did anything with it for about 40 years.

Yep. A very sad situation. When the cost of a truck roll exceeds the value of the equipment being adjusted, the equipment soon becomes uneconomical to service. We've been there for at least 20 years.

Thanks for the details.

You can have it either way. A bimetal with spring or magnet action has mech anical hysteresis, and the resistor reduces the hysteresis. A bimetal witho ut spring or magnet action has no inbuilt hysteresis, but also near zero co ntact pressure at switching - they are used for some things.

about 0.75F no less. Don't forget in practice the bimetal is in a case atta ched to metal, so if you breathe on it once it won't respond.

we use about 0.5C here

temperature hysteresis plus slow stat response is all you need

Hang on. More people can afford ac now than in the 60s because the equipmen t is cheaper, and run cost lower. Same reason kit doesn't get serviced, its price has fallen, meaning we do better today. Its a shame to see so much k it get junked, but it is economic progress.

NT

it brings "off" closer to "on", technically that's only working in one direction, but the net effect ix that at both on and off periods will be reduced.

thermostats have themal mass, it'll take a minute or so to react to a 1 degree change.

There is a hint: Those little heaters are called anticipators.

The purpose is not to eliminate hysteresis, it is to roughly model the thermal inertia of the room, to reduce overshoot.

I would look into the old patents on the Honeywell round thermostat, the ones with bimetal spiral plus mercury switch (which are still in use in my home).

Joe Gwinn

You'd need a lot of thermal inertia inside the stat to get it to warm up from the compensation R as slowly as a room.

NT

NT

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

MAN! That technology has probably been around since to 30's, and your asking questions like that, it's not even a controller even though they call it one. Replace it with a thyristor circuit and a temperature sensor so that it really is a controller! This is 2015! You could get fancy and skip the temp sensor and have a microcontroller monitor turn on current of the heating element, with that you could determine the temperature by the resistance of the heating element and how fast it reaches full temperature / minimum current.

Shaun

On Wed, 19 Aug 2015 23:33:00 -0500, "Shaun" Gave us:

You're right. $23 is pretty damned cheap...

Actually its an 1800s invention, 1830s & 1880s. Well over a century of technological development has given us less reliable electronic stats.

NT

On Thu, 20 Aug 2015 04:27:44 -0700 (PDT), snipped-for-privacy@gmail.com Gave us:

Well, that was a boredom tube post.

If reliability were the end-all, we'd still be writing on stone tablets.