hvcmos abc 130 implications francis 20/02/14fa1. standard stereo 20/02/14fa2 tonys slide
DESCRIPTION
Standard Stereo 20/02/14FA3 BC L0L1/R3 L0 PipelineL1 Pipeline DCLSER 256 Hits Positions encoding Hits positions Option : ABC130 with no FE ABC130 Digital functions kept as now, no changeTRANSCRIPT
1FA
HVCMOSABC 130
ImplicationsFrancis
20/02/14
2FA
Standard Stereo
20/02/14Tony’s slide
3FA
Standard Stereo
20/02/14
BC L0 L1/R3
L0 Pipeline L1 Pipeline DCL SER
256 256
Hits Positionsencoding
Hits
pos
ition
s
Option : ABC130 with no FEABC130 Digital functions kept as now, no change
4FA
Binary
20/02/14Tony’s slide
5FA
Binary
20/02/14
The way I understand it :
One hit position in 256 strips is encoded into 8 bits 8 wire bonds 8 inputs to ABC130 expansion to 1 hit position in 256 channels in ABC130
Two hits : second hit is encoded into 8 (next) bits 8 (next) wire bonds 8 next inputs to ABC130 expansion to 2nd hit position in 256 channels in ABC130
Etc … up to 16 hits
6FA
Binary
20/02/14
BC L0 L1/R3
L0 Pipeline L1 Pipeline DCL SER
88
88
1st hit
2nd hit
Option : ABC130 with no FE, stores 8 bits hit addresses instead of 256 bits hit positions image ABC130 Digital function simplified by the DCL unnecessary
7FA
Z-Encoded
20/02/14Your slide
8FA
Z-Encoded
20/02/14
The way I understand it :
One hit position in 256 strips is encoded into 12 bits (7 for X and 5 for Z) 12 inputs to ABC130 expansion to 5 bits ( z position) in one of the 128? (7 bits) memory locations (ie each of the location is 5 bits size ..)
Two hits positions in 256 strips : second hit is encoded into 12 (new) bits (7 for X and 5 for Z) 12 (new) inputs to ABC130 expansion to 5 bits ( z position) in another of the 128? (7 bits) memory locations
Etc … up to 8 hits
9FA
Z-Encoded
20/02/14
BC L0 L1/R3
L0 Pipeline L1 Pipeline DCL SER
1212
1212
1st hit
2nd hit
Option : ABC130 with no FE, stores 12 bits hit addresses instead of 128 bits hit positions image ABC130 Digital function simplified by the DCL unnecessary
10FA
Z-Encoded
20/02/14
The occupancy (~1hit per 128) is such that in average there is one hit per BC to store : the economy of data driven memory compare to time driven (every BC) looks minimal …
In Z binary or Z Encoding the complexity is in the HVCMOS periphery, where bitspositions at every BC have to be encoded to an address number.
- Easy : clocked and pipelined (ie min latency of a few BC)- Less easy : clocked and encoded in 1BC (no reason to get this constraint ?)- Less Less easy : unclocked, asynchronous ….