Why do more noise hits show up in the 2dh than in 1dh for this event?
If you take a look at the similar work going on at LBL
(see Dieter's page and click on "Pattern Recognition" and then
where he says they tag the merged particles), you may say (as Dieter did :) "it looks like there
is less noise hits in 1dh than in 2dh.
It is true! (for this event). Looking into this, I found that a few of the unused 2dh hits could be
eliminated by prohibiting one- and two- pixel hits (safe enough!), and by fixing a (small) bug that
had to do with the interaction between the algorithm
that cleans up noise at the TPC bottom and the removal of electron sequences (see the
old description of the overall algorithm).
But there is still more noise hits in 2dh for this event.
Let's see where those unused hits are:
Here is a plot of y vs x of the unused hits in 1dh (red)
and 2dh (green). Note that:
- Both have a lot of "noise" hits right around x=0, where the beam goes through.
- This is due to delta electrons which spiral down tightly to the bottom of the TPC, and to the
subsequent saturation of the electronics. 2dh has an "electron rejection" algorithm that helps
minimize the effect of this for central Au+Au collisions, but here we are driving a Mac Truck
through the TPC!
- 1dh misses a whole track! (had to point that out... :)
- 2dh seems to be sensitive to 3 bad pads (right hand side of pic) which apparently are NOT in badpads table.
- 2dh has a lot of unused hits for all x at about y=0, where the beam goes through.
- This is almost certainly due to "wire bounce," because of a TON of charge hitting the
(not-totally-decoupled) wires all at the same time. TPC's are not made to measure 2 GeV collisions
in extreme reverse kinematics!
To see the wire bounce at a pixel level, you can set the debugging switch in 2dh that makes row-wise
pixel histograms (set/val h2d_sw.dbg_histo 1), and see for padrow 7 and 99 for example:
(again 1dh hits in red and 2dh hits in green)
The wire bounce puts unused hits at all x (but a particular y)
and makes the 2dh look much noisier when plotting x vs z, as Dieter and I did.
Here is a picture of the unused hits from 1dh (triangles) and 2dh (circles) for this
event.
Since this is a problem only with the extreme reverse kinematics data (we do not see this with the
central symmetric data), and does not have a large effect on tracking (other than
to give some extra very short tracks which are easily cut away), I do not plan to put more attention
onto it. But it is good to have some idea where it comes from, and many thanks to Dieter for pointing
it out.
Mike Lisa lisa@mps.ohio-state.edu 06 dec 1997