I'm attempting to use a SI1133 without a diffuser. I'm using the equation:
UV Index = k((m * input^2) + input)
I'm using the "5. Simplest, Basic Performance Mounting Technique" configuration mentioned in http://www.silabs.com/documents/public/application-notes/AN968-Si1133-UV-Index-Sensor-Electrical-and-Optical-Design-Guide.pdf page 9, using these settings:
k = 0.0104
m = 0.003910
HW_GAIN = 1 (48.8us)
SW_GAIN = 7 (128 samples)
Decimation Rate = 3 (512 clocks)
Post Shift = 0
When I plug the raw reading from the FORCE command into the UV Index equation, my result bounces all over the place and doesn't make sense (UV Index of 0.3-0.5 in a dark room, 30's and higher outside in the sun). I thought maybe I should be dividing the raw readings by the 128 samples and passing the result into the UV Index equation, but that results in extremely small numbers.
Could anyone help me understand how to calculate UV Index in this scenario?
Use k = 0.008284, m = -000231 instead.
Do not divide the raw data by 128.
When you're testing the sensor under direct sunlight, make sure the sensor faces upwards. I would also recommend you put a rubber ring around the sensor to avoid optical leakage to the side.
Thanks for the fast reply. Could you help me understand how you calculated those new values for k and m? My customer is experimenting with a number of optical housing configurations and will likely be tweaking measurement parameters, and I'd like to give them some guidelines.
My customer also has a requirement to be able to take a UVI reading while facing the sensor directly at the sun. Is this possible with the Si1133? Would a diffuser be required in this case?
Those number comes from our testing. Ideally, the customer has to calibrate the result by themselves. If they change the configuration, they should take multiple measurements at different UV index levels and compare the result against a UV meter to generate their own coefficients.
UV index reading is not supposed to be the done with the detector facing the sun directly. They can still do that with Si1133, but the result will be very large and won't make too much sense to users.
Thanks Tony. With a negative 'm' value as you mentioned, the output of the UV Index equation is negative in moderate light. I have a UV meter reading UVI ~0.7 in today's cloud cover.
The Si1133 with the above configuration in the same position is giving a reading of 640-700 counts, and with k = 0.008284 and m = -0.002310 per your above suggestion, it's supplying a UVI of -2.5 to -3.5. Is that m value supposed to be negative?
m is -0.0002310
Thanks Tony. That's brought the numbers closer to the UV reader output, though it's still off by a bit. I'll play with k and m.
Per your earlier comment about not pointing the Si1133 directly at the sun: At around noon at low latitudes, the sun can be pretty close to directly overhead. In fact, anywhere between 23.5 degrees north and 23.5 degrees south latitude will have the sun directly overhead at some point in the year (Hawaii being one such place). That mean that “pointed upward” is directly toward the sun in some situations.
Is the device not intended for use near noon in low latitude regions? Are there some modes of operation (e.g. turning on the HSIG bit to divide sensitivity down) in which a an accurate UVI measurement could be possible under these conditions? Does Silicon Labs offer any other parts that are more appropriate for this type of measurement?
Without the diffuser, the UV index accuracy can only be approximate +/- 2 UV index.
I think you misunderstood my comment on not pointing the sensor directly at the sun. The sensor will work at noon and in low latitude regions. I'm just saying that at any time of the day, UV index is the measurement result with the sensor facing upwards instead of facing the sun. If you're measuring the UV index in the morning, you should not place the sensor facing towards the sun on purpose.
I see, that makes sense. Thanks for your help as we get to know this part.