Low values (e.g. 2-5) are suitable for mixtures than contain substantial amounts of quenchers. Nearly pure noble gases call for larger values (e.g. 10-20) while very large values (100 and more) are needed for pure noble gases. But in the latter case, it is preferable to use Magboltz\ 1 (see ANALYTIC-INTEGRATION).
The statistical accuracy of the drift velocity calculation improves with the square root of this parameter, while the CPU time consumption increases linearly. Since the accuracy depends on the gas mixture and on the field, it may be worthwhile in critical applications to estimate the statistical error of the Magboltz calculations for a given gas and a given field by repeatedly calculating the transport parameters:
Call book_histogram(vhist,50) Call book_histogram(dlhist,50) Call book_histogram(dthist,50) Global nrndm=200 For irndm From 1 To nrndm Do Say "Iteration {irndm}/{nrndm}" &GAS magboltz argon 90 methane 10 coll 5 e-field 125&MAIN Call drift_velocity(125,0,0,v) Call fill_histogram(vhist,v) Call transverse_diffusion(125,0,0,dt) Call fill_histogram(dthist,10000*dt) Call longitudinal_diffusion(125,0,0,dl) Call fill_histogram(dlhist,10000*dl) Enddo
Call plot_histogram(vhist,`v<SUB>D</SUB> [cm/microsec]`,`Drift velocity`) Call plot_end Call plot_histogram(dthist,`σ<SUB>T</SUB> [micron for 1 cm]`, ... `Transverse diffusion`) Call plot_end Call plot_histogram(dlhist,`σ<SUB>L</SUB> [micron for 1 cm]`, ... `Longitudinal diffusion`) Call plot_end
From the histograms, one concludes that, for argon 90\ % methane 10\ % and E=125\ V/cm, COLLISION=5 leads to a statistical error of 0.4\ % on the drift velocity, 4.5\ % on the transverse diffusion and 7\ % on the longitudinal diffusion.
[By default: 10\ \×\ 960.000 collisions.]
Formatted on 21/01/18 at 16:55.