Experimental Physics and
Industrial Control System

Ned Arnold <[email protected]> · Thu, 17 Mar 2005 07:13:42 -0600

Two suggestions ...
1) Since you are looking for a transition, choose a TTL input module
whose inputs can be configured to generate an interrupt on a
transistion, have this interrupt process a calc record that simply
increments, and have another calc record that makes sure the first calc
record increments at least 100 times over a second (or 10 times every .1
second).
2) OR, add a little FPGA circuit (or a timer relay) that expects a
transition every 10 milliseconds (a simple watchdog) and use this output
to drive the digital input. This is preferred because it eliminates
100Hz processing of records.

 Ned

Emmanuel Mayssat wrote:

I am convinced that some of you were faced with the same situation where
you had to read a timed input with an asynchronous device.

Well, apparently I am no different and would like to know what your
suggestions are.
Here is my situation:
The input signal is binary and periodic with period T=10ms.

The duty cycle is ~60% ( or the TTL level high is maintained for 6ms
while the TTL level low is maintained for the remaining of T)
If the signal becomes aperiodic ( always TTL high or TTL low ) then
there is an error ( wrong cable connection or device error )
I have an epics driver for the asynchronous device (binary input).
Here are the solutions I came across:
1/ The obvious one: write a new epics driver that will query the binary
input 10 times over 10 ms. If I get at least a 1 and a 0 then everything
is ok => pbi->val=1
2/ I looked at soft records that could run on the top of the existing
INSTIO record. That is my prefered way.

I came accross the seq record in the record manual. Is it an appropriate
use for it? Am I on the right direction?
Do you have any other suggestions?
Thanks,

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System