D0note 3537

Draft 2.3

B. Baldin, O. Bardon, J.M. Butler, M. Fortner, S. Hansen, E. Machado, N. Parashar, C. Rotolo, S. Uvarov, D. Wood

Differences from V 1.0:

1. All new L2 formats (1 16-bit word of complete address + 1 16-bit word of data for each hit)
2. Major revision of MDT L3 format
3. MCEN formats added.

Differences from V 2.0:

1. Superfluous word counts removed.
2. MCEN formats updated per JMB/EM
3. "DSP" removed from document title

Differences from V 2.1:

1. VBD word count and padding trailer added to L3 header
2. Psi offsets added to alignment constants
3. Second crossing number word added to keep track of all 3 version of the crossing number
4. Some bits defined in error/status register (calib, 1/n)
5. Pad word counts per wire included in PDT L3 format
6. Explictly state which word counts include themselves

Difference from V 2.2:

1. Extra crossing numbers removed from L2 and L3 headers (back to V2.1)
2. Fixed small errors:

• Spurious "Data Word Count" removed from MCEN L2 format
• MTD tube address tables fixed
• CMSC module numbers corrected
• Several typos fixed

FE word count: Inclusive number of 16-bit words in the data block.

The header length (8 words) is included in the count, including the word count word itself.

The module ID follows a convention similar to that used in Run 1 for the PDTs. It is a 3-digit decimal number, with the first digit being the layer:

A-layer = 0

B-layer = 1

C-layer = 2

the second digit being the eta "barrel" for 0-4 and an arbitrary identification number for the other systems:

CMSC = 5

MDT north, MCCM north = 6

FMSC north = 7

MDT south, MCCM south = 8

FMSC south = 9

and the third digit is the octant 0-7 (where appropriate) and "8" or "9" for those systems which span all several octants:

MDT = 0,2,4,6

CMSC = 0,1,2,3,5,6 (0 includes oct 7, 3 includes oct 4)

FMSC east = 8

FMSC west = 9

MCCM = 8

So specifically, the module ID’s are:

PTD: 010, 011, …035,036,100,101,…,146,147,200,201,…246,247

CMSC: 050,051,052,053,055,056

MDT: 060,062,066,160,162,164,166,260,…,266,080,082,…286

FMSC: 078,079,178,179,278,279,098,099,198,199,298,299

MCCM: 068,168,088,188

Crossing # :

An error bit is set for the TFW if the Local crossing # from readout controller differs from the L1 event crossing number from the TFW.

Turn #:

Local turn # from readout controller.

Error/ Status bits:

• Bit 0 = calibration data
• Bit 1 = 1/n data (MCENs)
• Bits 2-9 to be assigned later.

Calibration const. set ID: 11 bits -> up to 2047 sets (~ 2/day for 3 years).

Alignment const. set ID: 6 bits -> up to 63 sets (~10 alignment sets in run 1).

DSP code version #: all for up to 8 versions

Common to L3 data from all muon subsystems.

Total word count: Exclusive # of words 32-bit following this word. The length of the rest of the header (8 words) is included in the count, but not the VBD count words themselves. If the total number of 16-bit words is odd, then a 16-bit padding word ($aa55) is added at the end as a trailer.

Crossing #: same as in L2 header

Turn #: same as in L2 header

Event Status Registers: same as in L2 header

Common to L3 data from all muon subsystems.

This trailer word is present only if the 16-bit word count is odd. It is provided to pad out the last longword in the data block, since the VBD executes 32-bit transfers.

The level 2 input DSPs build up 32-bit words, each of which must fully specify the address and relevant data for a hit. The readout DSPs feed these to Level-2 as two 16-bit words, the first of which gives the complete address of the object hit (PDT cell, scint counter, or MDT 8-cell tube), and the second word of which contains and other relevant information (drift distance, time of hit...). Since the addresses in the first word are complete, they contain some information which is redundant with the module ID in the header. The module ID is kept in the header for diagnostic purposes and to make it uniform with the Level 3 headers.

Type: 0=drift tube, 1=scintillation counter

The local wire number in a chamber is given by bits 6-0. The local wire number increases with the readout order. The local wire numbers are show below for a 4-deck PDT (similar scheme for 2- and 3-deck PDT’s, with increasing r layers removed):

Decimal:

Hex:

Drift distance perpendicular to wire with respect to wire position; includes T0 subtraction.

The local phi divides each octant into 20 equal phi slices. This is comparable to the Delta-time resolution, and it is convenient for matching with the CFT and the A-phi systems, which divide octants into 10 slices.

Status: bit 13: 0 = single hit; 1 = multi-hit

bit 14: 1 = missing wire signal

bit 15: 1 = spare

Distance of reference point for this PDT from its nominal position (as defined by geometry constants).

7 bits + 1 sign bit -> 0.1 mm resolution over up to +/- 12.7 mm.

Distance of reference point for this PDT from its nominal position (as defined by geometry constants).

7 bits + 1 sign bit -> 0.1 mm resolution over up to +/- 12.7 mm.

Theta offset: angle of PDT theta reference with respect to its nominal angle (as defined by geometry constants).

Phi offset: angle of PDT phi reference with respect to its nominal angle (as defined by geometry constants).

7 bits + 1 sign bit -> 0.01 mrad resolution over up to +/- 1.27 mrad.

Number of wires with hits.

Max. # of wire channels = (4 FEB’s)*(24 channels) = 96 -> 7 bits.

Wire # : Same as for L2.

Wire Hit Count = number of time data words following this word for this wire.

< 8 hits recorded per wire -> 3 bits.

Pad Hit Count = number of pad data words following this word for this wire.

T0-subtracted time.

1.2 ns bin width -> units of 0.1 ns to avoid truncating significant figures.

~500 ns max drift time -> 13 bits.

Pedestal-subtracted (peak – base) signal.

10 bits of ADC information for ~ 10 pC max. integrated charge -> units of 10 fC.

Flag words:

Bit 15 = F1 = if set, missing peak signal (report base)

Bit 14 = F2 = if set, missing base signal (report peak)

Empty or otherwise flagged word for missing pad signals (i.e., 2 pad words will always appear for PDT’s with pads).

Note: For PDT’s with no hits, the header, alignment words and hit wire count will still be sent to L3, with hit wire count = 0.

Type: 0=drift tube, 1=scint

Region: Central=0, N=2, S=3

Suboctant: Most octants are read out by two SFE’s, so they each correspond to one suboctant. Only in the A-phi (CMSC-A) system does the division actually split the octant evenly in phi.

SFE#: each SFE in the system has a unique 8-bit number, which is embedded in bits 7-14 of the phototube address. With the bits defined as above, the SFE numbers are (hex):

CMSC-A: 00-0F

CMSC-B: 10-1F

CMSC-C: 20-2F

FMSC-A,N: 80-8F

FMSC-B,N: 90-9F

FMSC-C,N: A0-AF

FMSC-A,S: C0-CF

FMSC-B,S: D0-DF

FMSC-C,S: E0-EF

PMT#: differs for each subsytem

Phototube readout order and numbering for CMSC A-layer suboctant (similar for bottom A-layer, with last 4 phi segments omitted):

CMSC-A PMT#’s (Decimal):

CMSC-A PMT#’s (Hex):

The heavy vertical lines show the PDT boundaries under the counters. So bits 5-6 (the highest two bits of the z-index) specify the PDT number. The readout order is also designed so that the gate delays are unique and independent for each set of 15 counters mounted on a single PDT.

The first SFE in an octant will read out 3-PDT’s of counters (24 counters = 48 PMT’s), and the second will read out 2 PDT’s of counters (16 counters = 32 PMT’s)

CMSC-C PMT#’s in Hex (example: octant #2):

FMSC ("pixels"):

Phototube readout order and PMT# for FMSC octant-plane (decimal):

Phototube readout order and PMT# for FMSC octant-plane (decimal):

T0-subtracted time.

Hit channel count: number of phototube channels with hits.

Max. # of channels = (8 SFE’s)*(48 channels) = 384 -> 9 bits.

Number of hit SFE’s gives of # of words of ADC data.

Up to 8 SFE’s / crate -> 3 bits.

Max. 8 SFE’s / crate -> 3 bits.

Distance of reference point for scintillator serviced by this SFE from its nominal position (as defined by geometry constants).

7 bits + 1 sign bit -> 0.1 mm resolution over up to +/- 12.7 mm.

Distance of reference point for scintillator serviced by this SFE from its nominal position (as defined by geometry constants).

7 bits + 1 sign bit -> 0.1 mm resolution over up to +/- 12.7 mm.

Pedestal-subtracted pulse height in fC (max should be about 5 pC).

10 bit ADC -> 12 bits for conversion without truncating significant figures.

PMT #: Same as for L2.

Time: Same as for L2.

Note:

For scintillator crates with no hits, the alignment and ADC data will still be sent to L3, with hit channel count = 0.

A "tube" is an MDT unit with 8 wires.

Wire readout order and numbering for a 4-deck layer (3-deck layers are similar, with z=3 layer omitted):

MDT tube # (bits 0-7) Decimal:

Distance of reference point for first MDT octant-plane in this module from its nominal position (as defined by geometry constants).

7 bits + 1 sign bit -> 0.1 mm resolution over up to +/- 12.7 mm.

z offset: Distance of reference point for first MDT octant-plane in this module from its nominal position (as defined by geometry constants).

7 bits + 1 sign bit -> 0.1 mm resolution over up to +/- 12.7 mm.

theta offset: angle of theta reference for first MDT octant-plane in this module with respect to its nominal angle (as defined by geometry constants).

7 bits + 1 sign bit -> 0.01 mrad resolution over up to +/- 1.27 mrad.

phi and psi offsets: angle of phi and psi reference for first MDT octant-plane in this module with respect to its nominal angle (as defined by geometry constants).

7 bits + 1 sign bit -> 0.01 mrad resolution over up to +/- 1.27 mrad.

Reference 2 offsets: position and angle offsets for second octant-plane in the module. The word definitions are the same as for the first octant-plane.

Tube #: Same as for L2.

3-bit time resolution over ~1 cm drift distance

Note:

For MDT crates with no hits, the common header, alignment words and data word count will still be sent to L3, with data word count = 0.

Note: The MCEN format is done in hardware rather than in a DSP, but it is included here for completeness. All words shown are 16 bit words.

@Mod = index of 96-bit hitmap

@Segment = a contiguous group of 16 centroids

Note: Layer=3 is used to denote data from an MCON (concentrator).

This is fixed in size, but is only read every 1/n events. Each channel contains data in exactly the same format as it is received from the serial daughterboards.

The zero suppressed data is variable in size. There is one word for each non-zero hit map

This word is the same as for L2, but here the module and segment information is needed.

"V" should be set for all valid data, and can be ignored.

One current proposal is to attach "module-wise" alignment constants to the header of each event. Module-wise alignment constants would be the offset of one or a few measured reference coordinates (x, y, z, theta, phi) for each module (PDT; quarter-plane of MDT’s; octant of scintillator pixels serviced by one SFE crate) with respect to the nominal reference coordinates defined in the permanent detector geometry file. The geometry file would also hold the information required to translate the reference coordinates into the coordinates of each individual channel (PDT wire; MDT wire; scintillator pixel). The number of reference points will be picked to suit each type of module; e.g.:

- 1 reference point per PDT

- 2 reference points per MDT module (1 per MDT octant-plane)

- 1 reference point per set of scintillator channels served by one SFE (= 5-8 reference points per module; alignment constants would be attached to the SFE "sub-header" containing SFE address and ADC information, and only attached for events in which a given SFE had a hit).

Another proposal is to attach no alignment constants in the DSP.

T0’s are defined as the reported time in a given detector component for a relativistic particle emanating from the IP on a given crossing. They include particle propagation time (i.e., path length), and signal propagation through electronics, cables, etc. They do not include propagation time through detectors. In particular:

- A PDT’s T0 will be defined as the leading edge of a distribution of all prompt muons in that PDT. PDT drift time and signal propagation time along wires will thus not be subtracted as part of the T0. They will be measured and used to calculate axial and radial positions with respect to wires.

- A MDT’s T0 will be defined as the leading edge of a distribution of all prompt muons in that MDT. MDT drift time will thus not be subtracted as part of the T0. It will be used to calculate radial position with respect to wires. Signal propagation time along wires will be considered negligible at this level; in principle, once the axial position of the track within the tube is known, the signal propagation time along the wire could be corrected for.

- A scintillator pixel’s T0 will be defined as the mean of a distribution of prompt muons in the center of that pixel. Once the position of a track within the pixel is known, light propagation time through the scintillator can be corrected for.

I.e., the T0-subtracted time for a relativistic particle should ideally be zero, modulo propagation time through detectors.

These are described in more detail in the note on calibration constants.