Last week I tried exporting some of the blend filters mods that I had developed at ETMX, but found that at least one of the blends didn't work. Today I worked on figuring out why.  I started by doing an ugly hack of Jeff's plot_current_blends script so I could look at the original LLO  blends in complementary form. When I used the hack to look at what I had installed, it became immediately clear why my blend didn't work. In the attached image, I show the complementary version of the "bad" blend (solid lines) and a "good" blend. The good blend is what is currently running on st2 on all of the BSC's in X,Y,Z and RZ. It's a modification of a 250 mhz blend that we stole from LLO (they run a similar blend). The bad blend is a modified version of the 01_28 blend we use on the HAM ISI's. You can see that at .4 hz I managed to introduce a notch ( in the solid red line) that did all kinds of band things to the plant.

This happened because I was designing in foton (blind), and couldn't see the complementarity of the blends from there. I thought I was safe because I was making small adjustments and was trying minimize any phase wobbles around the blend frequency. Turns out that is not enough. I think fixing the complementarity will be relatively simple,  the first image in my alog  17790 shows why I would like to get this to work, roughly an order of magnitude reduction at 1 hz, with some low frequency injection.

The amazing part is this blend runs on ITMY and ETMX.

The 2 attached pdf's are the original blends (1st pdf) and modified blends (2nd pdf) for both stages X, Z, RY, RZ , shown in both installed and complementary versions.

The successful mods I made are adding elliptical filters to the St1 90 mhz Z blend (alog 17702), a more effective elliptical to the St2 250 blends (alog 17488) and an elliptical to 750mhz St1 RZ cps blend. I haven't found the to do much good, but it doesn't do any harm that I've seen.