I tested the working temp of the ballast in this floodlight as i mean to repaint it (it is rusted), and i want to estimate the temperature the paint will have to withstand

The ballast is rated dT70 tw130 - 1.8A 150W HPS ballast

I powered it up for 4 hours in the enclosed fixture, in a room with ambient temperature about 15C. The ambient temp inside te fixture was maybe 30-40C

After 4 hours of work, i could still touch the ballast core (on the limit but i could) - id guess about 60-80C



How much different is Tw typically from the temp of the core outer surface ?

Are 4 hours for a fixture like this enough to reach final temperature ?

Is a "enamel paint for washing machine and fridge exterior surface" in a spray can good for this ? Is a standard white graffiti paint good ?

Does it withstand the (probably less than 100C) temperature ?

Does it seal against moisture (so that condensation of water vapors on cold ballast won't result in it rusting) ?

Does it crack from thermal expansion / contraction of the ballast ?

Does it thermally isolate the ballast significantly, which might overheat it ?

The temperature difference between the ballast core and it's surface is hard to say, the highest would be somewhere inside the winding in the center (length wise) of the ballast. It depend, how well is the ballast impregnated.

The term "for washmachines,..." is very vague, I guess without more accurate paint specification nobody can tell from this. If I could compare, washmachine s are extremely wet, but by far not as hot as the ballast could be on the surfaces expected to be painted.

I know some powder paints (solvent less) do endure quite high heat and still are quite elastic, but you need a special electrostatic equipment and a curing oven to apply them. Nothing for home use.

But with some materials not the heat, but freezing may be more of the problem - the coat become fragile and start to crack.

The paint does not specify any usefull data. I do however assume that it can keep water off the core, and that if i can hold my hand on the core than it is not too hot for the paint either

But thats as long as the ballast's temperature does not rise as result of the added paint (thermal isolation...) layer - which i dont know how severe is going to be

This is just a spray paint

The coldest you get here in winter is down to -5C (very rare down to -10C). Is this a problem ?

4 hours will be pretty near thermal equilibrium, but you will have to measure the surface temp with a thermocouple to get a more accurate indication with the cover on. Hand hot just means below 60C, so Duco-in-a-can ( any paint in a spray form you buy over the counter) will survive for a while before it browns. It may react with the original paint, which is probably a solvent based paint anyway, to form ripples and crazing. Standard white PVA ( your paint that you apply with a brush, and clean with water before it dries on the brush) will not work at all.

Apply only a thin coat, and allow to dry well ( at least overnight in a well ventilated area) before applying power, then leave it on for a few hours ( 4 hours will do if you are there to check if the paint bubbles) to bake it to a hard finish. This will then provide protection for a while, depending on the humidity and cycles it gets in use. Here I am in the second worst place on the planet for corrosion, a close second to the Skeleton Coast in Namibia. Bare steel will rust in minutes here.

And to determine the average winding temperature: Measure the winding resistance (by an Ohm-meter) when the ballast is cold (so you know the temperature), then run it to stabilize the temperature and then measure the resistance again. From the known thermal coefficients of copper you calculate the temperature. But it would be about the average within the winding. Now if you add to it the difference between the ballast copper average and the surface, you get pretty accurate estimate of the peak core temperature (Tpeak ~ 2*Taverage - Tsurface).

And to determine the average winding temperature: Measure the winding resistance (by an Ohm-meter) when the ballast is cold (so you know the temperature), then run it to stabilize the temperature and then measure the resistance again. From the known thermal coefficients of copper you calculate the temperature. But it would be about the average within the winding. Now if you add to it the difference between the ballast copper average and the surface, you get pretty accurate estimate of the peak core temperature (Tpeak ~ 2*Taverage - Tsurface).

Interesting! The winding resistance changes that much to measure with a standard multimeter?

If you start from the generic equation for the resistance:
Rtemp = Rref * (1 + Coef * (Temp - Tref))
The Rtemp is resistance at the required Temp temperature
The Rref is resistance at the Tref temperature, where the thermal coeficient Coef is defined.

So when playing with the equation, you get the required temperature:
Temp = Tref + (Rtemp/Rref - 1)/Coef

The thermal coefficient of the copper resistance is 0.0039 K-1 at 20degC, so:
Temp = 20 + (Rtemp/R20 - 1)/0.0039

So e.g. when the resistance increased by 31% compare to 20degC, the winding temperature is 100degC.

You may improve the accuracy using the fact, then after removing the power, the temperature of the winding and core equalize pretty quickly, so if you put the thermocouple just between the winding and the core, you may correlate the winding resistance with the thermocouple on the temperature closer to the operating one, so eliminating some measurement errors (e.g. the resistance offset,...).