|Re: Raw data read and write [message #3839 is a reply to message #3802]
||Wed, 06 December 2006 10:47
Registered: November 2006
Concerning raw data simulation, I produced one date file
with 18 DDLs data from TRD in and Christian tried to analyze
this data using DATE. He fuond that baseline is not properly
set and it is zero.
In TRAP chip, normally ADC value itself has few ten to
few hundred baseline, and it is changing acording to
time by very low frequency time drift. This noise is
filtered by digital filter (pedestal filter). ADC value
should be unsigned integer of 10 bits dynamic range.
If we subtract the baseline, we simply cut off all ADC
values below the baseline. To avoid this problem,
artificial baseline is again added after pedestal
subtraction which is nominally 10 (but can be changed
On the other hand in AliROOT, additional baseline is not
added. Baseline is simply subtracted and negative value
is simply cut. This is of course not good because noise
appeared only on positive ADC value. Effetively we change
the baseilne little bit above if we define the baseline
as average signal.
I think we must do like TRAP is doing which is:
Define some ADC intrinsic baseline (define randomly).
Apply baseline subtraction as TRAP is doing.
Add artificial baseline (10, as TRAP is doint).
Then digits value will fluctuate around 10, and it presents
even negative value (below 10) and cross talk as well.
I think this gives impact to reconstruction part.
Somone has to change reconstruction part to take into
account the baseline which is not ADC intrinsic baseline
but artificial baseline added by TRAp.
How do you think?