DC3 Specifications and Plans

Detector Geometry

We aim to to describe the geometry to about the 1% level in terms of radiation length in the region eta < 2.5. This means that the amount of material in regions extending a few centimeters should be described to an accuracy of 1% of its total radiation length. This requirement is more important the closer we are to the interaction point. The motivation for this comes from the desire to have a good E/p measurement for calibration. It was determined that that the material needs to be know to the 2% level in the SCT and even better in the pixel for this purpose. This is further described in Ch 7 and 12 of the TDR. One will also get calibration from z->ee which has more relaxed requirements, but the E/p measurements is an important cross-check.

While we aim for 1%, we will be satisfied if we get to about 2%.

Many items are being weighed which provides a good means of cross checking that we have accounted for all material and the weight can also be used to scale material were necessary. In some cases it is the only means of determining how much material was used for items like solder. In order to get the correct radiation length it is necessary to know the composition of materials. This is were most uncertainty will probably arise as some material composition are not exactly know and we will have to use our best guess in such cases.

The level of detail is hard to quantify. We feel it is important to reproduce variations in eta (to what extent has not been exactly quantified), while in phi we can tolerate more smearing of the material. Some rough guiding principles used for determining what level of detail is that objects greater than a few cm3 or that have large local radiation length will be described by a separate volume. In practice, its a matter of someone making an intelligent judgment based on all the available information.

The approach being taken to reach these goals is essentially a material audit. The detectors components have been broken down into their hierarchy. For each component or assembly we ask for a document to be filled in (a pro-forma is provided) with information such as references to drawings, general description, location, material compositions and so forth. The format is a Word document to allow flexibility as the type of information varies from case to case. Different sections have been assigned to a person responsible for collecting and filling in the information. The priority has been to ensure components are weighed before it is too late.

More information on the progress with the updates can be found at InDetGeometryUpdate.

Current geometry: It is estimated that the current geometry is probably accurate to about 10-15%. The level of detail is not too bad. Material compositions, however, require a lot of improvement.

Simulation specific

Silicon

  • Delta ray cut: Currently 1mm in DC2. We anticipate using a much smaller value. It is expected something less than the pitch is needed (ie < 50 um) which will lead to more hits. 10um was used in the SCT test beam and validation studies. Still need to study what is an optimal value.

  • Max step length: Set to 50 um in DC2. This needs to be optimized, though we do not expect to need to increase it. We may be able to decrease it as the step is divided up in digitization if it is too long (although we lose the fluctuations along the step).

TRT

No changes anticipated for particle cuts. TR and PAI model as currently in release.

Alignment.

Currently we can only simulate a perfectly aligned detector. We can misalign the database and so have misaligned reconstruction.

The inner detector alignment community consider it necessary to simulate a misaligned detector in order to have full confidence in their alignment procedures . This is restricted to rigid volume movement. There are currently no plans to include distortions at the simulation level.

While the infrastructure will support simulating a misaligned detector, doing so with the current geometry will cause volume clashes.

Alignable nodes have been put in the GeoModel description at various levels in the hierarchy

  • Level 1: Whole Pixel, SCT barrel, SCT endcaps, TRT barrel and TRT endcaps.
  • Level 2: Whole layers and disk in silicon, TRT modules.
  • Level 3: Individual silicon elements. (Nothing for TRT at this level)

Alignment shifts are accessed from the conditions database using the IoV callback mechanism. The alignable nodes in GeoModel are then updated. This infrastructure has been used successfully in the CTB and by the alignment community for the ATLAS alignment.

What still needs doing from Inner Detector side:

  • SCT modules need to be treated as a single alignable object. At the moment each side is moved independently.

  • Leave enough space around objects so that they can be moved without causing volume clashes. ATL-COM-INDET-2005-001 documents the sorts of movements that we need to consider. (Typically 100s of microns for module tolerances)

  • Decide on what are the alignable objects in the TRT endcap.

From the ATLAS side:

  • The simulation must support geometry initialization at begin of run. (This was tried with a private copy of G4SimAlg and seems to work OK.)
  • Geo2G4 must see top level transforms.

Requested data sets: Any data sets with high energy muons (> 2 GeV) such as a trigger sample should be sufficient. Require:

  • 1 set with know misalignments. Random within tolerances.
  • 1 set with a systematic shift of whole structures (such as the whole SCT wrt to the pixels.
  • 1 blind set.
  • 1 set with other systematic shifts such as sagitta distortion.

Digitization. - Calibration/Conditions

Ideally we want to have all conditions type data handled in the same way we expect it to work in the real ATLAS in order to exercise the whole system. This is also important for commissioning needs.

What is required for reconstruction is generally less than what is needed in order to reproduce run conditions. For example reconstruction doesn't care what the thresholds were in the silicon detectors but they are needed in order to reproduce the detectors efficiency and cluster widths.

Silicon

  • In order to reproduce run conditions we need
    • Noise
    • Dead/Noisy channels.
    • Thresholds
    • Bias Voltage
    • Depletion Voltage
    • Temperature of Sensor (has to be inferred for Temp measured on hybrid)
    • Timing
    • Other FE settings (eg variations in gain)

  • For reconstruction we need
    • The Lorentz angle which depends on * Depletion voltage * Bias voltage * Temperature
    • Dead/noisy channels.

TRT

  • In order to reproduce run conditions we need:
    • Noise
    • Dead straws
    • Cross talk.
    • Thresholds
    • HV.
    • Other resolution smearing parameters

  • For reconstruction we need
    • Calibration (tzero and r-t)
    • Dead/noisy straws

While we should have as much conditions type information coming through the proper channels many of the effects are already included in the code.

Possible improvements and effects to include in digitization (these wont all necessarily get done):

  • Pixel:
    • Become time aware.
  • SCT:
    • Time offset to be correctly handled. Important for Cosmics. (Not much impact for DC3)
    • Use mag field service for Lorentz angle. Code already in place.
  • TRT:
    • Effect of mag field on drift time.
    • Pressure effects

Level of noise and dead channels

As a baseline we will simulate realistic levels of noise, while still allowing for extra noise via job options. We will include real noisy/dead channels maps real levels of noise where know, and use typical values elsewhere.

Setting of default conditions

This is currently a mixture of hard coding, job options and numbers in the DD database. Clear guideline were these are set would be useful. Something like the DD database seems a reasonable approach.

Bytestream

Get realistic cabling and retrieve the cabling from the Conditions DB. Redundancy links to be taken into account (this has no impact on physics and just effects the BS converter).

Pileup

We should make sure Digitization handles it properly. We are more or less OK. For Si, currently we will stick with only one bunch crossing until someone studies if there is any significant effect from previous bunch crossings.

Whats not covered.

The degradation with radiation is not covered. Radiated detectors will lead to more variation across the detector. Si detectors will have different bias voltages and likely have different efficiencies. There are no plans to simulate these effects for DC3.

Summary

  • Geometry: Aim for accurate rad length to about 1% in eta<2.5.
  • Simulation cuts: Anticipate smaller delta ray cuts in silicon -> More hits. Requires optimization study.
  • Alignment: Aim to simulate misaligned detectors.
  • Digitization/Conditions: Aim for all conditions type data from Cond DB.

Core support needed

  • Simulation support for reading of geometry at begin of run. Reading top level transforms.
  • Clear guideline for handling default conditions.

Other information:

-- GrantGorfine - 19 May 2005

Edit | Attach | Watch | Print version | History: r3 < r2 < r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r3 - 2020-08-30 - TWikiAdminUser
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    Sandbox/SandboxArchive All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2023 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback