The CMS Phase 1 Pixel Detector

Transcription

1 BPIX FPIX The CMS Phase Pixel Detector Julia Gray University of Kansas On behalf of CMS Tracker Collaboration BPIX supply tube: Module connections Optical links DC-DC conversion Cooling loop FPIX Service Cylinder Exploded view of the upgraded pixel detector. The figure shows the p Julia Gray IPRD3 Talk

2 Content Current Detector Run Conditions Phase Pixel Detector Upgrade Pixel Sensor Pixel Readout Chain Upgrade Timeline Conclusions Julia Gray IPRD3 Talk 2

3 The CMS & Pixel Detector Current Pixel Detector Performance is Good! Julia Gray IPRD3 Talk 3

4 Run Conditions The LHC design was to operate at x 34 cm s - with 25 ns bunch spacing and 25 interactions/bunch crossing (pile-up) Year s (TeV) Int. Lumi. fb - Peak Inst. Lumi. x 34 cm s - Peak (Ave.) Pile-up (2) 5 >LS * (25-5) 25-5 >LS2 * 4 5 >2. (5) 25 *There are three planned long shutdowns (LS), with LS in 234, LS2 in 28, and LS3 in 222 Bunch Spacing (ns) Julia Gray IPRD3 Talk 4

5 Goals for Phase Upgrade To run as well at high luminosities (2x 34 cm s - ) as the current detector at current conditions Will need to survive 5fb - Optimize layout for 4-pixel-hit coverage in <2.5 Reduce material budget High efficiencies, low fake rates at 5 or more pile-up Minimize degradation due to radiation damage Minimize data loss with new pixel readout chip (ROC) Julia Gray IPRD3 Talk 5

6 FPIX and 2 Upgrade.7 detector detector 2.38 Upgrade 4.76 BPIX 3. BPIX2 BPIX BPIX3 BPIX BPIX4 BPIX FPIX 3 BPIX The Phase Pixel Detector Four barrel layers instead three.36 FPIX 3.9 of current.8 3-disk forward system instead of current 2-disk New smaller diameter beam pipe needed to accommodate innermost pixel layer Upgrade = =.5 =. =.5 Outer rings Upgrade 4 barrel layers =2. =2.5 Inner rings 5. cm =2.5 Current = =.5 =. =.5 =2. Current 3 barrel layers Figure 2.: Left: Conceptual layout comparing the different layers and disks in the current and upgrade pixel detectors. Right: Transverse-oblique view comparing the pixel barrel layers in Figure 2.: Left: Conceptual layout comparing the different layers and disks in the current and IPRD3 Talk 6 the two detectors. Julia Gray upgrade pixel detectors. Right: Transverse-oblique view comparing the pixel barrel layers in the two detectors.

7 Pixel Hit Optimization Additional barrel and end-cap layer will give four-hit coverage over its whole range of ±2.5 The forward detector has outer and inner half disks Both half disks use same module type Can use quadruplet track seeds which will greatly reduce the fake rate Julia Gray IPRD3 Talk 7

8 Reduce Material Budget Move electronic boards and connections out of acceptance Use two-phase CO 2 cooling Lightweight mechanical support Volume Present Detector Mass (g) Phase Detector BPIX < FPIX < Pixels radlen Current Pixel Detector Upgrade Pixel Detector eta igure 2.2: The amount of material in the pix Julia Gray IPRD3 Talk 8

9 Efficiencies and Fake Rates Studies done with efficiency vs sample Expected ROC data loss was added to simulation Number of truth tracks matched to reconstructed tracks Track efficiency = Number of truth tracks Number of reconstructed tracks not matched to truth tracks Track fake rate = Number of reconstructed tracks Expected % Data Loss BPIX 2.38 BPIX2.46 BPIX3.8 BPIX4.8 FPIX-3.8 Current Pixel Detector Upgrade Pixel Detector (a) fakerate vs E34 cm s - (25 ns) (b) (c).7 Julia Gray IPRD3 Talk 9 t fakrate vs p (d).5.4.3

10 Radiation Degradation Performance Simulated 2% inefficiency in tracker inner barrel and 2 Uses a sample, with average pile-up of 5 Efficiency loss for current tracking detector at %, upgrade at 4% Julia Gray IPRD3 Talk

11 6.. Silicon sensor requirements 6 electrons which is not sufficient for reliable operation with the present electronics, where the in-time threshold is around 3 electrons. However ence exceeds the expected.5 5 neq /cm2, before Layer substitution, a factor of two and the new ROC allows for a significantly lower threshold of 2 trons. Both these factors insure us that the Layer modules can be efficiently during the expected lifetime with a bias voltage not higher than 6 V. - Barrel will increase pixel modules Signal [ke ] Pixel Modules and Sensors 25 Not irradiated from 768 to 84, pixels from 48 to 2 79 million Forward total module count will be 5 constant at 672 with larger size, pixels increase from 8 to 45 million Module active area: 6.2 x 64.8 mm2 5 Pixel size: x 5 µm Uses the 24 same n+-in-n type Voltage [V] Charge collection Chapter on irradiated and non-irradiated 6. Pixel Modules 4.3 4neq/cm2, Pions 6.2 4neq/cm2, Pions 6. 4neq/cm2, Protons 5. neq/cm2, Protons neq/cm2, Protons Figure 6.4: Charge collectiondoses in non-irradiated and irradiated sensors for different and particle types sensors with different d different particle types (sensors of 285±5 µm thickness.) Barrel Pixel Cells The good spatial resolution of the pixel detector is reached by an analogue inte method. The set of pixels showing a signal from the same particle is called a Interpolation between neighbours is only possible ifforward the cluster size in each dir at least two. Pixel Cells In the polar direction (orthogonal to the 3.8 T field) charge sharing is induced by of charge carriers in the magnetic field. The value of this drift is given by the angle which is a function of the charge carrier mobility. The mobility is a fu the electric field and therefore of the sensor bias. Presently the pixel barrel are run at a bias of 5 V which results in a Lorentz angle of 22 and therefo Figure 6.3: Photograph of four pixel cells in the sameofdouble column fortalk BPIX FPIX fraction two pixel clusters in r(left) This leads to a good spatial resolution Julia Gray IPRD3 f.and (right). 3 µm. increasingfor radiation damage, the bias voltage should be increased to com cause inefficiency. In order to predict thewith performance high fluences, sensors have for the increase of the source space charge within the sensor bulk. Eventually the m been irradiated and thoroughly investigated with a radioactive [44] (BPIX) and bias provided by the power supplies (6 V) is at a beam test [45] (FPIX). In both cases the results confirmed that the sensors canneeded to obtain sufficient signa

12 Front End Chips and Readout Chain Pixel Module ROC TBM ROC ROC ROC POH xlt xlt LLD LLD Fanout PP Ribbon Patch PP Multi-Ribbon Cable 2 96 Pixel FED DRx2 DAQ Processing I2C TTC Clock, Reset, Fast I2C CCU CCU CCU CCU LLD Rx4 PP PP 2 96 Pixel FEC DTRx Processing TTC LLD Rx4 PP PP 2 96 Pixel FEC DTRx Processing TTC Julia Gray IPRD3 Talk 2

13 Design Improvements for Pixel ROC Uses a 6 Mbit/sec LVDS data link istead of 4MHz analog 24 time stamp buffer cells (2x more than present) Data buffer size increased from 32 to 8 Add an additional buffer on the Readout Chip (ROC) level Reduce the operational charge threshold from 35e Detector Radius % Data loss for (cm 2 ns) (cm) Current detector BPIX BPIX BPIX FPIX and Upgrade detector BPIX BPIX BPIX BPIX FPIX Julia Gray IPRD3 Talk 3

14 Digital vs. Analog Readout Chip PSI46V2 PSI46DIG ROC size 7.9 mm x 9.8 mm 7.9 mm x mm Pixel size µm x 5 µm µm x 5 µm Smallest radius 4.3cm 2.9cm Settable DACs / registers 26 / 2 9 / 2 Power Up condition not defined default values pixel charge readout analog digitized, 8bit Readout speed 4 MHz 6 Mbit/s Time stamp Buffer size 2 24 Data Buffer size 32 8 Output Buffer FIFO no yes Double column Speed 2 MHz 2 MHz (4 MHz) Metal layers 5 6 Leakage current compensation yes no in-time threshold 35 e < 2 e PLL no yes Data loss at max Operating flux 3.8% at 2 MHz/cm 2.6% at 5 MHz/cm 2 Analog ( 3% at 58 MHz/cm 2 ) Digital Julia Gray IPRD3 Talk 4

15 Readout Chip Studies The new chip was produced this year High rate studies ongoing Studies with X- rays show that the readout thresholds as low as 5 electrons can be seen Julia Gray IPRD3 Talk 5

16 Token Bit Manager Development Token Bit Manager (TBM) controls readout and operating constants of ROCs; distributes clock, trigger, and reset Upgrades proposed use single data stream encoded at 4Mbps sent through a single fiber to FED Julia Gray Figure 5.4: Two TBM7s with Data Keeper. IPRD3 Talk 6

17 Optical Link 8 Chapter 5. Front End Chips & Read New high speed laser diodes, Transmitter POH Optical Output Receiver Electrical Output Threshold current (ma).4 Figure 5.8: Full link eye diagram results for different data patterns. 2. 4B/5B NRZI 4 Mb/s 4.6 4B/5B NRZ 4 Mb/s AOH Device G FP TOSA Event data 32 Mb/s 6.8 Relative Slope Efficiency 8 PRBS7 32 Mb/s Optical Sub-Assembly (TOSA), are more radiation hard than current Analog Optohybrid (AOH) The Pixel Optohybrid (POH) has been prototyped using a TOSA candidate POH uses same chipset as AOH using a 6 Chapter 5. Front End Chips & Readout Chain Analog Level Translator and Linear Laser Driver. 3. Fiber plant - Optical fanout cables to connect the new POH & DOH to t PP (Patch Panel at the end of the CMS vacuum tank) Pion fluence ( cm ) Digital Receiver Module (DRx2) - the twelve-channel digital receiver mo will sit on an upgraded Pixel FED in USC55 Figure 5.5: Comparison of radiation damage in between lasers used in the present AOH and Julia Gray IPRD3 Talk lasers for the upgrade. The radiation damage effects are much smaller for the new devices. 7 Deliverable is the responsibility of Fermilab, while deliverables 2, 3 & 4 ar with the current AOH to ease in-system testing. This footprint and/or of connector sponsibility CERN.type An overview of the project flow is shown in Figure 5 may change in subsequent iterations of the design that may also be required to fully the attendant PBS is shown in Table 5.2. The first part of the project, the match new mechanical constraints as the overall system design evolves. stration of feasibility of the POH concept, has been carried out by CERN. Th has been handed over to Fermilab for final implementation that will take into

18 Front End Driver readout Replacement module in counting room Captures, buffers, synchronizes, and packs data readout before transmission to CMS and DAQ Needs to match requirements of optical input link, digital data format, and bandwidth output link A 48 channel FED with 2Gbits/s DAQ output link could handle reading a nearly saturated pixel detector at 4Mbps FRONT END 48 x 4Mbps TBM data in 2 channel POD SFP+ SFP+ 2 x Gbps FED data out 2 x DAQ throttling CENTRAL DAQ IO IO... IO GTX GTX PERIPHERALS DDR/QDR RAM (option) IO CPLD MMC IO Xilinx Series 7 FPGA JTAG IO IO GTX GTX IO GTX LEDs, Sensors, IPMI USB 2., microsdhc AMC3 4.8MHz TTC clock TTC A/B channel stts throttling Gbps local (crate) DAQ GbE control clocking up to 2 x Gbps (spare) backplane communication MCH Julia Gray IPRD3 Talk 8

19 Phase Upgrade Timeline The plan is to install the new Phase Pixel Detector by LHC Machine 7 TeV LS 3 TeV TS 3 TeV XTS 3 TeV CMS Openings open Central Beampipe Old New Present detector maintenance Pilot blade installed with FPIX Upgraded pixel installation tests at P5 CO2 cooling plants construction ROC production and test DC-DC converter production Optohybrid production Sensor production Module production Service tubes production Modules integration onto mechanical support System assembly and system tests FED firmware, software, hardware production/test Pixel installation, commissioning, operation gure.3: Overview Julia Gray of the construction schedule for the Pixel IPRD3 Phase Talk Upgrade 9 projec

20 Conclusions The upgraded pixel is expected to be installed in CMS during an extended winter shutdown between 26 and 27. A few modules (pilot detector) are being inserted into the CMS forward region in 24 to assure that the data acquisition is fully integrated into CMS before the full installation The detector will provide better performance than the present pixel detector in the higher data rate environment. Julia Gray IPRD3 Talk 2

21 Backup Slides Julia Gray IPRD3 Talk 2

22 .4 Tracking Efficiency with Muons E34 cm s - (25 ns) 2E34 cm s - (25 ns) 2E34 cm s - (5 ns) (a) efficiency vs.8 efficiency vs.8 fakerate vs Upgrade D E34 cm s - (25 ns) 2E34 cm s - (25 ns) 2E34 cm s - (5 ns) (a) Current.6 Figure 2.4: Tracking efficiency (a,c) and fake rate (b,d) for the mu fakerate vs. E34 cm s - (25 ns) 2E34 cm s - (25 ns) 2E34 cm s - (5 ns) track h, for the current Upgrade detector Detector (a,b) and the upgrade pixel detecto for zero pileup.5 (blue squares), an average pileup of 25 (red dots), an Julia Gray IPRD3 Talk 22.4 Upgrade (c) (d) diamonds), and an average pileup of (brown triangles) with expected at the given luminosities as detailed in the text.

23 .4 Chapter 2. Expected Performance & Physics Capabilities efficiency vs efficiency vs Tracking Fake Rate with Muons efficiency vs fakerate vs fakerate vs.6 (a) E34 cm s - (25 ns) 2E34 cm s - (25 ns) 2E34 cm s - (5 ns).4 fakerate vs E34 cm s - (25 ns) 2E34 cm s - (25 ns) 2E34 cm s - (5 ns) Current Detector Current Detector E34 cm s - (25 ns) 2E34 cm s - (25 ns) 2E34 cm s - (5 ns) (b) E34 cm s Figure - (25 ns) 2E34 cm s 2.4: Tracking.6 efficiency (a,c) and fake rate (b,d) for the muon sample as a function of - (25 ns) 2E34 cm s - (5 ns) (c).5.4: Tracking efficiency (a,c) and fake rate (b,d) for the muon sample as a function of for the current detector (a,b) and the upgrade pixel detector (c,d). Results are shown pileup (blue squares), an average pileup of 25 (red dots), an average pileup of 5 (black ds), and an average pileup of.4 (brown triangles) with ROC data loss simulation (b) (a) (c) (d) fakerate vs track h, for the current.5 detector (a,b).5 and the upgrade pixel detector (c,d). Results are shown Upgrade Julia Gray Detector IPRD3 Talk 23 for zero pileup (blue squares), an average pileup of 25 (red dots), an average pileup of 5 (black diamonds), and an average pileup of (brown triangles) with ROC data loss simulation expected at the given luminosities as detailed in the text fakerate vs Upgrade Detector (d) Upgrade Detector (d)

24 .4 Tracking Efficiency(Pileup) E34 cm s - (25ns) 2E34 cm s - (25ns) 2E34 cm s - (5 ns) efficiency vs.8 (a) efficiency vs.8 (c) fakerate vs Upgrade De E34 cm s - (25ns) 2E34 cm s - (25ns) 2E34 cm s - (5 ns) E34 cm s - (25ns) 2E34 cm s - (25ns) 2E34 cm s - (5 ns) Current.6 Figure 2.3: Tracking efficiency (a,c) and fake rate (b,d) for the t t sam fakerate vs (c) (d) h, for the current detector (a,b) and the upgrade pixel detector (c,d zero pileup (blue.5squares), an average pileup of 25 (red dots), an av Julia Gray IPRD3 Talk 24.4 Upgrade Upgrade Detector diamonds), and an average pileup of (brown triangles) with R expected at the given luminosities as detailed in the text.

25 .4 Chapter 2. Expected Performance & Physics Capabilities efficiency vs efficiency vs Track Fake Rate(Pileup) efficiency vs fakerate vs fakerate vs Current Detector (b) E34 cm Figure 2.4: Tracking efficiency (a,c) and fake rate (b,d) for the muon sample as a function of s- (25 ns) 2E34 cm s - (25 ns) 2E34 cm s - (5 ns) (c) (c).6 (a) E34 cm s - (25 ns) 2E34 cm s - (25 ns) 2E34 cm s - (5 ns).4 fakerate vs E34 cm s - (25 ns) 2E34 cm s - (25 ns) 2E34 cm s - (5 ns) Current Detector E34 cm s - (25 ns) 2E34 cm s - (25 ns) 2E34 cm s - (5 ns) for zero pileup (blue.5 squares), Julia Gray an average pileup of 25 (rediprd3 dots), Talk an average 25 pileup of 5 (black.4: Tracking efficiency (a,c) and fake rate (b,d) for the muon sample as a function of for the current detector (a,b) and the upgrade pixel detector (c,d). Results are shown pileup (blue squares), an average pileup.4 of 25 (red dots), an average pileup of 5 (black ds), and an average pileup of (brown triangles) with ROC data loss simulation (b) (a) (c) (d) fakerate vs track h, for the current detector (a,b) and the upgrade pixel detector (c,d). Results are shown Upgrade Detector fakerate vs Upgrade Detector (d) Upgrade Detector (d) diamonds), and an average pileup of (brown triangles) with ROC data loss simulation expected at the given luminosities as detailed in the text.

26 Readout Chip Studies Use x-ray fluorescence test set-up Distribution width smears threshold scan Re-trim for every threshold point " " Actual! mean)charge) Julia Gray IPRD3 Talk 26