Measure solar radiation (W/m2) using SI1133

[Dave_PT]

I'm moving forward with a project that has to be closed in a short time for a prototyping of a few hundred units.

An important part is the measurement of solar radiation. I decided to use the SI1133 and make the sum of IR, visible and UV. For the application I'm doing is perfect.

However I am studying more deeply the operation of the sensor and I am very confused. I think I got even more when I saw some online sw libs for it.

What are the ADC's settings and the conversion calculation of the channel readings to W/m2?
How can I relate to table 8.3 (datasheet)?

ADCCONFIG0 = 0x02;  - IR
ADCCONFIG1 = 0x0D;  - VISIBLE
ADCCONFIG2 = 0x18;  - UV

I can not move forward from this point because I do not understand what kind of result the sensor will give me.

My strategy would be to get the direct values with the sensor without a lens (to be able to use the ratio of table 8.3) and use another sensor with a Teflon diffuser lens to calculate the lens attenuation.

The goal is to do something similar to a pyranometer.

Can anyone give me more infos? Some help? Is my thinking correct?

Links:
SI1133 datasheet
AN968
MBED_LIB SI1133

Thank you.

[Dave_PT]

The box where I will house the circuit will have a 8mm diam hole to insert this lens that I machined in Teflon.

If I can take sensor readings (in W/m2) without a diffusing lens and use these measurements to "calibrate" the sensor with a diffusing lens, it will be fantastic.





But for now the difficulties remain: how to configure the ADCs so I can relate the measurements with the table 8.3.

[Dave_PT]

Has anyone used any of these sensors?
SI1142/43/44 have the same operating principle. Has anyone used any of these?

I think an Application Note is needed to clarify how these sensors work.
The datasheet has some inconsistencies and the only AN available is about mechanical sensor mounting.

[RobK_NL]

If I were you, I would post my questions on the SiLabs user forum. https://www.silabs.com/community/sensors/forum
My experience with the MCU forum shows their applications engineers to be very helpful.

[ace1903]

Is teflon UV transparent? I have development board think it is with si1133/si1153 and teflon cover is like small porous membrane.
Just to diffuse incoming light. Think teflon  would not work if it is not specially prepared to be extremely thin or wound like sieve(process to get textile like material).
Haven't checked on wikipedia but think lens made of teflon will block large range of light spectre or modify sensitivity characteristic drasticly.

[Dave_PT]

Thank you both.

I already contacted Silabs (open a case) and I am waiting for response ... but I do not know the response time (with STMicro round 1/2weeks).
I can try the Silabs forum to see if I have a faster response.

I found in this LINK a lot of useful information that made me realize the basics of the sensor and solar radiation measurement.
According to them, Teflon is the ideal material for this type of application ...

Teflon has been used because it is a good diffuser and is also resistant to the elements and ultra-violet (UV) radiation, given its capability to diffuse transmitting lights nearly perfectly. Moreover, the optical properties of PTFE (Teflon™) remain constant over a wide range of wavelengths, from UV up to near infrared.

In table 8.3 of the datasheet they write ADC_GAIN, but in the registers it is possible to configure SW_GAIN and HW_GAIN. How do they relate? This is an example of information that does not make sense to me.
Another thing that confuses me: if I set up a channel for LARGE_IR (for example), when I read that channel how do I relate the value read with W/m2?

[ace1903]

Link is indeed interesting. It is also interesting that all parameters for teflon are for thin films.
Is teflon available in  clear form? On ebay i found only tubing and not parts for machining.
I would like to see how final part looks like(plastic often is difficult to make with good surface finish).

SW gain is simple accumulator. if you skip shift operation then it can be seen as gain. if you shift result then it is averaging.
HW_GAIN is actually integration time for ad converter.  For same input if you put longer time counter will count more and result will like as gain is applied.
In my opinion it is misleading to call them adc gain as they aren't that. I always relate adc gain with presence of PGA.

Result from the link with bpw21 are impressive. Waiting to see results you will get from si1133.

[Dave_PT]

Link is indeed interesting. It is also interesting that all parameters for teflon are for thin films.
Is teflon available in  clear form? On ebay i found only tubing and not parts for machining.
I would like to see how final part looks like(plastic often is difficult to make with good surface finish).

The diffusers are white.
In the link I left above, they used a teflon sheet. I used a 10mm rod similar to THIS.

SW gain is simple accumulator. if you skip shift operation then it can be seen as gain. if you shift result then it is averaging.
HW_GAIN is actually integration time for ad converter.  For same input if you put longer time counter will count more and result will like as gain is applied.
In my opinion it is misleading to call them adc gain as they aren't that. I always relate adc gain with presence of PGA.

I agree. Calling "ADC GAIN" was making me confused ...

Result from the link with bpw21 are impressive. Waiting to see results you will get from si1133.

I'm studying the situation and I've had direct support from a Silabs AE.

I just want a "rough" approximation of the value in W / m2 ...
If I point the sensor to the sun (it will receive light at various wavelengths) and set the sensor to "ADCMUX = 13", can I assume that the result (roughly) will be 980 ADC_Counts/(W/m2)?
This is the doubt for now ...

[Dave_PT]

I already have all the hardware assembled and I am already communicating correctly with the sensor.

But I'm getting surprised at the ease of saturation of the sensor.

At sunset, I pointed the sensor (without any difusor, only the "naked" sensor) to the a sun ray and (ADCMUX = 13, with output in signed 24bits) and the sensor as saturated ...  the readings were 0x7FFFFF.
This should only happen if the solar radiation were 8559 W/m2 (2^23/980)...

It seems very high.

The Sun's rays are attenuated as they pass through the atmosphere, leaving maximum normal surface irradiance at approximately 1000 W /m2 at sea level on a clear day.

wiki

[EDIT]
If I consider a distribution per percentage, then in fact 980 counts will correspond to 5W/m2.
This would be a worse scenario.

I'm not sure how I can do this conversion ...


Any help or ideas?


Thank you!

[Dave_PT]

Just to follow-up ...

I ended up calibrating with a weather station nearby and got a good parameterization of the sensor.
At this point, using only a piece of diffusing film, it is working properly (error ~3%).

But in the next project I do not use these sensors, but by PIN diode. Cheaper and with more spectrum coverage ... what is the necessary for this application.

 

[Zero999]

Just to follow-up ...

I ended up calibrating with a weather station nearby and got a good parameterization of the sensor.
At this point, using only a piece of diffusing film, it is working properly (error ~3%).

But in the next project I do not use these sensors, but by PIN diode. Cheaper and with more spectrum coverage ... what is the necessary for this application.

 

If you don't need to distinguish between wavelengths, then how about a bolometer?

A black (in the frequency range of interest) surface which will absorb the light hitting it, heating it up. If the ambient temperature is known, then the temperature rise of the surface is dependant on the power level of the radiation hitting the surface.

[Dave_PT]

Exactly, that's how pyranometers work.

But mechanically it is more difficult to assemble. I think the black surface can not be exposed to "ambient air". This is why pyranometers always have a dome to protect the surface of the air stream.

So it's a more "scientific" and better resolution path, but for my application there was something very complex to build to get an approximation of W/m2.


Thx