Add non-linear correction?

[caternuson]

See this App Note for details:
https://www.vishay.com/docs/84323/designingveml7700.pdf

It details the non-linear correction for higher lux conditions. There is also an automatic adjustment method (see Fig. 24) that could be implemented.

This has been done for the Arduino library, which may help as a reference:
https://github.com/adafruit/Adafruit_VEML7700

[marcusds]

Is this the relevant code?

float Adafruit_VEML7700::computeLux(uint16_t rawALS, bool corrected) {
  float lux = getResolution() * rawALS;
  if (corrected)
    lux = (((6.0135e-13 * lux - 9.3924e-9) * lux + 8.1488e-5) * lux + 1.0023) *
          lux;
  return lux;
}

https://github.com/adafruit/Adafruit_VEML7700/blob/master/Adafruit_VEML7700.cpp#L358

Edit:
Am I missing something, or is adding non-linear correction as easy as:

import time
import board
import busio
import adafruit_veml7700

i2c = busio.I2C(board.SCL, board.SDA)

veml7700 = adafruit_veml7700.VEML7700(i2c)

veml7700.light_gain = veml7700.ALS_GAIN_1
veml7700.light_integration_time = veml7700.ALS_100MS

def computeLux(lux, corrected):
    if corrected:
        return (((6.0135e-13 * lux - 9.3924e-9) * lux + 8.1488e-5) * lux + 1.0023) * lux

    return lux

def autoLux():
    gains = list(veml7700.gain_values.keys())
    intTimes = list(veml7700.integration_time_values.keys())

    gainIndex = gains.index(veml7700.light_gain)
    itIndex = intTimes.index(veml7700.light_integration_time)
    useCorrection = False

    ALS = veml7700.light

    if ALS <= 100:

        """increase first gain and then integration time as needed
        compute lux using simple linear formula"""
        while ALS <= 100 and (gainIndex < 3 or itIndex < 5):
            print("gainIndex, itIndex:", gainIndex, itIndex)

            if gainIndex < 3:
                gainIndex += 1
                veml7700.light_gain = gains[gainIndex]
                print("Increasing gain to", veml7700.light_gain)
            elif itIndex < 5:
                itIndex += 1
                veml7700.light_integration_time = intTimes[itIndex]
                print("Increasing integration time to", veml7700.light_integration_time)

    else:
        """decrease integration time as needed
        compute lux using non-linear correction"""
        useCorrection = True
        while (ALS > 10000) and (itIndex > 0):
            itIndex -= 1
            veml7700.light_integration_time = intTimes[itIndex]
            print("Lowering integration time to", veml7700.light_integration_time)

    print("Gain, it: ", veml7700.light_gain, veml7700.light_integration_time)

    if useCorrection:
        adjustedLux = computeLux(veml7700.lux, True)
        print("Lux (standard, corrected, raw):", veml7700.lux, adjustedLux, ALS)
        return adjustedLux
    else:
        print("Lux (standard, raw):", veml7700.lux, ALS)
        return veml7700.lux


while True:
    autoLux()
    """print(veml7700.lux, veml7700.light)"""
    time.sleep(1)

I have no way of knowing if these numbers are correct. Sometimes my adjusted lux is over 900,000, which doesn't make sense since it should only ever max out at 120,000? Edit 3: Sample code is working better, now it is making out raw value at 65535, that is with the my phone flash directly on the the sensor, which I would think should be higher? I'm not sure. (oh that not why it's being limited to 65535...)

If we can get this sample code working, shouldn't be hard to get it into the library. But I am not capable of that myself.

[caternuson]

That looks essentially correct. The "non-linear correction" is simply the extra math in computeLux(). The "automatic adjustment method" is all that looping/logic/adjustment stuff in autoLux().

[FoamyGuy]

Resolved by #29