HPGe drift maps

[tdixon97]

There seems to be an issue with the HPGe drift time maps:

• some points with very large dt in the bulk
• NaN dt for points close to bottom surface
  Any ideas @hervasa2 ?
  

[hervasa2]

Lets huddle with Luigi, he modified my code a bit. Do you mean low dt in the bulk? I see 4 points at the top which seem to have issues. At the moment I extract drift time from the actual length of the drift paths. This is subject to drift path convergence, something that was carefully vetted. However, I would expect if there is an issue with convergence, there should be dt values which are abnormally large and probably equal to the set max drift time (I think I set 10000ns). To not depend on drift path convergence one can always calculate $dt = T_{99.9999} - T_0$ from the waveforms but this would be more computationally expensive. As for the NaNs I think Luigi assigns NaNs to points outside the bulk. Just by the shape of the image it looks like those points are in the n+ surface

[tdixon97]

Sorry the points being 0 is an artefact of how I plotted it, they are actually 10 us (the maximum drift time), this just messed up the color scale so I set them instead to 0.

[hervasa2]

Ok that make sense now. Those drift paths are not converging, this should not happen. We need to look at those drift paths in detail. Are you using a relatively new version of SSD (>v0.8)?

[tdixon97]

I am using the most recent version.

[hervasa2]

Could you please send me your ssd config file (or is this V99999A?) and code that generated the dt map?

[tdixon97]

Regarding the bottom surface. I checked and the points at $z=eps()$ that should be just inside, have:

> p in sim.detector.contacts
> True

How thick is the contact in SSD? I guess we have to replace $z = eps()$ with $z = eps()+thickness$

[tdixon97]

It is V99000A

[hervasa2]

dl_thickness_in_mm = haskey(meta.characterization.manufacturer, :dl_thickness_in_mm) ? 
meta.characterization.manufacturer.dl_thickness_in_mm : 0
li_thickness =  dl_thickness_in_mm
pp_thickness = 0.1

[tdixon97]

Ok, so in this case li_thickness is probably 0 but maybe it has some size +/- eps() ?

[hervasa2]

eps() is not good enough. We need to use csg_default_tol(T) from SSD. I am modifying make_dt_map.jl. Also removing unnecessary stuff and checking the drift convergence

[tdixon97]

I am not sure if we want to force the li_thickness to be 0 ?

[tdixon97]

I am a bit worried about the points on the tapered surface too, maybe we need to check its working ok there?

[hervasa2]

Yeah I'll change the default value of li_thickness in ext/LegendDataManagementPlotsExt.jl. As for the geometry checks this is done really well in SSD. Note that in uses csg_default_tol(T)

in_idx = findall(x -> x in sim.detector && !in(x, sim.detector.contacts), spawn_positions)

[tdixon97]

For example imagine points like this, outside the dt will be NaN (set to 0 by reboost) and inside will have some (reasonable) not NaN value. This will give strange results for any energy deposit in this square.

[hervasa2]

Ah you mean a check that the dt heuristic is working well, I though you meant the drift maps. Yes there for sure there needs to be some check there.