PWM'ing a 250W heater.

[paulca]

Do not focus too much on heating, loosing main objective.
And main objective is to not to burn your house down. Make sure that in case of any controller failure the losses are contained. If the failure of some component (fan) brings fire hazard - this is not what you want. If you still need that controller, use some safety rated components, like thermal cut-offs. So peak mains voltage, lowest tolerance heater element, triac dead short and seized fan should not cause havoc there.

Obvious risk is a 240V heater in a humid environment with water around, in a sealed plastic tent.  That said, it is a reputable domestic brand and has it's own thermostat (set to half way), an normal overload thermal cut out on the element and a "fall over" switch which means it has to be sitting upright and if it falls over it auto shuts off.  It's cover proof and knock-over proof.

There is spill risk against the electricals at the bottom of the tent.  Mitigation there would be to put the electrics up off the floor, hung or on risers at least.

Anything that can be outside the tent is outside the tent, only things that have to be in there are contained in it.  Heater, Fan, Humidifier, Heat mat... sensors.  The extractor is currently inside, but I am moving it outside to save space and risk.

The tent itself is in the garage and the mitigation there is to move the junk boxes away from it.  If there is a fire, the tent will quickly shrink and drip into a burning mass of plastic.  Unless it can reach the wooden 2.5 meter ceiling beams before it burns out... the afore mentioned junk boxes are the biggest risk for spread.

I should have a smoke alarm in there anyway.  A good smart one would be nice as I'm not sure I would hear a smoke alarm in the garage from bed, but if I can repeat the sounded inside bonus points.

[paulca]

The way I have implemented these control systems for temperature and my home heating in the past has been more strict that this project.

In my heating control I do not rely on anything, or any device to turn something OFF.  OFF is just the default state and if nothing demands otherwise it turns off an stays off.  That extends down to the firmware on the boiler controller noting the expiry on the control message to turn it ON and if that time expires and nobody has updated it, the firmware switches the relay off.  (I confess this code is still not deployed on the relay device... shame.. it is contained in the HTTP controller for it)

They way this works is, something sends an update to a temp, that results in a need for heating.  A "demand" is raised with a time stamp and an expiry of 5 minutes.  The heating will now be on for 5 minutes unless something specifically turns it off.  The system will constantly strive to maintain that it stays ON, even if you try and manual turn it off or reboot something.  It will not give up.  However, if the Gods strike the house and the server goes on fire, that demand will expire and the heating will turn off.  However, assuming the world is normal and nice, the demand will be continually refreshed until it's no longer necessary, when it will be left to expire.

It's fault tolerant is most cases as it defaults back to safe.  The only component that could fail and leave the heating on is the MCU controlling the relay locking up with the relay closed.  Not yet, but I have plans for that also, certainly in terms of monitoring the health of my sheep (wifi devices) to keep tabs on them and raise alarm if one of them is offline.... especially if it's in a dangerous state when it did so, such as "ON".

[paulca]

Seems to be functioning as expected these days.  The attachment is 24 hour period.  The lights are OFF between 10am and 5pm. 

You can see the heater "accordian" showing the coolest part of the night and also when the light is out, the total heat drops by 100W, so the heater kicks back in.

It's odd though.  When it gets warm enough to not need a heater, I will be done with the propagator for the year....  Unless I use it to play with fully automated, electronic controlled hydro!  More hobby, more $$$.  It's always the same.

EDIT:  The "garage temp" in the graph is recorded by a sensor a little close to the south facing wall, it reads artificially high.  If the garage ambient was actually above the propagator temp, I would just flip the ventilation to high

[Microdoser]

plenty of triac-based temperature controllers out there - why reinvent the wheel

Actually I have a Triac controller but it's a manual knob.  I'd need to hack that and produce a voltage in lieu of the potentiometer so I can automate control of it.

I think you had something like this in mind though.
https://www.amazon.co.uk/Inkbird-Temperature-Controller-Thermostat-Thermocouple/dp/B01N2XZ4HV/

On reinventing the wheel, I always find that sentiment a little ironic, considering how many different types of wheel have been invented.

I have one of those I bought from Ebay for £11.99, although the price has since gone up to £15.99. Very similar controller, uses a thermocouple (included) and a 40 amp SSR (included). The SSR states 40 amp but BigClive dot com recently did a teardown of one and found the internals only rated to half the stated amps, but that is still 20 amps.

Works fine. Starts full on then reduces the pulse width to maintain a set temperature to within 1 degree C.

I use it to control a toaster oven for SMD reflow, copes fine with the ~800W heater in there so 250W would be fine.

https://www.ebay.co.uk/itm/Digital-LCD-PID-REX-C100-Temperature-Controller-Set-K-Thermocouple-Max-40A-SSR/313203913699