Hi,
I am tring to build my Meteostation, but i have much problem with the TSL2591. I see that values are not stable and change to much from one to other, in the same conditions. If i compare these values with an Unihedron SQM device, the values from the Unihedron device is the same or change of 0.02 but values that came out from the TSL2591 are completily different. Wht these difference??? Someone has compared the values from a TSL2591 with an Unihedron???

I use this device for my sqm without any problems. Here are the particulars
stargazerslounge.com/topic/345219-sky-quality-meter-with-wifi/
Not arduino, bu I hope it helps anyway.

Cheers

I also use a TSL2591 in RDBeck's DIY SQM

It works flawlessly here.

I also tried this, but the values are not stable

I'll check my MeteoStation (TSL2591 based) and come back to you.

Hi G_Gagnon,
do you have news for me???

Hi,
I‘m not sure whether we are talking of two distinct problems: stability and accuracy. And maybe both have the same root cause, a defective sensor.

What I know is that the indiduino weatherstation uses the same formula as published from Unihedron. But I would not be surprised if they have their calibrations before sending their device to customers.

-Wolfgang

I did also notice variations in SQM values, important ones sometimes, with myTSL based MeteoStation.

In order to try to find out the source of the problem, I stripped down the software from a WiFi weather station that I have that uses the same sensors (removed all WiFi, Web and file logging functions), and flashed my INDI MeteoStation with it. I found out that there is still some variations but the range of those variations is much less and the readings are more stable. This leads me to suspect that the firmata software used in the MeteoStation firmware may send some wrong data to the indiduino driver, for some non-evident reason. As I am not an expert in firmata, I can't confirm my hypothesis.

As Wolfgang mentionned in his message, this could also be due to a defective sensor but, as I also noticed some variations, they could be caused by some other issues, harder to investigate.

Not being very quick at it (lack of expertise and aging brain cells), I am leaning into writing a specific driver (not based on firmata) for the INDI MeteoStation that would be more streamlined and that would support more sensors (e.g. wind speed and direction).

Everyone that uses the TSL based indiduino MeteoStation should keep in mind that the SQM feature of the MeteoStation was meant to replace the irradiance sensor (solar cell) and give an estimate of the SQM and, while being based on the same formulas, likely lack some proprietary functions that Unihedron uses for calibration or other data massaging. That being said, I will try to find a solution to the data variation issue.

Gilles

As a followup to my message, here is a plot of the SQM values of the stripped down MeteoStation firmware. Taken through a window on a cloudy nights. Data points are every 10 seconds and there is a slight variation that could be smoothed out by doing running average or other smoothing algorithm, over a number of data points.

Gilles

I am tring to replace the code related to the TSL2591 with a code that use a TSL237, that is the one used by Unihedron

Just another doubt, where can I find a wiring schema for cabling the meteostation? I want be sure that all wirings are ok

You can look at this tutorial ( indilib.org/support/tutorials/177-howto-...no-meteostation.html ). It tells everything but it needs to be updated for the new hardware, which I am very slow at doing, unfortunately. The TSL connects through the I2C bus, as the BME280 and MLX do.

Gilles