EPICS [no subject]

[Matt Bickley <bickley@cebaf.gov>]

A bug was found at CEBAF today which affects 3.12 CAMAC device support for the
records ao, ai, lo and li.  The problem is associated with the PARM field.
It has to do with the standard problem when specifying the maximum output of an
ADC, namely, should one specify 2^N or 2^N - 1?  There is code in the device
support to allow the user to specify either one, and the device support is
smart enough to figure out what the user meant.  That is, if the user
specified 2^16 for a 16-bit ADC, the device support should set ESLO to
(EGUF - EGUL)/(2^16 - 1), so RVAL = EGUF at ADC maximum value
(assuming EGUL==0)

Unfortunately, that's not what the code did.  Instead, if the user specified
PARM to be 2^16 - 1, ESLO was set to (EGUF - EGUL)/2^16, so at ADC maximum
value, an ao had RVAL = 0.

The fix, for CAMAC device support for the records ao, ai, li and lo:

<            if (!(fsd & (fsd+1))) fsd++;
---
>            if (!(fsd & (fsd-1))) fsd--;

Matthew Bickley                                      Email: bickley@cebaf.gov
Computer Scientist                                   Voice mail: 804-249-7347
CEBAF