For lamp-controlled exposures, please only check/measure the following exposures:

• exposures with single lamps.
• the first exposure in any set with duplicate>1

The second problem has been fixed and the first ameliorated, but there is very little data after the fixes.

I have added an image showing evolution of the flux from different lamps.

Halogen lamp keeps relatively constant flux with time. New exposures after shutter mitigation show quite a lot of scatter, but that might be due to exposure time control issues cloomis? 

Mercury Argon is actually brighter now than at the beginning of the year. On June 2, when stability sequencing started, that lamp is much much brighter, and after exposure 18628 drops like a stone. Could this be due to the stability sequence doing something interesting +  and drop after Argon lamp being installed?

We were not taking Neon data all the data. The only thing that I can confirm is that the lamp is much dimmer than at the beginning of the year. We need around 50 sec now to get fluxes comparable to what we got at the beginning of the year in 3 sec. 

The flats are 3s. I think we now have the exposure times controlled about as well as we can with the existing PDU, but before that work I could certainly see up to 1s "jitter".

cloomis Do you have an idea why the exposures were so much brighter when you started stability sequences? Is there something there that could be suspicious?

Julien correctly pointed out that the time stamp on the images are quite confusing. The last bright exposure (18628) entry in the operational database is: 

expose arc exptime=3.0 duplicate=4 switchOn=hgar switchOff=hgar name=stabilityTest,18628,446,arc,2020-06-02 23:26:05.477073,,,"",stabilityTest,complete,arcs,1,b,1
expose arc exptime=3.0 duplicate=4 switchOn=hgar switchOff=hgar name=stabilityTest,18628,446,arc,2020-06-02 23:26:05.477073,,,"",stabilityTest,complete,arcs,1,r,2

My understanding is that this corresponds to 1:26 pm on June 2.