Flux versus Temperature from the Blackbody Source with and without the ZnSe Collimating Lens

We measured the flux off the blackbody in 4 wavenumber settings (587cm-1, 731cm-1, 1640cm-1 & 2142cm-1), using a range of bb temperature set points. We compare the results of a) blackbody with ZnSe lens installed, b) looking at the bare blackbody source, c) looking at the bare bb source with some misalignment so that we see some warm bolt heads and washers. The pupil images of a, b, and c shown below.
The short conclusion to this post is that all past flux calibration which relies on the BB_TEMP keyword to calculate the blackbody's flux to EXES is in gross error. However, the actual fluxes should be recoverable for all past observations.

Pupil images of a) Blackbody with ZnSe lens, b) bare blackbody, aligned, c) bare blackbody, mis-aligned

We looked at the wavenumbers 587cm-1, 731cm-1, 1640cm-1 and 2142cm-1. For each setting, we chose an order, slit and OTPAT and only changed the blackbody set point, using values of 260, 270, 280, 290, 296, 300, 310 K. Then we choose a box of pixels on well illuminated parts of the 2-D spectrum and measured the median count value.

High-medium setups used to compare blackbody counts, 587cm-1,731cm-1,1640cm-1,2142cm-1. It shouldn't matter what slit, mode or OTPAT we used, as long as we used the same config for each set of measurements at different temperatures.

If the bb is operating ideally, the relative values of this median count value at different temperatures will follow a Planck function. We calculated the predicted count values with the mplanck.pro function. Then we scaled the counts for all configurations to be equal to 1.0 at T=296 K.

The scaled flux v temp curves were all shallower than the expected counts. If we assume that this is due to room temperature scattered light (i.e. T=296 K) we can see what "emissivity" value works for producing the different curves. 

The dashed lines that pass through the BB count measurements are produced by modifying the expected blackbody flux values: 

B'(T,lambda) = B(T,lambda) * (1-e) + e * B(296K, lambda)

where B is the scaled mplanck flux, e is emissivity and lambda is 1e4/wavenumber (microns) used in the mplanck function.

However, the contaminating flux is probably different for the three blackbody configurations and this treatment could be misleading. But anyhow:

No lens, aligned BB: As hinted by the Fluke thermal imager gun measurements, we pick up flux from the flat Al turning mirror on the cal source turntable. This is inevitable, since the mirror isn't perfectly reflective. From the Thorlabs site, we expect about 5% dilution with the temperature of the ambient environment. For 587cm-1, we instead see about 12% to match the trend. Time to wash the mirror? This will contribute to all blackbody configurations. 

Reflectance of the cal source BB turning mirror (Thorlabs)

 No lens, mis-aligned: 

In addition to the 5-12% addition from the turning mirror, we have to account for the fact that the area of the pupil now has warm metal parts that don't follow the blackbody setpoint. The are close to ambient temp. If we treat them like they are at 296K, we get an emissivity of e=0.23 to scale the expected Planck counts. This says that the pupil area coverage of the washers is about 11%, which seems plausible, looking at the images 

ZnSe lens: We found that we'd need to assume e=0.57 to get the observed flux v. T curve for this blackbody config. In this case, it seems that the lens may be directing warm off-axis from the hot components of the blackbody, like the lens barrel and the T-exchange Cu block, into the pupil. These components feel far warmer to the touch than room temp, so using T=296K as the sole contaminant temperature is probably not accurate here. 

The other wavenumber settings: 

(And, I forgot that we didn't do 2142cm-1 with the ZnSe lens yet).

The excess due to the mirror is not the same for the three settings. We need to take other wavenumber setting data with the ZnSe to see what's happening. When we took blackbody data with bare aligned, bare, mis-aligned source we did: 387cm-1, 587cm-1, 731cm-1, 811cm-1, 1387cm-1, 1423cm-1, 1640cm-1, 1748cm-1, & 2142cm-1. We'll start with those settings to get an idea of how to explain the counts directly off the BB and ZnSe lenses. 

Other questions: 

What is the repeatability of the calwheel to the bare BB source? 

Does the entrance window need to be considered?

Is water vapor between the blackbody and window skewing the lab data? Can we purge the blackbody cavity to check?

Does cleaning the mirror flat make a difference?

When the collimating lens is removed, is the warm barrel and large copper block a source of contamination?

Does the lower ambient temperature have a big effect on the in-flight blackbody counts?