Re: Controls & Physics

["Steven M. Hartman" <hartmansm@ornl.gov>]

Korhonen Timo wrote:

> The calculations are put into a layer of EPICS records that uses the 
> hardware layer records as inputs and outputs.
> Basically we work(ed) together with the physicists and translate(d) some 
> of the most used applications into epics
> templates; those have then bunches of calcout records for the formulas 
> and other stuff to distribute/concentrate the values
> as needed. One can achieve quite neat results with epics records....
> Once this is done, the physics parameters are available for all 
> applications. For changing the parameters you then need
> just a simple screen - medm panel for instance.
> Of course, all this can also be done using a server. I think however 
> that the implementation in the records has paid off.

That is similar to the approach I took. PVs for beam energy (MeV), 
quad focusing strength (m^-2), betatron tune shift, orbit 
correction (mrad), undulator K/gamma, etc. are the PVs which are 
available at the operator interface layer (e.g. medm). The logic 
for translating this to engineering units (e.g. power supply Amps) 
is all contained in the EPICS database layer. In most cases, it is 
not a one-to-one mapping. For instance, the betatron tune shift 
knob drives 9 families of quadrupoles involving 16 (if I recall 
correctly) power supplies. The wiggler knob drives even more power 
supplies with built in feed-forward compensation for orbit shifts 
and tune shifts. And the beam energy knob touches nearly every 
magnet power supply (with the quadrupoles, for instance, 
maintaining the same effective focusing strength for the different 
energy beam).

Within the EPICS database, it is basically an ao record for a 
high-level parameter, a bunch of calc, calcout and fanout records, 
breakpoint records (for non-linear mapping), and then to ao 
records to drive the hardware, with event records used for 
synchronization.


--
Steven Hartman
hartmansm@ornl.gov || 865-466-6473