VME Fifo modules for the OD-DAQ: C.A.E.N. V533

Pipeline FIFO buffer unit 256 x 32 bits (expandable to 4096 x 32 bits)

Documentation links:

  1. C.A.E.N. model V533 product description and data sheet
  2. Fifo-loop subroutine flow charts for the Anti-Collector software (by J.George):
  3. This document:

Table of data and flag input usage and VME addresses

This is the current setup of the Fifo input channels at the OD-DAQ since the upgrade to version 2 [8/1/96]:

Fifo #VME slotbase addresstrigger inputdata input bitsdata / flag name
0slot 50xF10000 G + B (*)0...31Local Time
1slot 8 0xF00000G + B (*)0...15Event Number
16global trigger (long) (*3)
17End-of-BIP flag
18GPS fine-tune flag (*4)
19global trigger (40 nsec) flag
20...23BIP hut 1...4 (inverted)
24AUX bit (*4)
25Veto flag (inverted)
26Reset Veto flag (inverted)
27...31-- free / not used yet --
2slot 60xF20000G + B (*)0...31Local Time (*2)
where:
(*)
= Global Triggers combined (OR) with End-of-BIP Triggers.
(*2)
= In the future, this will be dedicated to time-stamp the End-of-BIP Triggers only, in order to measure the TDC deadtimes. Currently, this fifo is only used for backup of Fifo 0.
(*3)
= This is the original trigger signal from the Event Number cable (17th bit), as originated from the EVENT COUNT DRIVER box. It's of approx. 250...350 nsec length.
(*4)
= Not implemented or used yet with the current OD-DAQ version.

Modifications on the V533 modules for the SuperK OD-DAQ

Note: Two modifications had to be done with the original C.A.E.N. V533 boards before they could be implemented into the OD-DAQ system!
  1. Each Fifo module was delivered by C.A.E.N. with a CERN-specific "Paux" connector, an additional special 30-pin DIN connector for the VMEbus backplane in order to supply the board with -5.2V for the ECL and NIM circuits. Standard VME crates on the US market were not equipped with the counter parts nor an additional -5.2V power supply, and the European versions we found were far more expensive than the standard crates. Instead, we found a simple solution by removing the Paux connector and replacing it with a $1 voltage regulator IC (type 7905) to derive the needed -5.2V (-5V) from the -12V standard supply voltage. See photo below for the position of the regulator on the board.

    simple -5V regulator circuit

  2. The termination resistors R1 and R4 - both 51 Ohms - for the NIM inputs TRIG and REFCLK had to be removed in order to allow hooking up the trigger/clock signals in a serial chain. The appropiate termination is then done with an external Lemo 50-Ohm Terminator.
modified V533 board


How to expand the fifo buffer depth

With the original manufacturer setup, the V533 module consists of two 256 x 32-bit fifo buffers. The first one is a continuously running circular buffer, latching 32-bit input data with each new REFCLK pulse (50 MHz). The second buffer then reads 3-15 words (programmable amount) from the first buffer with a programmable shift of 0-255 clock ticks at each TRIG pulse. In other words, the first buffer acts as a programmable delay and the second buffer is an "event buffer".

For the OD-DAQ the event buffer latches three 32-bit words at each trigger. Thus, with a fifo depth of 256 words, we can save up to 85 trigger events (*) without reading it out through the VMEbus. If more than 85 events occur between VME readouts then the module sets a BUSY flag, telling the user that the buffer is full and more events are ignored from now on.

In order to keep up with higher trigger rates and trigger bursts at SuperK we soon have to expand the fifo depth of the event buffer. These are the positions U26...U29 (replace IDT72201 with IDT72241) and position U33 (replace IDT7201 with IDT7204). Latter IC is used as an internal event count register buffer of the V533 module. With the new ICs the event buffer is then 4096 x 32-bit wide, allowing us 16 times as many events - 1365 instead of 85 (*) - to pile up before a BUSY state occurs again!

V533: expand to 4k-word fifo depth

(*)
actually, in the current OD-DAQ hardware the Fifo Trigger inputs are fed with an OR of the Global Triggers and the "End-of-BIP" triggers. Latter ones occur each time at the end of a TDC busy period and are used to measure the deadtimes of the Fastbus modules. In average, almost every Global trigger is followed by a End-of-BIP trigger a couple of microseconds later. Thus, a fifo buffer of 85 events means in average a maximum of only 42 Global Trigger events can be allowed between consecutive VMEbus readouts.
Dito for expanding the buffer to 4k words means, we'll then have a Global Trigger buffer capacity of approx. 682 - which is still 16 times more than before ...

last edited 3/31/97 [HGB]