Transcript HCAL_Upgrade_AC-DC_Nov08_2011_S_Los

HCAL Upgrade 2016(18?)
SiPM to QIE10 Coupling / FY2012
Sergey Los
FNAL/CMS/HCAL
S. Los
HCAL Upgrade Workshop, Nov. 08, 2011, FNAL
1
Introduction
Damn AC !
Customer is ALWAYS
right!
• AC coupled scheme for SiPMs was a success at the test
beam tests, and this scheme works for HO Upgrade
• Now that time is running towards SiPM upgrade for HB/HE,
it’s time to look again, and select appropriate coupling for
that application
S. Los
HCAL Upgrade Workshop, Nov. 08, 2011, FNAL
2
Basic AC/DC Coupling for SiPM/QIE
• DC coupling is just it, but with some complications:
• SiPM output is directly fed into QIE
• Full charge, no additional shaping
• QIE input has an offset from 0V
• AC coupling uses a series capacitor , which attenuates, and shapes signal going into QIE
• External drain resistor is used to drain SiPM leakage current
• Performance of capacitive attenuation scheme is well understood, but compromises
have to be made (say undershoot amplitude vs. BV drop across the drain resistor)
S. Los
HCAL Upgrade Workshop, Nov. 08, 2011, FNAL
3
AC/DC Pros and Cons
Issue
DC
AC
Attenuation
Not practical, the best solution is to have
QIE sensitivity/dynamic range tailored to a
specific SiPM gain
Simple and stable
Leakage current
measurement
High end measurement, not trivial for 100V
bias voltage, but doable, some nonlinearity should be expected
Low end measurement across the
drain resistor
Voltage drop on drain
resistor
Virtually does not exist, as QIE input
impedance is much smaller than typical
drain resistor
Luminosity dependent, can be
adjusted with slow control. High value
drain resistor is needed to lower
undershoot value, and improve
accuracy of leakage current
measurement
Radiation induced leakage
current effects (10uA for
Zecotec, 3E12, G=50,000,
300uA for a 2x2mm
Hamamatsu, G=650,000 )
Pedestal goes up (QIE10 80 LSBs for 10uA)
Baseline goes up,
RMS increases
Pedestal stays the same
Baseline stays the same
RMS increases
(same as DC)
Luminosity dependent
effects
(Minbias event spectrum
per single diode?)
Pedestal goes up
Signals sit on top of leakage current, and
signal pile-up
Signal pile-up depends on luminosity
Pedestal stays the same
Signals sit on average on the same
pedestal value
Signal pile-up depends on luminosity
Pedestal – what you measure with a random trigger
Baseline – moments when there are no signals, tails, or undershoots from minbias events
S. Los
HCAL Upgrade Workshop, Nov. 08, 2011, FNAL
4
AC/DC Pros and Cons
Individual signal effects
Signal sits on top of leakage current
(made mostly by single pe signals),
and the rest of minbias events
Signal has an undershoot, which can peg
pedestal value into zero, duration of
undershoot is anti proportional to it
amplitude
SiPM recovery time
Diode recovery time is equal (almost)
to the individual cell recovery time,
determined by the device itself
After a signal, when a big percentage of
cells was fired, BV sags for all cells for a
long time (Cd x Rdrain), similar effect
happens for SiPMs with fast recovery, and
multi-hit capability
A solution was found for this in a form of a
ballast capacitor in parallel with the diode
Att = Catt / (Cd + Cballast + Catt)
Cballast
S. Los
HCAL Upgrade Workshop, Nov. 08, 2011, FNAL
Rdrain
5
Rate, and Pulse Shape Effects
• Radiation induced leakage current (10uA for Zecotec at 3E12 p/cm2, G=50,000)
– That is 30 single pe signals per TS at the end of the life (40fC/5pe noise per TS)
– For DC coupled solution that means pedestal increase by 80 LSB counts, and being in bigger
size bins. Might be o.k. as is, as the noise at that point is 13 LSB. Leakage current can be
compensated for inside the QIE10, or externally, or not at all
– For AC coupling this current generates a voltage drop across the drain resistor, and has to be
compensated for (say 10uA*10Kohm=100mV, at 25mV resolution of a 12bit, 100V BV DAC).
Pedestal value does not change, noise increases the same way as in DC case
• Luminosity dependence (up to 1uA (25fC, 8LSB, 10mV@10Kohm) per tower for a
Zecotec diode at G=50,000, and SLHC luminosity
– Most of those signals are MIP signals (around 10 pe, or 30 LSB, and 12.5MHz)
– For DC coupling that gives a pile-up of MIP-style signals, effective pedestal increase, and
noise increase
– For AC coupling there is no pedestal increase, but the same noise increase
• Large signal effects
– For DC coupled system there are no surprises here besides familiar pile-up
– For AC coupled system the mostly notable issue is that for a small pedestal value, and a
reasonably sized undershoot (1-10% of signal amplitude), QIE readout is going to be pegged
to zero for a certain duration after signals above some amplitude. My gut feeling is that this is
going to introduce a negligible dead time, but it is better if somebody can run a simulation for
this (spectrum of higher amplitude signals times undershoot time, which is proportional to
the amplitude)
S. Los
HCAL Upgrade Workshop, Nov. 08, 2011, FNAL
6
Technical Issues
• BV generation/regulation
– HO solution for this is applicable to both AC/DC coupling, and has been tested to 4.2E12
p/cm2 (going to BV <100V will help a lot, KETEK rocks!)
• Leakage current measurement
– HO solution will work for AC coupling
– DC coupling will require high end current measurement, which is not as trivial, given a 100V
BV (think have a reasonable solution, will be ready for the 2012 Summer TB)
– Do we need more than 12 bit resolution in sight of higher radiation induced leakage currents?
• Signal attenuation
– Passive component attenuation for AC coupling, not sensitive to the QIE Rin
– A whole R&D is required if we want to have attenuation for DC coupled QIE10, we’d better
just redesign QIE sensitivity for the selected SiPM gain:
– QIE10 input impedance changes with input signal amplitude (105 ohm)
– QIE10 input is not referenced to zero, but to a 1.2V reference voltage
– Current mirror splitter is probably the best option for DC attenuation, but linearity, bandwidth,
and real estate are the issues (which Tom Z. has already solved inside the QIE)
• Signal saturation
– Main AC coupling disadvantage of having additional saturation source due to voltage drop
across the diode, when many pixels fire was figured out how to deal with during the summer
2011 TB – ballast capacitor in parallel to the diode (extra gain of SiPM only helps here, as it
allows to lower the value of the attenuating capacitor, speeding up the QIE response!!!
S. Los
HCAL Upgrade Workshop, Nov. 08, 2011, FNAL
7
Summary
• There is still pretty much a single issue with AC coupling – undershoot
after a big pulse
• Signal saturation due to a voltage drop has a ballast capacitor solution
• Undershoot effect should be studied with numbers in hand
(estimating dead time caused by undershoot for different undershoot,
and pedestal values)
• DC coupling presents a number of technical challenges
• Signal attenuation scheme
• Hopefully not an issue anymore /Leakage current measurement/
• Pedestal shift compensation (may be not needed for radiation induced leakage
current, as it is stable, but can be a plus for luminosity dependent shift)
• Are there indeed issues with physics information extraction when
using AC coupled design
S. Los
HCAL Upgrade Workshop, Nov. 08, 2011, FNAL
8
Summary
S. Los
HCAL Upgrade Workshop, Nov. 08, 2011, FNAL
9