• HTR    -    10% spares, $2.5k/card, 16 fibers, 2 channels/fiber, 40MHz
• DCC    -    10% spares, $7.8k/card, same number of cards as VME crates
• HRC    -    10% spares $6.5k/card, ditto
• VME    -    calculate crates based on half detector occupancy, $5.5k/crate, $3k/rack, 3 crates/rack, no spares
• L1 trigger cables  - 2 TPG/twisted pair, 2 channel/TPG, 4 pair/cable, gives 16channel/cable or 2 cables/HTR, $131/cable

Since then, the following has changed:
 

1. HTR costs have gone up due to not having properly taken into account optical/electrical transducers.
• depends upon number of fiber links, 40MHz vs 80MHz, etc.
2. HTR count
• take properly into account total channels per detector half
3. TPG cables
• depends on number of HTR cards, not total number of channels.  the cost has gone up
4. DCC cost has been reduced due to cost scrubbing, mostly of LRB but in general
• from $7.8k to $5.0k
5. HRC cost has been reduced due to reduction in scope
• From $6.5k to $5.0k - this will be a commercial CPU card (~$3k) and surrounding PC board to fanout TTC
6. VME crate and rack count has been more accurately determined
• for example, in baseline we had 10, 6, 8, and 6 for HB, HO, HE, and HF respectively.  Due to roundoff error, when you do the calculation correctly using 4968, 2280, 3672, and 2412 (total 13,332) for HB, HO, HE, and HF we get 2 fewer crates in the HO.

 

Item	"Fast/Dense"	"Baseline-like"	"Fast/Sparse"	"Easiest"	Baseline
Chan/fiber	3	2	3	2	2
Frame rate (MHz)	80	40	80	40	40
Fiber/HTR	12	16	8	12	16
Chan/HTR	36	32	24	24	32
HTR uit cost ($k)	$2.7	$2.9	$2.1	$2.6	$2.5
DCC unit cost	$5.0	$5.0	$5.0	$5.0	$7.8
HRC unit cost	$5.0	$5.0	$5.0	$5.0	$6.5
HTR total cost ($M)	$1.12	$1.33	$1.30	$1.58	$1.16
HB crates: HTR/crate	18	16	18	18	16
Channels/crate	648	512	432	432	512
HO crate:  HTR/crate	16	18	16	16	18
Channels/crate	546	576	384	384	576
HE crates:  HTR/crate	18	15	16	16	15
Channels/crate	648	480	384	384	480
HF crates:  HTR/crate	17	13	17	17	13
Channels/crate	612	416	408	408	416
Total crates (HB,HO,HE,HF)	(8,4,6,4)  22	(10,4,8,6) 28	(12,6,10,6) 34	(12,6,10,6) 34	(10,4,8,6) 28
Total HTR cost ($M)	$1.12	$1.33	$1.30	$1.58	$1.16
VME total cost ($k)	$148	$187	$223	$223	$198
DCC+HRC total cost ($k)	$260	$330	$400	$400	$501
TPG cable total cost ($k)	$96	$96	$96	$96	$73
TOTAL COST ($M)	$1.62	$1.94	$2.02	$2.30	$1.93
Additional cost due to loading	$94	$26	$0	$0
GRAND TOTAL ($M)	$1.71	$1.97	$2.02	$2.30	$1.93

Comments:
 

• As you can see from the above table, the "Fast/Dense" solution is the cheapest.  It is also the one that will require the most work in terms of additional engineering to get the 80MHz chips to work.
• The "Baseline-like" scenario costs just about as much as the real Lehman 2000 "Baseline" because some of the costs went up, some down.
• The "Fast/Sparse" scenario, where we go to 80MHz but reduce the number of channels, causes the HTR unit cost to go down alot, but causes the number of cards to go very high.
• The DCC baseline estimate has full 18 slots of HTR, 576 channels per crate.  This is too close to the theoretical DCC bandwidth.
• All scenarios with the exception of the "Spares/Dense" seem to be able to accommodate this except in the HO, where the crates are full.
• To make things fit in the HO, we'd have to reduce the number of cards by 2 (1/9^th, or 11%), which would necessitate a 11% increase in the number of crates, racks, and DCCs.  However, HO is only about 1/7 of the total, so this is not a big effect on the total cost
• For "Sparse/Dense", we would need to increase the number of crates by 2 in the HB, 2 in the HE, and 1 in the HF, or 5 crates in all.  Due to the fact that there are 3 crates per rack, we would need an additional 2 racks since the HF would have a spare place for an extra crate.  I've calculated the total increase to be $94k to give us a bit of breathing room.