[Gb-ccb] 17feb04 CCB telecon minutes

[Martin Shepherd]

On Thu, 19 Feb 2004, Martin Shepherd wrote:
> On Thu, 19 Feb 2004, John Ford wrote:
> > extra complexity invites extra problems.  I suggest we trust the
> > reference signals.
>
> How am I supposed to ensure that this thing works if it unnecesarily
> depends on external conditions which rely on the lack of human error?

Apparently this came across as impolite. I apologise if this was the
case. It wasn't intended. I am just a little exasperated, because the
design seems to be getting more and more complicated, compared to what
I originally designed, seemingly introducing more and more
liabilities, and then when I suggested a simple synchronization
circuit, it got thrown out as being a liability, when the whole point
of it was to make things more reliable.

The fact is that we have gone from a single board with one outer
shielded box and one shielded computer box, to between 8 and 10
individually shielded modules, all with their own fallible connectors
and cables, 4 separate FPGAs which now need to be synchronized, where
no synchronization issue existed before, the need for USB 2.0, which
requires a more powerful CPU and will rely on less mature USB 2.0
support in Linux, and the need for more complex software to deal with
the 4 USB interfaces.

Martin