1. Mircomegas - Gas gain simulation studies -

2. Goals of the gas gain studies
Comparison of amplification-factors with experimental Data
Measurements will be done in Jan/Feb 2013 at the LMU Munich for non
resistive chambers (resistive chambers depend on availability)
Voltage Scan from 480V to ? 550V ?
Upper boarder of the voltage scan depends on CPU capaticity, and efficiency of the code, finals value not jet settled
Dependence on:
temperature (Simulation & Experimental Data available)
pressure ( ‘’ )
Layout of the simulated mesh
Test of a CPU optimized algorithm (p. 6)
Avalanche profiles (ne/z; x-y distribution /z)
26. Nov 2012
Goals of the gas gain studies

3. Simulating a MM mesh
Woven structure of the mesh causes increase in CPU time, due to:
More complex meshing in 3D models using FEM (ANSYS)
More and smaller elements using neBEM (Garfield)
No analytic calculation of the field possible
Question: Is that effort necessary?
Or: Do more simple layouts result in the same distribution?
Microscope-photo of a mesh
26. Nov 2012
Simulating a micromegas mesh

4. Simulating a MM mesh Properties to simulate: periodic mesh pitch
cylindrical wires
woven mesh structure
mesh flattening (flattened tops, compressed contacts) X
Microscope-photo of a mesh
planar approximation
flat - cubic elements
flat - cylindrical elements
woven - torodial elements
26. Nov 2012
Simulating a micromegas mesh

5. Simulating a MM mesh Contours of the potential flat - cubic elements
flat - cylindrical elements
Contours of the potential
woven - torodial y =0
Mesh:
upper edge
lower edge
Contours of the potential
woven - torodial y =pitch/2
26. Nov 2012
Simulating a micromegas mesh

6. Idea for a CPU optimized simulation
Divide the avalanche simulation in two steps, by insertion of a “virtual” ~ half distance:
1.) - between mesh & virtual plane
- inhomogeneous field due to mesh geometrie
- crucial part of the avalanche formation with a
smaller number of electrons (< 103)
microscopic avalanche calculation in Garfield
2.) - between virtual plane & readout strips
- highly statistical part ( ) in an approx.
homogeneous field
using a parameterized function for each
electron at the plane, assuming a uniform field 1 2
26. Nov 2012
CPU optimized simulation

7. Idea for a CPU optimized simulation
Open questions concerning in this method:
Where to insert the “virtual” plane?
(trade-off: enough electrons after step 1 to obliterate
the statistic fluctuation in step 2 vs. CPU cost)
- Minimum number of electrons in the avalanche?
(= minimum voltage)
- How severe is the effect of ignoring space
charge effects in step 2?
Method will be tested in a mid range voltage and
if successful used to simulate higher voltages. 1 2
26. Nov 2012
CPU optimized simulation

8. Open questions Other parameters of interest?
Which dependences to study? (temperature, pressure, gas mixture?)
Range of the voltage scan?
(depends on success or not with CPU saving method)
How to take resistive strips into account?
- full simulation in Garfield++
- Using signal from non-resistive simulation + LTSpice
26. Nov 2012
Open questions