Re: rapid scans

[David Maden <David.Maden@psi.ch>]

D. Peter Siddons wrote:
> What aproaches are people taking to performing on-the-fly scans? What I 
> mean is, a scan in which one launches a motor move and reads a detector 
> as the motor is moving, rather than the more typical step-scan.
> Pete.

I presented a talk on how we do this at the powder diffractometer of the 
SLS Material Science Beamline at the EPICS meeting at Bessy in May 2002. My 
slides are probably still on the web somewhere.

During a scan, we have a program (data taker) running in the ioc. It is 
driven by watchdog interrupts to minimise timing jitter. This reads either 
the encoder or the motor step counter (depending on whether the motor has 
an encoder or not) and 6 scaler values directly from the VME registers to 
minimise overhead. The read frequency is specified by the user as a number 
of ticks with the tick rate of the ioc set to 4 msec.

The data taker buffers the data via a vxWorks pipe to a data writer and 
this, in turn, writes the data in ASCII SDDS format to an NFS-mounted file.

The whole thing is glued together with EPICS soft channels, medm panels and 
a python script.

An 8 hour run with a 10 Hz sample rate generates something like 25 Mbytes 
of data, if I remember correctly. The data is then rebinned offline.

David

--
                               David.Maden@psi.ch
     _/_/_/   _/   _/_/_/     Beamline Control Group, SLS Project
    _/       _/   _/         Paul Scherrer Institute
     _/_/   _/     _/_/     CH-5232 Villigen-PSI
        _/ _/         _/   Tel: +41 56 310 3616
   _/_/_/ _/_/_/ _/_/_/   Fax: +41 56 310 3294