Experimental Physics and
Industrial Control System

I'm looking into using the ASLO and ROFF fields in my ai and ao records.
These records are used for current and voltage values which are in ADC
and DAC counts, and which I'd like to convert to amperes and volts.

Our setup for current (voltage is equivalent) has the following
structure (SP and RB are setpoint and readback, respectively):

Current-SP (ao) -> CurrentConvSP (calcout) -> CurrentRaw-SP (longout and
asynInt32) -> hardware

and

hardware -> CurrentRaw-RB (longin and asynInt32) -> Current-RB (ai)

CurrentConvSP is necessary because, as I understood it, the calc and
calcout records don't have device support. It and Current-RB use another
two PVs in their calculations: CurrGain (conversion factor between
amperes and ADC counts) and CurrOffset (raw current measured when
current is zero).

My intended structure using the ASLO and ROFF fields was:

Current-SP.VAL (ao and asynInt32) -> Current-SP.RVAL -> hardware

and

hardware -> Current-RB.RVAL -> Current-RB.VAL (ai and asynInt32)

My initial attempt at using ASLO and ROFF was with database links to
CurrGain and CurrOffset, which didn't work. It seems the record only
really expects numeric values there. I have worked around this so far by
turning CurrGain and CurrOffset into dfanout records, but this requires
that whenever I add a new record that uses some current information from
the hardware, I update the dfanout outputs. It doesn`t seem scalable and
feels like a mistake would be easy to miss.

These values, especially CurrOffset, can be configured in the field, so
it isn't enough to define a variable for them when reading the template.

*Is there some alternative I`m missing for propagating the values I'd
like to have in all ASLO and ROFF fields?*

On top of that, my IOC has two types of users:

- operators and high level scripts, who use current in amperes
- experts, who use currents in amperes and in hardware units

The experts require being able to control the hardware using hardware
units, and I expected this to be supported by writing into the RVAL
field, but it isn't (caput writes, succeeds, and the value doesn't
change). I had to change my setpoint chain to the one below, to work
around this:

Current-SP (ao and Raw Soft Channel) -> CurrentRaw-SP (longout and
asynInt32) -> hardware

It is still better than the previous version, since the math and input
links don't have to be copied for each new record every time and we have
one fewer record, but being able to write into Current-SP.RVAL seems
like it would be the ideal solution. It would even have the advantage of
supporting updating VAL for the new RVAL, which would work around the
issue we currently face (with the new or old scheme) of writes into raw
setpoints not being propagated to normal setpoints.

*Are writes into RVAL supposed to work / going to work in the future, or
should I stick to the intermediary record, or is there some better way?*

Cheers and thank you for the help,
Érico