Things are somewhat better today. It's still at the somewhat frustrating stage, but it has improved past the "please make it stop" phase that we spent much of yesterday in.

The day shift had little luck it would seem. When we left we requested a new sample because we'd found sufficient evidence that the current sample was trash. Once they installed the new sample they had trouble getting the geometry back on the experiment, basically only restoring the experiment by about 8pm. A rough day indeed.

We've been doing something a bit dangerous today, at least in my opinion. No, not physically dangerous..... Instead of taking the data in the way we've done in previous experiments, we're using a new CCD detector. While this will certainly give, in the long run, better quality data, I'm not convinced that we're taking an unnecessary risk by using it right now. We did not use the new camera very much and we're not set up to easily do the data reduction and analysis for it yet. Necessity is often the mother of invention and so we're working hard to get the routines moved over to the new detector. The necessity in this case is so that we can interpret our results while we take data. Without it we're flying blind.

So long as we've got a grasp on what the data is telling us, we are usually more sensitive to mistakes and interesting twists. If we're unable to read the results while the experiment is in progress, then we've got to just proceed and hope everything is working as intended.

For instance, the image at the top of this post is one of our scattering peaks. This IS a picture in reciprocalspace, though I'll have to explain that in a later place. In the center is a blob/blur that is an example of the scattering we observe. The intensity, shape, and position all tell us important information. Unfortunately it is often difficult to find such a nice looking peak. This is especially true when the peak is smaller than any background noise.

We though this was indeed the case for our set up today. The data looks very, very noisy and the scattering peak doesn't look very strong compared to the background. Very often with counting type experiments (and our detectors simply just do that, count photons), you have a statistical random variation. The simplest way to get around this is to simply count longer.

Our images appeared to have just this kind of noise in them, but after a few attempts at counting longer, nothing really improved. The signal did get stronger, but the background also appeared to get stronger. So I pulled up two different images from very different places and put them side by side. Zoomed in you get something like the picture below. Indeed, it looks like noise.




but what I hope you can notice is that the two pictures are almost identical. The noise is almost perfectly correlated even though the sample and the experimental geometry have both changed quite a bit between the two. The only thing that is the same is the exposure time....

So what you see is that our exposures are dominated by a "dark current" in the detector. The longer you wait, the more current is accumulated in each pixel. It's not really "bad" news because it means that (so long as things behave) we can simply subtract some "background" images and the result should be a nice exposure. And that's what you get at the top of the post(though not from subtracting the two frames in the second image).

Anyhow, all this is fine, but I sure wish we were not figuring it out while the data collection is in progress. Ideally we would have known this coming in. But then again, it's better that we know it now rather than figuring it out much later after the run is over. I suppose part of it is that I'm just a little concerned about changing the "data pipeline" when we already had it flowing and just needed a little more to finish the experiments we'd been doing. But perhaps this will improve our data sufficiently and ease things in the future.

We're taking reasonably good data (we think) at this point. Everything seems to be working. We are a little concerned that the sample may be trashed again, but that's only a mild suspicion at the moment. We'll know soon enough. Very, very soon we'll begin to illuminate our sample with UV radiation and the effects should be quite clear.