These data were then compared to the response of the same towers / PMT's, operating at the same HV, when pulsed with the green LED the previous week by Renee.
The correlation between the LED peak channel (pedestal
subtracted) and the crudely estimated laser pulse height is surprisingly
good. On the plot, I've thrown in `error bars' of 5% for the LED values,
and 7% for the laser. The errors don't represent true uncertainties -
the LED peak position is determined much better than 5% - but are shown to
suggest that tower-to-tower variations of this order are all that is
needed to establish a very tight correlation. Unless this sub-sector
represents an anomalously `good' set of LED fibers and laser scribes, this
would indicate that both the LED and the laser are injecting
light more uniformly than had been expected.
Moreover, the correlation shouldn't be perfect even under ideal conditions. While the LED peak should show no eta dependence, the laser injection scheme was designed to increase by about 30% with increasing tower number. This is seen in the plot below |
This is the same data, now presented as the ratio of laser pulse height to LED peak channel, for each tower (tower 9 was not working for either data set). An error bar of 10% has been added for each point - again, this was arbitrary, but it is the level of tower-to-tower variation required to yield a chi-squared close to one. A linear least-squares fit shows about a 22% increase with tower number, not far from the 30% measured for some individual megatiles early on in the process. |
Conclusions