Experimental Physics and
Industrial Control System

Allan Honey <[email protected]> · Thu, 4 Dec 2003 14:52:36 -1000 (HST)

In a more recent update to devAscii, the returned value is written
directly into the record's VAL field, thereby bypassing the RVAL/ESLO 
conversions. To do this the device support must also return the
"don't convert" status. This may be a better solution if you are
creating a device specific support layer.

AH

> X-Authentication-Warning: hotrats.mit.edu: [email protected] 
[127.0.0.1] didn't use HELO protocol
> To: "Mark Rivers" <[email protected]>
> cc: <[email protected]>
> Subject: Re: aiRecord, RVAL vs. VAL 
> Date: Thu, 04 Dec 2003 17:42:35 -0500
> From: "Brian McAllister" <[email protected]>
> X-Mts: smtp
> List-Help: <http://www.aps.anl.gov/epics/tech-talk/index.php>
> 
> 
> >>> On 12/4/2003 at 16:30:38 CST, Mark Rivers wrote:
> 
>   > I agree that having the RAW field be an integer is a limitation.  For
>   > example, I might want to read a voltage from a digital multimeter,
>   > which returns a floating point number, but then have that number
>   > converted to some other units, for example pressure, by the ai record
>   > conversion routine.  As it stands I think device support would have to
>   > do that conversion directly, it could not write into the RVAL field and
>   > ask the record to do the conversion.
> 
> One work-around for this (which is used by devAscii, for instance) is to
> scale the value read from the device up in the device support so that it
> can be represented as an integer while maintaining useful precision, and
> then to scale it back down in the conversion to your desired units.
> 
> - brian

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System