GCT meeting

1
Duplicating data onsource card
Summary of situation so far
Wesley requires that we use the original 3x3
sliding window algorithm. This is fine everywhere
except at ?0.
2
Old scheme
Only had data from ?- source cards to ?- leaf
cards and only had data from ? source cards to
? leaf cards
New scheme
Data from both ? ?- source cards to ?- leaf
cards and data from both ? ?- source cards to
? leaf cards
This requires duplicating data across ?0.
3
Regional Calorimeter Data
Cables 3 4 are still fine
Cables 5 6 need to be (partially) duplicated
Cables 1 2 are still fine
4
THIS IS BASED ON OLD (CURRENT) SCHEME. NEW SCHEME
EXISTS BUT STILL BEING FINALISED. Yesterday John
forwarded me a scheme Magnus proposed which looks
very similar to new scheme.
5
Hardware count
e- leaf ?-
Cables 12
Source Card
Cables 34
Source Card
jet leaf ?-
?- normal
Source Card
Cables 5
?- normal
Source Card
Cables 6
?- normal
? duplicated
?- duplicated
? duplicated
Considers 1 region in (say) ?- and f
?- duplicated
6
Hardware count (cont.)
4 source cards per RCT crate
72 source cards total
Electron leaf cards unaffected
still 27 fibres to each
Each jet cards now takes 10 fibres per unit f
now 30 fibres to each (was 24)
234 fibres total (excluding muon fibres)
7
Old Layout
8
New Layout (if 4 crates)
9
Source card current
If regulators are running flat out then each card
could draw 12A. 18 cards in a crate
216A per crate flat out. Could expect to draw up
to 5A. 18 cards in a crate 90A per
crate expected. Madison crates that they want us
to use are rated 50A ie no more than 10 cards per
crate or 8 crates required. If can show that
current no more than 4A per card then up to 12
cards per crate or 6 crates required. These
calculations dont include USB hubs
either! ??Could Wiener crate handle cards
(Crates I found are rated 115A)?? Would still
need five.
10
Conclusion
Signal duplication at source card works in
theory. Serious issues of location/power supply
need discussing.