Hi,
Measurement of low light levels is probably more significant to us here in the UK than in Portugal or Australia.
Presumably the source resistance (18 ohms) is built into the sensor and I would be the last to suggest that you should "hack" a precision, calibrated sensor (even if 15 years old). So the options are a little limited:
The AD822 Op Amp certainly has a "better" Input Offset Voltage than many "ordinary" Op-Amps, but even the "nominal" tolerance of 100 uV represents around 1% (or 10 W/m2) of the Didcot's maximum (UK) level and the "worst case" is 800 uV, or 80 W/m2! The maximum (noon) level in mid-Winter down here is less than 200 W/m2 (onto a horizontal surface) and I guess may be approaching 100 W/m2 in Scotland?
There are Op-Amps with lower Input Offset tolerances (probably using bipolar rather than JFET inputs, but then you'd need to look at the input current specifications) and chips with additional "input offset balance" pins, but on the KISS principle I suggest the following:
Add a potentiometer that can "sit up" the voltage across the sensor (i.e. on the Op-Amp input pin) by up to 1.8 mV. That needs 0.1 mA into 18 ohms, or a 47k resistor to the wiper of a pot (any value up to say 10k) wired across the 5 volt rail to Earth. 1 mV (nominal) will then increase the Op-Amp Output Voltage (to the ADC) to around 55 mV, which also overcomes another potential issue that a "rail to rail" Op-Amp will normally not be able to pull accurately all the way down to true zero (milli-) Volts.
Then you will need to modify the software to subtract about 55 mV (I make that a value of 50 with a 10-bit ADC and 1.1 volt reference) and finally adjust this new "balance" pot so that "zero" light level is just reached when the sensor is in darkness.
Cheers, Alan.
Measurement of low light levels is probably more significant to us here in the UK than in Portugal or Australia.
Presumably the source resistance (18 ohms) is built into the sensor and I would be the last to suggest that you should "hack" a precision, calibrated sensor (even if 15 years old). So the options are a little limited:
The AD822 Op Amp certainly has a "better" Input Offset Voltage than many "ordinary" Op-Amps, but even the "nominal" tolerance of 100 uV represents around 1% (or 10 W/m2) of the Didcot's maximum (UK) level and the "worst case" is 800 uV, or 80 W/m2! The maximum (noon) level in mid-Winter down here is less than 200 W/m2 (onto a horizontal surface) and I guess may be approaching 100 W/m2 in Scotland?
There are Op-Amps with lower Input Offset tolerances (probably using bipolar rather than JFET inputs, but then you'd need to look at the input current specifications) and chips with additional "input offset balance" pins, but on the KISS principle I suggest the following:
Add a potentiometer that can "sit up" the voltage across the sensor (i.e. on the Op-Amp input pin) by up to 1.8 mV. That needs 0.1 mA into 18 ohms, or a 47k resistor to the wiper of a pot (any value up to say 10k) wired across the 5 volt rail to Earth. 1 mV (nominal) will then increase the Op-Amp Output Voltage (to the ADC) to around 55 mV, which also overcomes another potential issue that a "rail to rail" Op-Amp will normally not be able to pull accurately all the way down to true zero (milli-) Volts.
Then you will need to modify the software to subtract about 55 mV (I make that a value of 50 with a 10-bit ADC and 1.1 volt reference) and finally adjust this new "balance" pot so that "zero" light level is just reached when the sensor is in darkness.
Cheers, Alan.