Calculating expected data from Spectral sensor AS7265x

I recently bought the Sparkfun Triad Spectroscopy Sensor - AS7265x. I am testing out the possibilities with the sensors on this chip. I noticed the provided datasheets for the light sources, which got me interested in calculating the expected values. Here I ran into a few problems and I hope someone can help me with them.

NIR and UV light source intensities are stated in mW/sr which is very helpful, although I presume the distance is somewhat of importance. I’ve found the formula for the solid angle to help determine the expected radiant intensity on a distance, because the formula for radiant intensity is I = P/A which in terms of units is [W/m^2] = [W] / [m^2]. Although with the IR and UV led the radiant intensity is given in mW/sr. Therefore I multiplied the [mW/sr] with [sr] and replaced [W] and replaced [m^2] with [cm^2] in order to obtain [mW/cm^2] which I can work with. The steps surely must sound vague, so ask away for clarification.

By using this formula and the given information I was able to calculate the expected values. I took in account the amps and wavelength for a different intensity, but my values still are lower then the device gives me. For example on the IR led on 860nm I have calculated 195 uW/cm^2 and my results give back roughly 19200 uW/cm^2. My measuring setup is not nearly perfect, but these values are extremely different. However on 900nm I should have lower values (because of the lower intensities the datasheet describe) and the results are roughly 1950 uW/cm^2. I wonder if I have messed up my calculations or if the sensor does something very unpredictable. It seems consistent in measurements, but it responds so much heavier to UV and IR compared to visible light.

Also, I’ve noticed the datasheet says the sensors are calibrated to reflective surfaces. Does that mean it is impossible to get accurate results with direct exposure to light? (I bought the LEDs and installed them a certain distance from the sensor.)

I’d be super helpful if anyone could give me more answers. It would also be okay to point me into the right direction. Feel free to ask for more clarification.

Update 1

I’ve noticed that the gain was set to 64X by sensor.begin(), which results into really large numbers of data. When I set this to 1X it seems to have dragged the 19200 to 200 uW/cm^2, which comes really close to my calculated estimate. Although this is the 860nm and when I compare it with the 900nm the results are still off… low but we’re talking a drop to 66 uW/cm^2. Which according to the schematics should only have 80% intensity compared to the 90% intensity on 860nm.

There is not enough information in your post to understand what you are doing and what calculations you have done. Please post a drawing of the setup, with distances indicated.

Are you pointing an LED directly at the detector, and hoping to make an accurate estimate of something like the LED emission power? That is basically what the manufacturer did to arrive at the sensitivity values, although some sort of diffuser was used.

  1. Calibration and measurements are made using diffused light

  2. Each channel is tested with GAIN = 16x, Integration Time (INT_T) = 166ms and VDD = VDD1 = VDD2 = 3.3V, TAMB=25°C

  3. The accuracy of the channel counts/μW/cm2 is ±12%

  4. The light source is either a 5700K white LED with an irradiance of ~600μW/cm2, or an incandescent light with an irradiance of ~1500 μW/cm2 (300-1000nm). The energy at each channel (V, B, G, Y, O, R) is calculated with a ±40nm bandwidth around the center wavelengths (450nm, 500nm, 550nm, 570nm, 600nm, 650nm).

In practice you can trust the manufacturer’s sensitivity values to be reasonably accurate. I’ve found that if my values disagree with my expectations (by more than, say, 20%), I’ve made a mistake in either my thinking, my assumptions or my calculations.

It sounds like calibration issues might be the problem. Ensure your sensor is correctly calibrated for direct exposure and check for potential discrepancies in your distance and angle calculations.