Vlad, Niko
Continuing from aLog:

Resolved Noise issue of [Exp3, previous aLog].:
    Niko replaced the incorrect Op.Amp, and reinstalled the related capacitor.
    We also checked and found that R2 resistor (resistor across transimpedance amplifier) on this board has been updated to a model with lower temperature dependence.
    Board performed as expected.

We repeated [Exp3, previous aLog] from the previous aLog with the fixed board:
    [Exp4.png] We observed a temperature dependence that meets the specification of the R2 resistor. The blue line is the calculated voltage change we expect from the resistor. The board has a mounted thermometer which we use to calculate this expected response based on its specifications. The resistor appears to perform slightly better than the specification.
    From this we are confident we know temperature dependence that arises from the PCB board itself - that it is due entirely to the R2 resistor, and that this about 50x less than what we see from [Normal_operation_voltage_change.png].

We go back to [Board1] and repeat test against the Gold Standard (GS) in the presence of optical light:
    [Exp5.png] We do observe the expected temperature dependant voltage difference from  [Normal_operation_voltage_change.png].

From here we continue to isolate the problematic component, given a much more limited number of variables:
    [Exp6.png] Here we heat only the black box that sits atop the integrating sphere, and repeat the experiment as above. We don't see any Voltage differences to the (not heated) GS. For this reason we can rule out the black box as being a major factor in temperature dependance of these devices.
    [Exp7.png] Here we heat only the integrating sphere and not the black box. I plot against the signal seen GS, and we once again observe the temperature dependant voltage difference.
    [Exp8.png] To confirm the result from the previous aLog, where [Exp7] was done in the absence of light, I repeat the above (heat only the integrating sphere) in the absence of laser light. The absolute voltage change is seen about a factor of two stronger than what was seen in the old experiment. Not sure why.

Conclusion for now:
We observe output changes attributable to an integrating sphere temperature dependence (which itself depends on input laser power).

More clever tests need to be designed to continue to understand how and why this works the way it does. It is not currently known why the magnitude of the response depends on whether there is incident light or not, and how the sphere's thermal state is coupling to that and the total output of the calibrators.

I've tried to read about the material in which the sphere is coated in: "Spectraflect".[Datasheet]
    The only thing that jumps out to me is the Laser damage threshold: we would be operating really quite close to that - perhaps this material is very sensitive to thermal effects related to relatively focussed incident power
    Still doesn't fully explain why we see output changes when there is no laser light incident - surely must be some thermal coupling to the PD's longer wavelegnth sensitivty region, and Vlad will attempt to model that.