Re: Manipulating time in records

[Luedeke Andreas <andreas.luedeke@psi.ch>]

 Di Maio Franck wrote on 10/21/2010 10:46 AM:
> [...] Clarification: 
> For ITER, we'll have a IEEE 1588-2008 network to provide time accuracy "below 1 micro-second", [...]
> [...] A ns precision for a software interrupt is somehow excessive and microsecond precision would fit [...]
Hi Franck,
interesting: cheap and rather powerful.
I'm pretty sure you are the first one in the EPICS community to use this approach.

Normal timing systems do use a separate interconnection of the IOCs, often a
dedicated fibre network.
A common solution these days is the Event-Generator/-Receiver system from
Micro-Research in Finland.
It provides a synchronization in hardware down to the pico-second range and
allows for software trigger with an accuracy only limited by your operating
system, typically in the microsecond range.
This system is then used as well for clock synchronisation.
Of course nobody needs picosecond resolution in a CW machine.

The only missing component for your system is some software "timer" record.
It will trigger execution of records (and other software) at a given time-stamp
or time stamp pattern.
I think if you use doubles for the time-stamps rather than a vector of seconds
and nanoseconds,
like Dirk suggested, your databases will look simpler:
double arithmetic is easier to understand than vector arithmetic.

How do you generate hardware triggers?
If you create an interrupt and write to a binary output, you would spoil you
clock precision, wouldn't you?


Andreas