Combined cloud and sky darkness sensor

I’ve got a combined cloud and sky darkness sensor up and running.

If you’re interested to see the results please click the “Daily Charts” button on my weather website and scroll down

With the cloud sensor (a Melexis 90614 IR thermometer) I plot the differential between the sky and ground temperatures. I notice that around -12C indicates cloud free skies, but since I am surrounded by tall buildings which impinge on the sensor’s field of view, the true differential is probably higher. It works both day and night but is less stable in direct sunlight, I think because of the tall buildings reflecting heat.

With the sky darkness sensor (a TAOS TSL237 light-to-frequency converter) I plot the darkness of the sky in magnitudes per square arcsecond.

Both of these are controlled by an ATMega168 and linked to the PC through RS232. These are quite simple circuits, I’m just pointing the sensors skywards and reading their values. I got the Melexis 90614 from Digikey and the TSL237 from Mouser, both of whom also have the respective datasheets available for download. I happen to be familiar with the ATMega168 microcontroller so used that for the interface, but I am sure many other microcontrollers would be up to the same job.

It would be great to see years hence, when many weather stations read the sky temperature, weather maps of the sky temperature differential being used to indicate cloud cover. :smiley:

Looks good to me. I’m waiting for the components from Maxim and Digi-key to make something similar myself, but based on a thermopile (MLX90247) and interfaced with a DS2438. This doesn’t give me straight temperature readings, but voltages to compare. I have been looking at the MLX90614 too, but since it is SMBus based I have no clue how to interface this with 1-Wire. I guess I’m just an analog guy :slight_smile: --Hans

I’ve had some initial success with the MLX90247. The signal fed into an AD8628 opamp gives me indoor readings between 0 and 5 Volts. I haven’t fed this into my 1-wire network yet as I first want to do some test measurements. I wonder what kind of casing you used for your set up. Since the sensor measures “body” heath I wonder what the optimal mounting position is. Thanks. --Hans


The sensor is waterproof so I drilled a hole in a plastic box and mounted the sensor through the hold from underneath. I used silicone bathroom sealant around the hole. Seems to hold together so far.

I have been reviewing this thread (all 33 pages) Detecting Cloudy or Clear night and have recently purchased a HT13 remote sensor for my iROX/Honeywell station. In the weekend I tried a few rough experiments with a jam jar stuffed with polystyrene, black metal capped with the sensor in intimate contact with it. Initial readings look very promising for both Solar and night cloud and can be currently seen on my website.
As a longer term solution I have come up with a design as attached. I know there have been many and varied designs with differing amounts of success. One of the problems I had with my experiments to date is that the neck of the jar is narrower than the body hence the poly is a loose fit. I was contemplating putting the glass through one of my diamond rock saws so I am left with a true cylinder in which I can core some poly to the right size and a tight fit.
Has anyone tried something like this and what results have you had.

Cheers Greg


Solar Ina Jar Design.pdf (9.38 KB)

For a cylindrical container it might be simpler to use a lab beaker.


I have been trying to find a supplier of lab equipment locally but cant find one. That was my initial thought as the glass would probably be somewhat thinner and more responsive to heat gain/loss.


You prompted me to try again. Looks like I can get some on Trademe (NZ version of Ebay). The ones I was looking at are made from Borosilicate glass. I did a google on this and found that this glass is used for Infra Red lights so should not screen out any IR and may well be better than a glass jar. Has anyone got any info on this product and it’s suitability for a solar in a jar sensor housing.


For a thermometer-in-a-jar solar radiation sensor project, you want the jar to act like a greenhouse. A greenhouse works because it’s glass roof blocks IR. The visible light passes through the glass, is absorbed by the contents of the greenhouse, and re-radiated as thermal IR, which is trapped inside the greenhouse by the glass. This causes the internal temperature to rise, roughly proportional to the amount of non-IR solar radiation that makes it through the glass.

Using an envelope that is transparent to IR would result in a thermometer-in-a-jar sensor that has significantly less temperature change per unit of solar radiation.

You also want the increased internal temperature to leak off through the envelope fairly rapidly. If the envelope insulates the air inside too well, it will stay warm for a long time after a period of sunshine ends. Therefore, you want the envelope to be thin and fairly small. There have been several posts here where people have successfully used small clear light bulbs for their envelope.


I think it would be more accurate to say that shorter wavelength IR passes in through the glass and is converted to long wavelength IR (heat) which does not pass back out through the glass.

Steve, thanks for that
Niko, A bit like low E (emissivity) glass.
The thing I like about the beaker idea is the glass thinness, hence faster heat loss and the better resistance to thermal stress. I have noticed that the inside temp on my experimental unit is up to high 30’s C and it is not that well insulated. The point at which thermal stress causes glass fracture is about 50 C differential with good clean float glass and no inclusions. Jars are not known for being manufactured from quality glass and have a few inclusions such as small air bubbles due to the forming process.
If I get the beakers I will try both methods and do a cross check.


I have done a bit of on line research on Borosilicate glass (Pyrex) and there is some evidence out there suggesting that it has a lower emissivity rating than ordinary glass. This means that short wave radiation (sunlight) passes through it but long wave radiation (from a heated surface) is reduced and reflected back in.
So that should be all good for a Solar In A Jar set up.