Comparing UV sensors?

Breakout boards, sensors, other Adafruit kits, etc.

Moderators: adafruit_support_bill, adafruit

Please be positive and constructive with your questions and comments.
User avatar
ibucky
 
Posts: 112
Joined: Wed Jan 11, 2012 10:40 pm

Comparing UV sensors?

Post by ibucky »

I see from the description that the Analog UV Light Sensor Breakout - GUVA-S12SD https://www.adafruit.com/products/1918 can detect UV from 240-370nm.

Wanted to compare it to the SI1145 Digital UV Index / IR / Visible Light Sensor

What's the UV range for the SI1145 Digital UV Index / IR / Visible Light Sensor https://www.adafruit.com/products/1777?

User avatar
adafruit_support_bill
 
Posts: 88096
Joined: Sat Feb 07, 2009 10:11 am

Re: Comparing UV sensors?

Post by adafruit_support_bill »

That ls not clear from the data sheet. The chart on page 14 indicates a pretty sharp cutoff below 400nm.

https://www.adafruit.com/datasheets/Si1145-46-47.pdf

User avatar
ibucky
 
Posts: 112
Joined: Wed Jan 11, 2012 10:40 pm

Re: Comparing UV sensors?

Post by ibucky »

If I'm reading page 15 correctly it is reading 280-400nm.

Please let me know if I'm not reading it correctly

Thank you.

User avatar
adafruit_support_bill
 
Posts: 88096
Joined: Sat Feb 07, 2009 10:11 am

Re: Comparing UV sensors?

Post by adafruit_support_bill »

I believe that the chart on page 15 is a standardized chart indicating the relative sunburn inducing effect of different parts of the spectrum.
Wikipedia has the same curve here: http://en.wikipedia.org/wiki/Ultraviolet_index

Image

User avatar
primexandy
 
Posts: 89
Joined: Sat Sep 06, 2014 11:09 am

Re: Comparing UV sensors?

Post by primexandy »

The Si1145/46/47 has photodiodes capable of measuring both visible and infrared light
and
The UV Index is a number linearly related to the intensity of sunlight reaching the earth ....
I presume that the sensor is not technically measuring UV. It is calibrated to a known light source and measures visible light and then the UV is "deduced" from that. Since the amount of UV is linearly related to visible light the UV readings it reports will be roughly correct but should anything be in the atmosphere that blocks some UV without blocking visible light to the same degree, at the time a reading is taken, then the UV reading will be less correct than at other times.

The sensor will probably take a definition of what the UV spectrum is, 280 to 400nm by the looks of it, and knows how much UV in that range there normally is when the visible light levels are at X.

Since one of the purposes of the sensor is to let you know what the UV index is, which is a crude measurement, ie UV index of 1 means that the UV levels are within a range between x and y and a UV index of 2 is within range from y to z the accuracy of the sensor isn't that important. It is more a "UV indicator" rather than a UV measuring device.

Looking at the graph on page 14 of the datasheet the range of the visible photo diode is influenced by some UV if it is setup to read low light levels but when put into the high range (for outdoor measuring) it looks as though it won't be sensitive enough to see any UV. Same goes for the IR photo diode. Both diodes have a large range.

I think if you use this sensor as an indicator and not a precise measuring device you will get what you want from it. If you want to know precise UV readings then you'll need a different sensor. For comparing the UV outside from day to day as part of a weather station, and remembering the readings as roughly correct, then it is a fine sensor for that purpose. Using it for anything more scientific and I think you'll be disappointed.

....at least that is my take on it.
Last edited by primexandy on Sun Nov 02, 2014 9:54 am, edited 1 time in total.

User avatar
adafruit_support_bill
 
Posts: 88096
Joined: Sat Feb 07, 2009 10:11 am

Re: Comparing UV sensors?

Post by adafruit_support_bill »

I presume that the sensor is not technically measuring UV. It is calibrated to a known light source and measures visible light and then the UV is "deduced" from that.
After reading through the spec sheet again, that is what I think is going on there too. There is nothing in the spec sheet that mentions any actual sensitivity in the UV range.
I think if you use this sensor as an indicator and not a precise measuring device you will get what you want from it. If you want to know precise UV readings then you'll need a different sensor.
Agreed.

User avatar
egutting
 
Posts: 297
Joined: Wed Nov 14, 2012 12:57 am

Re: Comparing UV sensors?

Post by egutting »

I've not been happy with the SI1145. I ordered both it and the GUVA sensor to do some UV level sensing. For my weather station I just need a UV index reading and the GUVA is just so much easier to deal with than the SI1145. (Plus the SI1145 didn't want to respond when placed at the end of 20 ft of cable [even with 1k pull ups on the SDA & SCL lines on the Raspi])

User avatar
primexandy
 
Posts: 89
Joined: Sat Sep 06, 2014 11:09 am

Re: Comparing UV sensors?

Post by primexandy »

You can get the SI1145 to work over that distance. You need to choose the cable correctly for one thing. The bigger the cable the better and multi stranded if possible. I used twin and earth mains cable as thin cable just wouldn't do it. The other problem you may have had was drawing too much current via the 3.3v supply from the PI. I took my power from the 5v supply and divided it down using 2 resistors. I have mine over a similar distance.

The analogue UV sensor you are using, if it worked over that range, may be giving inaccurate readings due to voltage loss over the cable even though it looks like it is working. Depending on the sensor and how it works you may be able to compensate for voltage drop via programming but you'd need to measure the voltage drop.

While comparing any sensor with another it's always going to be a "horses for courses" issue but in general a digital output sensor would be a better bet against an analogue output sensor over long cables when voltage drop is happening. If better cables don't work you should increase the power supply to make sure the correct voltage gets to the sensor and perhaps buffer the sensor's output either to allow higher current draw or to amplify to output signal but over 20ft with cheap cables I think it would be easier to get more accurate results using a digital sensor.

Taking voltage drop into account I would have thought the analogue sensor, in lower light levels anyway, even though it measures UV better, would also be nothing more than a UV indicator and not a precise measuring device.

If you are wanting to compare devices, unless you go into the science in much greater depth, every sensor within a cheap system design will have issues making it a rough indicator guide....which is still perfect for a weather station but not for serious scientific measurement of UV. I think it boils down, for most people, what is easier, most cost effective and fun for you own purposes. If you are going to worry about the spectrum response curves and accuracy of devices I think you need to put the same effort into overall system design including cables and temperature compensation and physical location. If you want to compare your readings with official readings then the angle of your sensor, and it's lens, and how it is positioned in relation to the sun makes a huge difference. The material used to weather proof your sensor also makes a huge difference......it starts to get very complicated if you want scientific results rather than a UV indication....something beyond me :)

User avatar
egutting
 
Posts: 297
Joined: Wed Nov 14, 2012 12:57 am

Re: Comparing UV sensors?

Post by egutting »

I'm using Cat5e cable to make the run. It works fine for the other items connected to the end (TMP007 sensor and Arduino Pro Mini). I don't have an issue with voltage loss with the analog sensor as it can be connected to the pro mini (a length of about 6 inches)

I wasn't saying the sensor is bad, I was saying that it didn't work for me. It's much easier to get a quick analog reading from the GUVA sensor through the pro mini than it is to adjust current going down the cable and making sure my I2C commands are correct.

User avatar
ibucky
 
Posts: 112
Joined: Wed Jan 11, 2012 10:40 pm

Re: Comparing UV sensors?

Post by ibucky »

Thank you for the replies. Indicator vs pure scientific, indicator will work for me. I've setup the SI1145 and GUVA.

The SI1145 gives a sample 'indicator' of the actual UV Index. Not any nm values, it outputs the UV index value in x.xxx so far we've seen 0.2 to 38. Amazing how strong some flashlights are and/or dangerously high on the UV Index.

Anyway regarding the GUVA, perhaps I might be missing something. Using the following code I get the voltage output. Can I calculate the nm values and not UV Index? Am I missing some math or this sensor's voltage output is another UV Index value and cannot be represented in nm values?

Code: Select all

 
const int UVSensor = A0; // the GUVA is connected to analog pin 0

int UVsensorReading = 0;      // variable to store the value read from the sensor pin
                            // Start at 0

void setup() {
 Serial.begin(9600);       // use the serial port
}

void loop() {
  // read the sensor and store it in the variable UVsensorReading:
  UVsensorReading = analogRead(UVSensor);    
  
    Serial.print("UV: ");         
    Serial.println(UVsensorReading);
  
  delay(100);  // delay 
}

User avatar
egutting
 
Posts: 297
Joined: Wed Nov 14, 2012 12:57 am

Re: Comparing UV sensors?

Post by egutting »

The GUVA calculates UV index. I believe this is the correct formula on an arduino (connected to Analog 0 and using 3.3v):

Code: Select all

sendVal = (analogRead(0) * (3.3/1023))/0.1;

User avatar
primexandy
 
Posts: 89
Joined: Sat Sep 06, 2014 11:09 am

Re: Comparing UV sensors?

Post by primexandy »

To work out what reading you are getting from specific frequencies you'd want something like "http://en.wikipedia.org/wiki/Fast_Fourier_transform" but whether you can do it on those type of frequencies I don't know. Out of my depth.

... in general you get a UV reading (as egutting's formula) and can only deduce a good guess as to what intensity you get from each frequency based upon the frequency response of the graph. If all you want is an "indication" level of accuracy for each frequency then you could make a lookup table based upon the response graph of the photo diode and do some simple maths....not sure what you'd gain though apart from saying (something like) the intensity at 300nm is 70% compared with 350nm etc. What you would be doing is scaling the graph on page 2 of the datasheet and just changing the Y-values of that graph.

I get the impression that you want something like an audio graphic bar graph type display / spectrum analyser thing.... if so, best of luck with that :)

User avatar
ibucky
 
Posts: 112
Joined: Wed Jan 11, 2012 10:40 pm

Re: Comparing UV sensors?

Post by ibucky »

Actually I was just looking for a simple log of nm samples taken over time that can be charted. A monitoring chart if you will. The light source maintained a range from x nm to x nm over some period of time.

Fast Fourier transform is a good bit beyond me to create, but thanks for the thought.

egutting's formula will give UV index value (thank you). I was initially wondering about the results from a SI1145 vs GUVA sensor, what they each output. From what I've understood they both output UV Index. Although I guess there is someway to calculate the nm from the GUVA sensor.

Question: Would the following formula calculate the UV Index for a GUVA sensor with 5v

Code: Select all

sendVal = (analogRead(0) * (4.3/1023))/0.1;

User avatar
egutting
 
Posts: 297
Joined: Wed Nov 14, 2012 12:57 am

Re: Comparing UV sensors?

Post by egutting »

Why 4.3 and not 5? That number would be the reference voltage. Normally for 5v it would be a 5 there.

User avatar
adafruit_support_bill
 
Posts: 88096
Joined: Sat Feb 07, 2009 10:11 am

Re: Comparing UV sensors?

Post by adafruit_support_bill »

None of these sensors are directly measuring the distribution of UV wavelengths. To do that you would need either a whole lot of sensors with narrow-band filters, or a spectrometer with a sensor that was sensitive to the whole range of interest.

All of these sensors are measuring some part of the light spectrum (which may-or-may-not include UV) and extrapolating the UV distribution from that. Sunlight behaves much like an ideal blackbody emitter with some filtering by the atmosphere. Since the wavelength distribution of sunlight is pretty well defined, given the measured intensity of one part of the spectrum, you can make a reasonable estimate of what the rest of the spectrum (including the UV range) looks like.

Locked
Please be positive and constructive with your questions and comments.

Return to “Other Products from Adafruit”