Experimental Physics and
Industrial Control System

Jameson Graef Rollins <[email protected]> · Sat, 19 Jul 2014 10:45:39 -0700

On Fri, Jul 18 2014, "Johnson, Andrew N." <[email protected]> wrote:
> This seems to be a rather unusual case. It's easy to implement this
> behavior using a calc record to do the equality comparison, but
> equality is actually a fairly rare comparison to use for floating
> point values (other than zero). If X is really an integer you could
> use longin or longout records instead, you just have to set the LOW
> and HIGH alarm levels to one either side of X and HYST to zero. If X
> is not an integer and is read from an ADC say, you could calculate the
> ADC values in engineering units adjacent to X and use them for the LOW
> and HIGH fields of your ai/ao record. If X really is a double value
> you could even calculate the next higher and lower double values and
> use them, but you need an understanding of IEEE floating point to do
> that properly.

Hi, Andrew.  Thanks so much for the response.

I imagine we are doing some a bit non-standard.  We use ai records for
system parameters that need to take on non-integer values, such as
digital gains.  As you suggest, we could calculate the next higher/lower
double values and set the alarms to those, but that requires more
intermediate work when setting the alarms.

It seems, though, that this case could be easily covered if either the
HIGH or LOW alarm were changed to strictly greater/less than the alarm
value.  Can you think of some reason this behavior would not be desired?

jamie.

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System