Maintaining constant overnight bedroom temperature with electric heater

[Caecilius]

I've got a large house with gas central heating which I use during the day. It gets cold overnight in the bedroom though, so I use a 2KW electric convection heater with the thermostat set to about 18 C on a timeclock which switches on between midnight and six am (boiler comes on again at six).

This was a great improvement last winter, and it only uses one or two KWh overnight which must be cheaper than heating the whole house with gas.

This year I want to improve the solution to keep the temperature more constant (there's a big hysteresis on the convection heater thermostat) and remove the bimetallic thermostat clicking noise.

I think I want electronic switching to remove the clicking noise and allow a lower hysteresis plus an external temperature sensor. Maybe something that's closer to a temperature controlled dimmer switch instead of a standard bimetallic thermostat.

My plan is to put this temperature controller between the timeclock and the heater, and set the heater thermostat to something like 25C as a fail-safe.

Has anyone found or built something like this? I'm sure I can't be the only person who want a constant overnight bedroom temperature without running the main house heating.

[The Natural Philosopher]

Mmm. Raspberry Pi with a solid state relay, real time clock and temperature sensor?

To be honest in most control situations its as good to have a hysteresis and longer cycle time on the element as to have a modulated in real time electricity supply.

So I would go solid sate, and push the hysteresis down, rather than go dimmer.

is one example I found.

Or you would use a wireless stat and put the receiver in a soundproof box.

Or suffer from Hives...I am sure they do something

[williamwright]

I've experimented by turning the central heating off overnight and it doesn't seem to reduce gas consumption much. It seems to mean that the house is horribly cold for the first hour of the day and the heating never switches off all morning and most of the afternoon. I think the thermal momentum of a brick building and its contents must mean that cooling down and warming up take an inconvenient amount of time.

Bill

[Caecilius]

I could certainly do that, although I'd probably use an arduino rather than a pi because it's a very simple task that doesn't need a general purpose OS.

That's an interesting view - I'd assumed that "dimming" the heater by chopping the waveform would be the way to go, but I guess it ends up making no difference if the hysterisis is low enough. Something like

0.5C would be unnoticible.

Thanks for that link: although the 0.2A capacity means it won't work with a 2KW heater there may be similar units available that do what I want.

I think the takeaway is solid-state switching plus low hysteresis.

Possibly. Sounds a bit more complex and heath-robinson that I need though.

Not sure how a hive would help TBH

[T i m]

<snip>

Or have both functions (clock / stat) combined?

I'm not sure the actual switch is silent though but it is very quiet.

The problem you might have is getting the heater itself to stay on without it's overtemp or main stat cycling, even though the remote timer / stat is calling for heat?

I did buy some mains level PWM power controllers with the intention of mating them with an Arduino and remote thermostats (one on the heater and one in the room) so that I could manage the power to a small oil filled rad (that couldn't dissipate it's heat energy fast enough so would modulate on it's overtemp stat) and manage the room temp both silently and more accurately (no temperature overshoot etc).

Not gotrountuit yet of course ... ;-(

Cheers, T i m

[Caecilius]

You don't have my gas bill!

I like my house, but it's expensive to heat, and having the CH off overnight makes a big difference. Having it come on again at 6.00am means its warm enough by the time I get up.

I need to do other things to reduce my energy consumption like replace the Aga with something from the 21st century and bring my EPC rating up from it's currently dire E rating. But for now I'm trying to live comfortably while working through the various improvements.

[Caecilius]

Yes, no reason not to combine the two functions. But that's not essential.

It's probably an improvement on the WW2 era bimetalic strip that my convection heater uses, but I suspect it's still a relay inside and it probably has a fairly high hysterisis to preserve contact life. Unfortunately the data sheet doesn't give much information about the switching method or the operating parameters.

I don't think that will be a problem. If I set the thermostat on the convection heater to, say, 25C then it shouldn't activate unless the room is unpleasantly hot. And if the internal overtemp operates then there's something badly wrong.

We've all been there with arduino/pi things I think. I know I have.

[T i m]

<snip>

It just stopped there being a 'tree' of plugin things. ;-)

Quite, the overtemp stats in the little (600W) oil filled rads are particularly noisy (in an otherwise quiet bedroom).

Yes, that sounds right (from the discrete 'Click' you can just hear now and again).

No, it is a pretty basic piece of kit but it's what's right in front of me now and in it's second season of use.

No, it *shouldn't* and being a convector, might not.

Well, that's how I treated the functionality of the (several) oil filled rads I have here but it's how they work nonetheless.

Even the 600W ones can generate more heat than they can dissipate and so do 'overheat' and then cycle on their overtemp stats, way before they may have reached the temp set by the main thermostat. I even considered (and may well try this year) or rigging up two sockets on an extension lead and wiring two of the 600W heaters in series so they become two x 300W heaters. Set one stat on Max and then use the other to manage the actual room temp.

I just had a look on eBay and there are quite a few choices of 2kW opto coupled module that should be very easy to interface. An Arduino Nano, a Dallas I2C external temperature probe and a few lines of code and you should be able to do what you need.

Mount it all in a plastic double box (like a trailing lead) with a double pole switch on one side and 13A socket on the other and with the stat coming out though a rubber grommet to somewhere suitable in the room and you should be good to go (if you are ok with the coding). ;-)

I was sorting though some electronicy stuff earlier and came across an ESP32 with external SIM card reader that was part of a project I was playing with previously. That has built in BT and WiFi so you might even be able to use one of those and set / adjust the required temp with an App! ;-)

Cheers, T i m

[Chris Green]

I think ours is similar. ... and gas isn't cheaper off-peak, especially ours as it's LPG in a big tank.

We tend to have the gas CH on when there are a lot of rooms in use and use electricity for quick short term top-ups and in bath and shower rooms etc.

[T i m]

<snip>

<snip>

Thinking on, isn't the hysteresis partly a function of using a convection heater, over say an oil filled rad?

A convection heater has no thermal mass and being 2kW gives a large burst of heat, reaches stat temp, and very quickly cools down again. If the same 2kW heater was in the form of an oil filled rad you would get the same heating efficiency but with a lower hysteresis, simply because of the heater design?

Also, have you tried it on 1kW OOI? We prefer to have a lower power heater that runs a slower duty cycle for partly the reason you are trying to minimise.

I don't know how easily your 2kW convector can cope but if it's more than adequate, even in the worst conditions and better, can also cope easily when switched to 1KW (assuming it can be), you might find you can get away with a 600W flat panel oil filled rad and be even closer to your desired destination 'naturally'?

We use one of these in the lounge:

It's more that capable at getting the room up to temp fairly quickly and keeping it there ... and it's much better than one of the more upright oil filled rads we have because it's radiating surface area better matches it's energy output (so it's doesn't 'overheat').

Also less chance of it setting the house on fire if something falls onto it when it's unattended?

Cheers, T i m

[Caecilius]

That's useful information. I was considering replacing the convector heater with a oil-filled rad to smooth the output by increasing thermal mass, but I don't think I'll bother now if that's how they work.

Yes, I'm fine with coding - I used to be a software engineer in the

1980s so I'm comfortable with C.

I'm still hoping there's a neater pre-built solution though. I would hope that solid-state switching add-on thermostats would be common - I can't be the only person with this use case.

This is the sort of thing I like to play around with, but don't want to integrate into "production" equipment around the house that I want other family members to be able to use.

[Theo]

If you don't mind DIYing a case, there are PID controllers, eg:

're all clones of clones of clones)

I wouldn't trust the SSR with 40A but 8A mains should be ok.

I don't know if the thermocouples would be accurate enough - quite a difference in +/- 2C for room temperature. You might be able to manually offset it (ie ask 20C and get 22, so ask for 18C instead).

The advantage of PID is that it should aim to keep wherever you place the thermocouple at a constant temperature - it'll 'learn' the dynamics of the room and the heater to avoid peaks and troughs as it overheats and then overcools.

Theo

[Caecilius]

I think it's more to do with the difference between the mechanical thermostat switch-on and switch-off temperatures. Last winter I monitored the temperature and the graph looked like a sawtooth - can't remember what the min/max spread was, but I think it was a couple of degrees.

I'm thinking that either controlling the output power with something like a dimmer circuit (chopping the AC waveform) or switching faster with a solid-state relay or triac would solve that problem.

It only uses around 1KWh overnight at the moment (I've got a power-measuring device in the stack-o-plugs). As it's on for six hours (midnight to six am), that means it's only on for around 8% of the time. So I think it's more than adequate, and a 1KW would certainly work.

It's a bit big though/

Yes, that's true.

[Caecilius]

That's interesting. PID control might be a bit OTT for my application, but it's cheap enough.

Yes, I've seen enough big Clive videos to know how optimistic these ratings are. But as you say a 2KW pure resitive load should be well within it's capabilities.

I'd hope that there's something with a smaller temp range (0 - 1300 C looks like it's designed for a kiln) and an associated increase in accuracy. I'd like something like 0.5 C or maybe even 0.1 C if possible.

I'm surprised that a PID controller uses a simple SSR output. I would have thought that a "dimmer" type circuit using a triac for switching would be better for that.

[Johann Klammer]

Have a look at triac datasheets. For the 2kw heater you'll need a rather large one. (18 amps? BTA20 maybe) and a heatsink. Not all triacs are insulated,

as to how to drive them... you'll need some power supply, xfrmr or resistive. and generate pulses or pulse train with correct timing. There are application notes.

Microchip AN958 Low-Cost Electric Range Control Using a Triac (very much what you want)

ST has a lot of appnotes. search their website.

On semi AN1048/D (Snubber networks) HBD855/D (Thyristor Theory and Design Considerations) Handbook, large There's a zero point switch shown on pg 137. supposedly there's IC versions.

OTOH There are perfectly fine thermostats and heaters available wot utilize triacs. maybe get one of those.

[T i m]

<snip>

That's how some of the cheaper ones work yes. We are currently running a more traditional upright oil-filled rad (on 1kW) here in the lounge (with the external timer stat) and it's fine. The point being it's surface area is better matched to it's output power so it works 'properly'. ;-)

<snip>

That's *exactly* what I thought when I was looking for the same a couple of years ago and I couldn't find one then. ;-(

Understood. That's why something even 'preset' at whatever temperature you consider reasonable might be a good solution, or add reading the value of a pot to the project and make it fully adjustable. ;-)

Cheers, T i m

p.s. On these little (and they are pretty small) 450W oil filled rads I was also going to fit one of the PWM modules inside the unit to limit the power to something they could actually radiate!

[Theo]

Might come in handy for other things too. I put one in a case to control the soldering oven, but it gets borrowed for other things.

You might be able to change the thermocouple type - not sure if there's a setting for that (it'll depend on which clone it is).

I don't think it matters - the thermal mass is such that giving a blast of 2 seconds on / 2 seconds off is just as good as 50/50 duty cycle at 50Hz - and less buzzing.

Theo

[Peter Burke]

Caecilius [e-mail] wrote in news: [e-mail]:

In a similar application I found my 2kw convectors had 2 paralleled elements and disconnected one to make them 1kW only and less agressive in their heat output.

For control, honeywell have been forcing P+I control and multiple switching cycles per hour on us in their room stats for years (wearing out over cycled zone controls in the process).

Their programmable stats (eliminating the time switch) are a bit overpriced but this one has all the P+I, default 6 cycles per hour (programmable up to

12 if desired) and is 8A rated (selectable prog option) so will direct switch. The mag assisted batt powered internal relay should be quiet enough.

Instructions downloadable from the spec tab of that page for a preview.

Could save a bit of faff in rolling your own.

[T i m]

<snip>

That's a big part of course, especially if it's physically in the heater itself.

But the thing is the speed of the temperature increasing in the room versus the thermostats chance to monitor / react to same. Imagine putting a 10kW heater in there and how hot it would get before the thermostat had chance to do anything about it?

The closer you match the heat losses from the room to the heat being applied to it the smaller the chance of wide temperature swings?

Yes, I think that would work ...

Or better still, actually 'hold' the temperature at whatever you want using PWM rather than Bang-bang?

Ok, suggesting that 1kW (or even less) should be ok (and obviously, less hysteresis).

LOL!

Agreed.

Well yes, it's fairly long but not 'big' as such, as long as you can find somewhere to position it. Under a window maybe?

If you are a good coder (especially in C) maybe you could make the solution we and I'm sure (as you say) many others are waiting for!? ;-)

Or even if you didn't want to do so commercially (not everyone can handle being a millionaire <g>) maybe it could be put on the net as one of those d-i-y walkthroughs?

Arduino Mini, 2kW opto coupled PWM controller, Dallas I2C thermocouple and a cheapo 5V wall wart ... I'd build one. ;-)

Cheers, T i m

[Caecilius]

You're the second person who's mentioned de-rating a 2KW heater. I initially thought this was pointless, but perhaps it would help reduce overshoot.

I certainly don't need anywhere near 2KW as the duty cycle is below

10%.

I assume P+I is like a PID controller without the derivative term? What's "multiple switching cycles per hour" mean? The DT90E datasheet shows a "Cycle rate" parameter with a default of 6 and possible values of 3,6,9 or 12 so presumably that's cycles per hour, but the datasheet doesn't explain what that really means.

36.50 is not that bad, but I note this comment on the screwfix site:

"I have marked it down on performance as the "click" when it turns on and off can be heard all over the house"

So maybe not that quiet.