(1) Igor Mandic (Oxford) reported on Bit Error Rate (BER) Tests of Data Links.
    The test system used LDC to drive an LED in a GEC opto package and SEQSI and
    LITMUS modules to measure BER. BER of less than 10**-9 could be achieved
    with LED optical powers of about 2 microW (cf 17 microW average before
    irradiation).
    Igor also reported on BER measurements of a TTC link. The link used the
    the BiLED chip to encode BiPhase Mark data. The data was received
    by a PIN diode and DORIC mounted in a GEC opto package. BER measurements
    were also performed with SEQSI and LITMUS.
    The results showed that
    (a) sending 2 "1"s followed by 14 "0"s worked with BER < 10**-9 provided
    the timing was adjusted carefully.
    (b) sending 3 "1"s followed by 13 "0"s worked with BER < 10**-9 only over a
    tiny range of timing
    (c) sending psuedo-random data the BER was never below 10**-4.
    
    These problems are all believed to be due to timing; the DORIC chip only
    accepts data if the appropriate transitions are within a 2ns gate. This
    problem will be fixed in the next itteration of DORIC by opening up this
    gate width (see talk by Marcus French).
    
(2) Marcus French (RAL) discussed the problems with DORIC. DORIC works
    correctly if the input signals have perfect timing as demonstarted by Dave
    White's tests at RAL and also by the tests at Oxford (see item 1 above).
    The very narrow gate clearly makes DORIC very difficult to use in practice
    and so the solution will be simply to widen that gate from 2ns to about
    10ns. Since we will use VCSELs for the TTC distribution, the light levels
    will be much larger. The new DORIC will be optimised for these light
    levels.
    
(3) Igor Mandic (Oxford) reported on VCSEL driver studies. The waveformed of
    Honeywell VCSELs driven by the BiLEd and LDC chips were studied with an
    optical probe. Some undershoot and overshoot of the signals were seen but
    these could be minimised by prebiasing the VCSELs with about 4mA (the
    threshold was 5mA). Dave White will perform a more systematic study of
    VCSEL drivers to determine the optimum circuitry to use for driving VCSELs.
    
(4) Tony Weidberg (Oxford) reported on results GEC had with MITEL VCSELs in
    their Silicon opto package. It was easy to obtain >500 microW optical power
    at 10 mA drive current. There were no modal problems. Good results were
    obtained for BER tests at 100 MHz and 1 GHz. A package with two VCSELs has
    now been provided to us for our tests.
    
(5) Radiation Tests.
    (a) Gilles Mahout (Birmingham) reported on the status of the high staistics PIN diode
    irradiation program. 15 tiles, each with 8 epitaxial Si PIN diodes have
    been delivered. They will be used as follows
    (1) 3 tiles kept for reference
    (2) 12 tiles will be irradiated at ISIS with 2 10**14 n/cm**2
    (a) 6 of these will then be irradiated to 6 10**14 n/cm**2
    (b) 6 of these will then be irradiated to 3 10 **14 p/cm**2 at the PS.
                  
(6) Links for System Tests in '98.
    (a) Tony Weidberg reported on the availability of parts for this work. Pedja
    reported on the dog-leg cables and opto-boards in another meeting later in
    the week. The first of the 16 GEC LED/PIN package has now been delivered
    and the rmaining 15 will be delivered in 4 weeks. Pete Shield (Oxford) has
    started a design of a very low jitter and adjustable BiPhase Mark encoding
    board. It should be possible to make the TTC links work with low BER with
    this board.                    
    (b) Alan Rudge (CERN) reorted on the status of the opto-reciver. The module
    will use the AME 8 channel PIN array which can be connected to an MT-8
    fibre ribbon. Simple commerical chips are used for the pramp and
    discriminator (see transparency copies for details). The boards will be
    avalible for BER tests in May.
 
(7) John Dowell (Birmingham) discussed the question of fibre instalation.
    Robert McLaren (CERN) will look at the fibre routing from the patch panel
    to the RODs. We need to investigate a second source of rad-hard fibre.
    There should be a first design of fibre routing for the June SCT week.
         
(8) Gareth Noyes (RAL) reported on the SCT links project plan. The project is
    divided into three sub-projects
    (a) System Tests '98
    (b) Development
    (c) Production
    A clear interface between the links and the off-detector electronics needs
    to be defined so that detailed engineering work can proceed. A more
    detailed project plan should be available for the next SCT week in June.