LAV front-end board test Silvia Martellotti, Francesco Gonnella Photo Veto Working Group 3 September 2014 Silvia Martellotti - Laboratori Nazionali di Frascati
LAV front-end 9U board Inputs Type Outputs Power connection VME J1 16ch IN Analog 32ch OUT LVDS CANopen or USB IN OUT VME custom power ToT mezzanine card (16 x) 2 channels and 2 threshold per board. Includes the circuits for: Clamp, Amplifier, Comparator & LVDS driver. Sum mezzanine card (16 x) Sum 4 analog channels (8 x) Sum 16 analog channels (2 x) Connected to a 50W LEMO out Test pulse controller (1 x) Can set for each channel: Pulse height (set/read) Pulse width (set/read) Pulse rate (set/read) Board controller (1x) Communicate using CAN or USB Allows setting and reading: Thresholds (set/read) Power connection (read) Inputs Type Outputs Power connection VME J1 64 LVDS output 2 x SCSI2 32 analog signals 2 x DB37 8 analog sums of 4 ch 8 x LEMO-00 2 analog sum of 16 ch 2 x LEMO-00 Silvia Martellotti - Laboratori Nazionali di Frascati 3 September 2014
LAV front-end electronics LAV electronics channel working principle High threshold Clamp Amp 6x Compare threshold Low LVDS Split to analog sum To TDC To TDC 16ch IN Analog 32ch OUT LVDS CANopen IN OUT VME custom power Split the input signal into two copies: 1 copy to comparator + 1 copy to analog sums Clamp the signal preserving its width, wide dynamical range is expected (MIP 70 MeV ~ 10 mV, high energy γ ~ 1V) Amplify the signal and compare with 2 thresholds Each threshold is independently adjustable up to 250 mV Produce an LVDS signal and send the signal to the digital read out board Silvia Martellotti - Laboratori Nazionali di Frascati 3 September 2014 Charge reconstruction algorithm The actual dependence of Q on ToT can be obtained using a fit to the distribution of Q vs. ToT. This relation is in principle the same for all the LAV PMTs. The sensitivity of the method is reduced at high charge due to the exponential dependence of Q on ToT. Measure ToT vs. charge using QDC and TDC only during calibration not during experiment Fit the function Q(ToT) (i.e. polynomial function) During data taking, measure the time using a TDC only Charge reconstruction in NA62 LAV
Test setup Hardware setup Hardware layout Software flaw chart LAV FEE x 16 LAV FEE Board TDC CAEN V1190B VME Bridge V1718 Linux PC x 2 32 ch VME Pulser 32 ch x 16 GPIB programming repeat n times Store serial ID DB repeat n times Repeat n’ times Set FEE Threshold Program pulser Run Daq binary raw to .root .root storage Data analysis Log File Borad report Silvia Martellotti - Laboratori Nazionali di Frascati 3 September 2014
Test stand at CERN The LAV front-end electronic test stand has been moved to CERN to allow in situ calibration and testing of the boards Test of all the 86 boards have been performed at CERN by F. Gonnella, V. Kozhuharov, S. Martellotti, M. Moulson The facility will be also used to characterize the boards we have provided to CHOD and CHANTI Silvia Martellotti - Laboratori Nazionali di Frascati 3 September 2014
Board modification is done All modified boards have been delivered, mounted and tested. 5 boards have problems, 1 is still in Frascati 83/89 boards are ready 81/81 are installed we have only 2 fully working spares Investigations to fix problems on the remaining 5 spare is ongoing Thanks to Katelyn Fariss and Mia de los Reyes! Silvia Martellotti - Laboratori Nazionali di Frascati 3 September 2014
Hysteresis (histə’ri:sis) Time of threshold is measured with two different pulse setups: The mean values of the ToT distributions in the two setups are evaluated Hysteresis is evaluated with the following formula: Note that this strategy does not imply any knowledge on the threshold. threshold 5ns 20ns Francesco Gonnella - Laboratori Nazionali di Frascati 11 February 2014
Hysteresis The hysteresis has been correctly modified Silvia Martellotti - Laboratori Nazionali di Frascati 3 September 2014
Threshold offset The new threshold offset is lower we gain some mV in sensitivity Silvia Martellotti - Laboratori Nazionali di Frascati 3 September 2014
Threshold offset channel by channel The offset is independent from the channel number Silvia Martellotti - Laboratori Nazionali di Frascati 3 September 2014
Calibration slope vs channel The amplification is dependent on the channel number, i.e. on the track length Silvia Martellotti - Laboratori Nazionali di Frascati 3 September 2014
Calibration-line slope The calibration-line slopes were not altered by the hysteresis modification Silvia Martellotti - Laboratori Nazionali di Frascati 3 September 2014
Strange Time-over-Threshold In few channels we have a peculiar ToT Silvia Martellotti - Laboratori Nazionali di Frascati 3 September 2014
Time resolution The time resolution is not effected from the hysteresis modification Silvia Martellotti - Laboratori Nazionali di Frascati 3 September 2014
Time resolution vs channel There is a clear dependence of the time resolution on the channel number and not on the mezzanine. This is under investigation. Silvia Martellotti - Laboratori Nazionali di Frascati 3 September 2014
Efficiency drop Before hysteresis modification: there were a drop in the efficiency curve for 18% of channel in leading only and 20% of channel in leading and training After hysteresis modification: the drop in all the efficiency curves disappeared! Silvia Martellotti - Laboratori Nazionali di Frascati 3 September 2014
Conclusions Boards were modified, re assembled and tested in time We have 81/81 installed boards and 2 spares We will work to repair the 5 boards with some problems The performance is improved, not only the hysteresis: Efficiency drops disappeared We can now safely go down in threshold to 5 mV (even 4 if we smartly swap high and low threshold) Silvia Martellotti - Laboratori Nazionali di Frascati 3 September 2014
Thank you for your attention 3 September 2014