Estimate page for FNAL nu detector

(a/k/a Daniel Boone -- concept drawings from HF/Chicago)

Revised 2-NOV-2011 // GSV

1 meter cube detector

Baseline assumptions:

1 row of 5 tiles form a module
5 modules per 1m^2 side
6 sides to the cube, for 30 total modules
Each module has 40 strips (at current pitch)
Readout at both sides: 1 module = 80 readout channels
Total (minimum) channel count is 30 * 80 = 2400 channels
Requires 600 4-chan, 300 8-chan and 180 16-chan (since 8 wasted channels/side/module)

A number of ASIC options exist, some number of which are captured in Table 1.

Table 1: A summary of potential ASICs for this application. Costs in blue are based upon quotations received 28-FEB-2011 from MOSIS.

ASIC name	Amplification?	# chan	Depth/chan	Sampling [GSa/s]	Time resolution	Analog BW	Vendor	Size [nm]	Ext ADC?	Cost
PSEC4	no.	6	256	5 - 15	TBD	TBD	IBM	130	no.
PSEC3	no.	4	256	1-16	TBD	0.3-1.5 GHz	IBM	130	no.	55k
DRS4	no.	8	1024	1-5	10's of ps	0.35-0.75GHz*	IBM	250	yes.	? ? (ADC also)
SAM	no.	2	1024	1-3	10's of ps	350MHz*	AMS	350	yes.	? ? (ADC also)
IRS2/IRS3	no.	8	32536	1-4	TBD	>= 1GHz	TSMC	250	no.	$ 47,500
BLAB3A	yes.	8	32536	1-4	TBD	350MHz	TSMC	250	no.	$ 47,500
TARGET	no.	16	4192	1-2.5	<40ps	150MHz	TSMC	250	no.	$ 41,500
TARGET2	yes.	16	16384	1-2.5	TBD	350MHz	TSMC	250	no.	$ 41,500
TARGET3	no.	16	16384	1-2.5	TBD	>= 1GHz	TSMC	250	no.	$ 41,500

For simplicity, can assume there will be 1 fiber optic/Giga-bit fiber transceiver per module. This can be handled by 8x quad-pair receiver cards, which are probably 2.5k$ each in small quantity. Extrapolation (modulo inflation) to larger numbers of channels can be done for the TSMC 0.25um process per the curve in Fig. 26 of NIM A591:534-545 (2008) [PDF link]

* = all of these ASIC analog bandwidths can be improved by adding an external, low-impedance output buffer amplifier

Last modified: 11/22/2011 -- GSV