Experimental Physics and
Industrial Control System

Hello all. We were having problems with Pmac 2-5-16 here at LNLS. The problem was solved by Augusto Horita (in copy), but the behavior was weird and we though maybe someone here knows why it is like that or maybe reporting it in tech-talk would be useful for other people.

The IOC was compiled with EPICS base 3.15.6.

The particular instance that we were having trouble with had a set of axis in the PVs such as:

BL:H:PB02:CS1:m1

the axis was configured with

pmacSetCoordStepsPerUnit("$(PPMAC_CS1_PORT)", 1, 1000)

and its MRES was set to 10E-3.

When initialized, the IOC sometimes showed the right values for .RBV and .VAL fields, but sometimes showed values 10 times bigger than what should be. Sometimes the *10 factor was only in .RBV, sometimes only in .VAL and sometimes in both.

Augusto suspected this was due to failure in pmacSetCoordStepsPerUnit since the default scale factor is 10000 (ten times bigger than what we were using). So we suspected that the function was failing for whatever reason and defaulting the scale factor to 10000 while MRES remained in 10E-3. Indeed, changing MRES to 10E-4 and

pmacSetCoordStepsPerUnit("$(PPMAC_CS1_PORT)", 1, 1000)
 to
pmacSetCoordStepsPerUnit("$(PPMAC_CS1_PORT)", 1, 10000)

Solved the problem.

In the middle of the debug process, we struggled to find any error message about the suspected failure in pmacSetCoordStepsPerUnit.
Is this absence of error message intentional? Is there a particular debug method we should have used?
Has this been reported before?

Thanks for the attention. Best regards,

Marco