CTFE LOOKUP MEMORY PROM's ----------------------------- 28-MAR-1989 The following is the official description of the PROM's that will be used for Energy and Momentum Lookup on the CTFE Revision B card. This is the result of the 12-JAN-1990 meeting. Energy Lookup PROM's -------------------- The input address to an Energy Lookup PROM consists of 8 bits of data from the flash ADC Mux-Latch and 3 bits of page select address. The 8 ADC data bits are typically scaled so that the LSB is 1/4 GeV with a full scale of about 64 GeV. The 8 bits of output data from an Energy Lookup PROM are also typically scaled so that the LSB is 1/4 GeV with a full scale of about 64 GeV. All page select address bits are controlled by independent lines on the Timing&Sync Bus (i.e. none of the memory pages are permanently associated with either the first or second energy lookup). The memory devices used are 16k bit 2k by 8 registered PROM's, Cypress CY7C245A-25WC at $18.47 each. The PROM address pins are used as follows: PROM Address Pin Signal Driving the Pin ---------------- -------------------------------------------------- A0 ADC Data LSB D0 . . . . A7 ADC Data MSB D7 A8 Energy Memory Page Select LSB from TS&S Signal D A9 Energy Memory Page Select MDB from TS&S Signal E A10 Energy Memory Page Select MSB from TS&S Signal J Momentum Lookup PROM's ---------------------- The input address to a Momentum Lookup PROM consists of 8 (or 9) bits of data from the output of the digital adder that combines the EM ADC Mux-Latch data with the HD ADC Mux-Latch data, and 3 (or 2) bits of page select address that come from Timing and Sync Bus signals. The output from the EM + HD adder is 9 bits wide and is typically scaled so that the LSB is 1/4 GeV with a full scale of about 128 GeV. The 8 bits of output data from an Momentum Lookup PROM are typically scaled so that the LSB is 1/2 GeV with a full scale of about 128 GeV. All page select address bits are controlled by independent lines on the Timing&Sync Bus. The memory devices used are 16k bit 2k by 8 PROM's, Cypress CY7C291A-35WC at $8.59 each. The PROM address pins are used as follows: Connections to the Momentum Lookup PROM's when using 9 bits of data from the digital EM+HD adder and 2 page select lines (i.e. 4 maps, each with 512 locations). PROM Address Pin Signal Driving the Pin ---------------- -------------------------------------------------- A0 EM + HD Adder Data LSB D0 A1 EM + HD Adder Data next to LSB D1 A2 EM + HD Adder Data nt,nt LSB D2 . . . . A7 EM + HD Adder Data next to MSB D7 A8 EM + HD Adder Data MSB D8 A9 Momentum Memory Page Select MDB from TS&S Signal L A10 Momentum Memory Page Select MSB from TS&S Signal M Connections to the Momentum Lookup PROM's when using 8 bits of data from the digital EM+HD adder (drop the LSB from the adder) and 3 page select lines (i.e. 8 maps, each with 256 locations). PROM Address Pin Signal Driving the Pin ---------------- -------------------------------------------------- A0 Momentum Memory Page Select LSB from TS&S Signal K A1 EM + HD Adder Data next to LSB D1 A2 EM + HD Adder Data nt,nt LSB D2 . . . . A7 EM + HD Adder Data next to MSB D7 A8 EM + HD Adder Data MSB D8 A9 Momentum Memory Page Select MDB from TS&S Signal L A10 Momentum Memory Page Select MSB from TS&S Signal M The following are short notes about using the lookup memory PROM's in the Calorimeter Trigger Front-End Cards. Low Energy Cutoff, Rectifier Effect and Pedestals ------------------------------------------------- If the lookup table in a PROM is to have a low energy cutoff then there are 3 possible scenarios for setting this up. Only last two scenarios as presented bellow are generally useful. 1. Make the low energy cutoff point within the energy range of the normal noise. For example, cut at 0.75 GeV when the width of the noise is 1.0 GeV. This is guaranteed to produce "rectifier effect" and thus the global sum over the detector will include "positive noise" and none of the "negative noise" to balance it. 2. Make the low energy cutoff point well above the noise. For example, if the normal noise width is 0.5 GeV then cut at 2.0 GeV. Because in this case the cutoff is well above the noise then there is no need to balance the positive and negative noise. Because we do not need to balance the positive and negative noise then we do not need to carry the negative noise through the Front-End cards and the Adder Trees. Because we do not need to carry negative numbers through the Front-End cards and Adder Trees then we do not need to set the calorimeter energy signal on top of a positive pedestal (i.e. we can tune the Front-End cards so that zero energy signal from the BLS results in a zero code coming out of the FADC's. 3. The third scenario is to have no low energy cut in the lookup table. This is not the same as having a low energy cut at 0.0 GeV, i.e. the system must be able to carry negative noise values through the Front-End and Adder Tree cards. To do this the BLS signal is put on top of a positive pedestal so that negative noise from the BLS will still result in a positive value coming out of the FADC's. The sum of all of the positive pedestals can be removed in the last stage of the Adder Tree. How Shall We Allocate the Tables -------------------------------- 1. If we use only one lookup for the Energy Adder Trees then we may want to use 7 tables for different vertex Z positions and 1 table for special conditions (e.g. crossings with main ring noise, or mapped 1 to 1). 2. If we use two Energy Adder Tree lookups then we may want to do one of the following: 1. Use 4 tables for the first lookup and 4 tables for the second lookup. In each lookup we could use 3 tables for different vertex Z positions and the forth table for special conditions. 2. Use 6 tables for the first lookup; 5 tables for different vertex Z positions, and the 6th for special conditions. The second lookup could use the last two tables; one for normal operation and the other for special conditions. 3. If we do not have a problem with main ring noise then we will not need to reserve any lookup memory tables for trigger during special conditions. In this case we may want to use 5 tables for the first lookup and 3 for the second lookup. How many lookup tables do we need --------------------------------- How many lookup tables, spaced along the Z axis, will we need for the Calorimeter Trigger to work well? In making this determination we need to consider not just the change in sin theta as the vertex moves but we must also consider the following: 1. Sharing between Trigger Towers. 2. How well is Level Zero going to do in providing the very fast vertex position information. 3. How projective are the Calorimeter Towers? There is no need for our sin theta to be more accurate than the Calorimeter Towers are projective. 4. Because of the finite size of the Trigger Towers it should not be necessary to have a sin theta resolution better than the spread of sin theta over a tower.