|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
In POL, skew is an important problem, limiting
the maximum data-rate of parallel transmission. |
|
Skew is generated by: |
|
the optical converters, |
|
the integrated electronic circuits, and |
|
the optical fibre (in this case fortunately SM) |
|
In ribbons, skew is mainly generated by
variations of the reflection index, close to 1.5, between the fibres.
Indeed, speed of transmission in the fibre is in function of its reflective
index. |
|
We have thus asked BICC for ribbon fibre where
each of the 8 fibres was drawn from the same bath. Very little change is
now expected in speed of propagation between the different fibres of the
ribbon. |
|
Price remained unchanged, no extra manipulations
were needed during the fabrication process. |
|
|
|
|
|
|
|
|
|
|
|
|
Bit errors are generated differently in short
distance digital parallel optical links and in long-haul single fibre (PTT)
transmissions. |
|
Where in PTT-type applications the detected
light signals are sometimes so small that any noise can cause bit-errors,
at CERN, parallel transmission signals remain always (exept dramatic
failure) sufficiently large. Errors are mainly generated by skew problems
at high transmission frequency. |
|
However, we tried to give a basic estimation of
BER under possible working conditions at CERN: |
|
|
|
|
|
|
|
|
pulse broadening detects single error by
latching the set-up to level 1, |
|
pulses of less 1ns can be transmitted and will
be detected, |
|
cascading 7 links permits to speed up BER
measurement time by a factor of 7, |
|
severe working conditions are generated by
injecting a 1Gb/s signal on remaining chan-nel, causing max. probability
for crosstalk, |
|
no error was detected after 72 hours, giving an
estimated BER of: 1 / (72 * 3600 *
10^9 * 7) |
|
BER for Hitachi POLC better than: 5.10E-15, this @ 25C. |
|
|
|
|
|
|
|
|
|
|
|
Notes
