FittingServiceFunnies

AnalysisService

[5PG offset] Five point gain problems... (under-constrained?)

From Chris

Wiggles in trim steps, possibly caused by trim scans being to restrictive on the threhold range scanned.

From Matt

Mail from Matt - not investigated yet.

[Fits] [Raw]

Date: Thu, 07 Oct 2004 14:26:51 +0100
From: Matt Palmer <SECRET>
To: Peter Phillips <AGENT>, Alan Barr <CLUB>

Hi,

I've just discovered something about the Fitting Service that you may
want to look at.  I've attached a couple of plots taken from module 278
(you can see the SCTDAQ report here:
http://www.hep.phy.cam.ac.uk/~silicon/d07/20220330200278/Module_Reception/index.html)

As you can see from the raw plot (I zoomed in a bit), something funny
happens in the first chip of link 1.  It appears that the occupancy
drops (but not to 0) for some bins.  Looking at the fits, there are 2
examples.  The one on the left managed to ignore the low bin and did a
reasonable fit (compare this with SCTDAQ which gave a mean of 55 -
somewhat low) but the channel on the right was poorly fitted (SCTDAQ
gave a similar answer for this channel).  I guess the fitter gets stuck
in a local minimum.  Neither SCTDAQ nor SctRodDaq reported any errors
for these.

The question is, what happened here?  Do we have any idea?  Could it be
related to the low injected charge (it affected the 0.5, 0.75  and 1.0fC
points tho it got better as the charge got higher)?
If it is bad, should the fitting service catch this sort of thing?

BTW, this doesn't happen much - its the only one of nearly 200 modules
I've found...