Title: CLAS12 Micromegas Tracker: FE electronics eric.delagnes@cea.fr
1CLAS12 Micromegas TrackerFE electronicseric.d
elagnes_at_cea.fr
2Introduction
- Nearly no work made on the FEE since the May
review. - 50 of the slides already shown.
- Real work cannot start before March 2010.
- Preliminary study for compatibility with SVT.
- Outline
- Main Specification for the MicromégasTracker FE
chip. - VFE expected performances (starting from AFTER
ones). - Selected Architecture.
- MTFEC for SVT ?
- Plans.
3Features common to all FE solutions Technology
choices
- Technology choices
- Use an existing chip there are not a lot of
available tracker chip adapted to both analog
readout and large detector capacitances - the APV0.25 designed for CMS could be an option
(under evaluation) - nearly a perfect chip
- we already use it.
- APV availability
- APV not designed for high detector capacitance.
- Large occupation time (RC-CR shaping).
- New chip
- Using a well known technology (AMS CMOS 0.35µm)
- very front-end part nearly already designed.
- Chip size if integrates a lot of digital
electronics. - Using a more recent technology
- long term availability.
- prepare the future for our lab.
- Less power consumption
- ? Less noisy.
4Features common to all FE Chip solutions
- Packaging, modularity
- For Mmegas Prefer a QFN/QFP package (no bare
die). - 32-64 channel/chip is the best modularity for
integration on FE boards. - 128 channel/chip gt big chip package difficult
to handle during test. - Power consumption
- As we are outside the magnet, the requirements
can be relaxed/ 5 mW/ch for the FE Chip. - Configuration (Slow-control) Link
- To program test modes, peaking time, ranges, etc.
- Test system
- Each Channel can be pulsed individually (or all
together). - For test purpose and not absolute calibration.
- Input Protections
- Designed to reduce the size (or even the need) of
the external protections.
5Requirements for the CLAS12 MM electronics (1).
- For the moment only the barrel has been studied
- Use of standard (without resistive sheet)
Micromegas assumed - 20000 channels
- Electronics moved away from detector using 0.8m
Kapton cables - Particle rate lt 20MHz
- Hit Rategt 48 kHz/strip (considering cluster
size4) - External trigger with Max Trigger Rate 20 kHz,
fixed latency 4.5µs - Inefficiency due to electronics 2
- Ghost hits/trigger lt 8/view. Noise hits rate
negligible - Main functionalities (not necessary performed in
this order) - Collect, amplify and filter the detector signal
- Discriminate pulses
- Timestamp pulses
- Select pulses within a L1W ( 100ns) window
around the L1 accept signal - Measure signal charge (for centre of gravity
calculation) - Many requirements very similar to those of
COMPASS tracker
6Requirements for the CLAS12 Micromegas
electronics (2).
- Channel Occupancy
- Tocc lt250ns to keep occupancy lt 1.2
- High order filtering (symmetrical shape).
- Shaping peaking time must be
- Large
- To avoid ballistic deficit .
- To Minimize noise.
- Small
- to limit occupancy
- to be Compatible with L1W100ns
From calculations and experience from
COMPASS 100ns peaking time should be ok gt
Tunable between 50-250ns.
7Requirements for the CLAS12 Micromegas
electronics (3).
- Dynamic Range
- 600 9-10 bit Max Signal over ENC required.
- - Max Charge 10 MIP
- - Threshold MIP/10 for efficiency
- - Threshold 6 Thresholds set to 6ENC (for
noise rejection). - Max range (and MIP) depends on the detector gain
- gt Variable gain front-end 4 ranges selectable
by slow control - i.e 160, 320, 640 fC for Micromegas.
- 40 fC range for Si detectors.
- Exemple
- For the160fC range
- gt MIP 100 Ke-
- gt Th 10 Ke
- gt ENC should be around 1500 e- rms (gives a
S/N60) - Feasible with our large detectors ( kapton
cables) ?
8What we can learn from the AFTER chip
- AMS 0.35µm technology.
- Designed for the TPC of T2K.
- Slow Readout (incompatible with use in trackers)
- But very versatile
- shaping time, dynamic range are matching with
our needs. - Front-end part could be re-used nearly as it is
associated with a custom back-end. - Modifications (50ns shaping) in progress for
another experiment. - Noise deeply tested a complete parameterization
has been extracted - Ability to predict the noise in other conditions.
- 1Mrad radiation hardness demonstrated in another
similar chip we designed using the same
technology.
9AFTER ASIC design for T2K
IEEE Trans. Nucl Sci, June 2008
- No zero suppress.
- No auto triggering.
- No selective readout.
- AMS 0.35µm techno
- 500000 transistors
- Main features
- Input Current Polarity positive or negative
- 72 Analog Channels
- 4 Gains 120fC, 240fC, 360fC 600fC
- 16 Peaking Time values (100ns to 2µs)
- 511 analog memory cells / Channel
- Fwrite 1MHz-50MHz Fread 20MHz
- Optimized for 20-30pF detector capa
- 12-bit dynamic range
- Slow Control
- Power on reset
- Test modes
- Spy mode on channel 1
- CSA, CR or filter out
10Requirements for the CLAS12 Micromegas
electronics (3).
- Noise
- Must be minimized to be able to operate at low
gain (if necessary to reduce spark rate). - Huge Flex detector capacitance of 60-80 pF.
- 1600-2000 ENC (for very low gain operation) seems
feasible even with short shaping time - From COMPASS experience.
- From measurements on the AFTER chip.
ENC versus input capacitance for different
peaking times (120 fC range, ICSA400 µA).
Measured on the AFTER chip.
11VFE part of the Chip same as for AFTER
123 possible options for the FE chip architecture
were proposed
- ONLY DEAD TIME-FREE solutions (with dual-port
L1 buffers) are proposed - ASD multihit TDC
- Similar to Micromegas COMPASS tracker readout.
- Time Stamping analog memory
- Trigerless Front-end.
- Selective Readout.
- Analog Memory L1-Buffer (APV-like)
- Similar to GEM COMPASS tracker readout.
- Solution selected
- Better noise rejection.
- Minimum work for us the only which could match
with the schedule and the available manpower.
13Analog Memory L1 buffer solution (APV-like
solution)
- A Switched Capacitor Array is used as a circular
analogue buffer - The analog signals of all the channels is
continuously sampled at Fs in a Switched
Capacitor Array (analogue memories). - When a L1-Trigger occurs it is sent to the chips
with a FIXED LATENCY (TLAT) - 3-4 samples on all channels are kept (frozen) for
each triggered event. - They are read and multiplexed towards an external
ADC _at_ Fread. - Cells are rewritten after readout or if no
trigger occurs during after TLAT. - Dead Time Free architecture
- No interruption of writing during readout of a
triggered event. - several triggered events can be stored in the
SCA waiting for readout. - No on-chip zero suppress all channels are read
for a trigger.
Triggered cells
14SCA Key parameters
- Fsgt 2/Tp (2 samples in the trailing edge)
- gt Fs 20 MHz for Tpeak 100ns
- SCA DEPTH Latency buffer extra cells
- 8 µs latency gt 160 cells.
- 10 events derandomizing buffer gt 40 cells
- SCA depth 256
- 512 cells is feasible but increase cost
15Main advantages of this solution
- Charge is directly measured.
- Oscilloscope-like operation makes diagnostics
easier. - The timing can be accurately calculated from the
samples - better than 1/Fs precision In ATLAS LARG ECAL
1ns rms timing performed with FS40 MHz (and
tp50ns) - Pile-up can be detected and even compensated.
- Common mode noise can be calculated and
subtracted. - Low frequency noise can be partially eliminated
(by subtracting baseline samples). - Operations are performed before zero-suppress
(discrimination)
16Analog sampling solution (APV-like solution)
- Data flow for the whole MM tracker 1600 MByte/s
_at_ the ADC output. - Becomes 20 MByte/s after zero suppress.
- Can be reduced by 3 if an online filtering on
timing is performed. - For
- Simple Proven
- Very robust to bad grounding pickup (common
mode node correction) - Expertise of Saclay on SCAs
- Against
- Need for high frequency ADC FPGAs close to the
very front-end. - Not self triggered
- Need for a L1accept fast and synchronous.
17Use of the Micromégas chip with SVT ?
- Possible issues
- Input DC current limited to 5nA. Can be a problem
with DC coupled Si detectors - AC or DC coupled ?
- Increasing DC current capabilities under study.
- Power consumption
- 5mW/ch planned for MM readout gt cooling issue.
- Low power mode for Silicon detectors ?
- Can we move away the electronics (as for MM) ?
- Noise
- Preliminary study made using
- AFTER parameterization.
- Data from the ENC calculations for Barrel
Modules of the SVT Note assuming there is a
mistake in the leakage current specification
(20nA/ch inst. of 5uA/ch). - A note (excell file) will be available soon.
18Few words about the noise
- Expressed as Equivalent Nose Charge gt input
refered noise. - Several sources, adding quadratically,can be
categorized - Parallel noise current noise at chip input.
Scales as tp1/2 - Serie noise voltage noise at chip input. Scales
as Cdet and tp-1/2 - 1/f noise 1/f noise of preamp Scales as Cdet.
Constant with tp. - 2nd stage noise constant.
- Analytical model takes into acount these noise
sources. - Parameters come from
- Measurements on AFTER
- Simulation
- Theory
19ENC Model
In red Chip Parameters (extracted from
measurements) In blue detector parameters
(calculated from theory) Tp free parameter
20AFTER measurement compared to model
Measurements (Ibias400µA)
Analytical model (Ibias400µA)
21ENC simul for SVT with AFTER-like FE
- Detector Parameters taken from equivalent Noise
Charge calculations for Barrel Modules of the
SVT (excepted Idet) - Simulations on
- 3 ranges 3 ranges with 40pF added (to simulate
a kapton cable) - 3 shaping times (50 ns,100ns, 200ns).
- 2 bias currents for input transistor (5 6mW/ch).
22SVT ENC simulation noise contributions,
5.5mW/ch/ tp50ns
ENC is clearly dominated by serie noise gt
Improvement expected for higher tp
FEMTC model (tp50ns) gt
SVT Note (tp65ns) (reference)gt
23SVT ENC simulation varying tp, IPOL
With tpgt100ns Power6.5mW gt ENClt 2000 e-
(S/Ngt11) for range 12 including 40pF kaptons
cables We could imagine to move SVT
electronics away
24Short term plans.
- No manpower in microelectronics for this project
before March 2010. - VFE part with 50ns shaping is currently designed
for GET. - Study the use of this chip with Silicon
detectors. - Definition of Digital/DAQ electronics and of
integration gt Irakli talk. - Before summer 2010 Submission of a small size FE
chip prototype - 16 channels x 128 cells for lower prototype cost.
- Test during the fall.
- Check the possibility to use APV
- successful test with new large Micromégas of
COMPASS last summer , but detailed analysis of
beam data are required to check the efficiency).