Tolerance on stats

[The Medway Handyman]

What's the +/- tolerance on a bog standard wall thermostat?

I mean, if you set it to 20c will it trigger on at 19c & off at 21c?

[<me9]

Depends much on the model and how it is set up. Older ones have accellerator heaters which increase cycling. Later ones go into a proportional(*) mode when near the set point.

Well sort of, mine has a ten minute cycle once it is near temperature, the on/off mark space ratio is altered depending on temperature.

[Roger Mills]

Are you sure that you mean tolerance - 'cos what you've described is hysteresis. If by 'bog standard' you mean a mechanical by-metallic stat, that will inevitably have a difference of several degrees between the switch on and switch off points. As someone else has said, these often have a built-in accelerator heater. This is a small resistor which is powered when the stat is 'on' and warms up the innards so that the stat switches off slightly before the target room temperature is reached. Of course, the rads don't cool immediately, so the room goes on getting hot for a bit - so the accelerator heater just helps to reduce overshoot.

In terms of absolute accuracy (which would be closer to my definition of tolerance), I suspect that the relationship between the number shown on the knob and the actual temperature at which it switches is probably not a very close one! It's probably best to ignore the numbers, and set it to give a 'comfortable' temperature - or hang a mercury thermometer alongside it if you want to calibrate it.

Now, digital stats are different altogether - and not *that* expensive.

[Grimly Curmudgeon]

We were somewhere around Barstow, on the edge of the desert, when the drugs began to take hold. I remember "The Medway Handyman" saying something like:

Do you mean accuracy? I'd be surprised if most domestic stats are any tighter than +/- 5% of indicated figure, and 10% on older ones.

[John]

I think the usual term is "Hysteresis" - the amount of slack in the system between on and off.

Tolerance would be associated with 20c meaning 20c and not 18c.. - ie the calibration of the dial or display relative to the real temperature

Hysteresis of 1 degree would imply on at 20c and off at 21c

[The Medway Handyman]

I mean hysteresis now I know the correct term :-)

Thats what I was after knowing - thanks.

Didn't know that.

[meow2222]

Unfortuately its not correct. Hysteresis in bimetals is typically around half a degree. The natural hysteresis of bimetals is indeed a few degrees, but the accelerator resistor warms the stat mecha slightly after it switches to reduce this down to half a degree or better in use. The acelerator resistor has nothing to do with preventing overshoot.

BTW you cant tell the hysteresis by turning the dial, as that will only show you the hysteresis of the bimetal mecha, which is then modified by the accelerator R in use to give much smaller hysteresis.

Bimetals never go into proportional mode.

NT

[Roger Mills]

I fear that we're getting into semantics! Yes, the accelerator resistor reduces the hysteresis rather than the overshoot per se - but the effect is very similar. By switching off *before* the set temperature is reached, the likelihood of overshooting the set temperature is reduced.

We seem to be in agreement that - in the absence of an accelerator resistor - a bi-metal stat will have a hysteresis of several degrees.

[meow2222]

But it doesnt do that. The hysteresis means it overshoots its set point every time. Not by enough to cause a problem.

NT

[Roger Mills]

It doesn't do what? Are you saying that it doesn't switch off before the set temperature is reached? Clearly it does! The innards will have reached the set temperature because of the effect of the accelerator resistor, but the surrounding air will be at a lower temperature - giving scope for the thermal inertia in the rads to raise the room temperature a bit more without overshooting the *true* set temperature.