Experimental Physics and
Industrial Control System

Michael Davidsaver <[email protected]> · Wed, 13 Aug 2014 10:31:21 -0400

On 08/12/2014 02:18 PM, Ryan Pierce wrote:
...
> Right now I plan to keep an Arduino (possibly a Mega) and a Pi in a
> cabinet that will house the power supply, relays, etc. The Pi would be
> networked, and a laptop or tablet would be the control panel. So lack of
> medm for ARM isn't all that concerning.

This seem reasonable.

> For low level IO, I think I need an Arduino. Flow rate sensors use Hall
> effect chips to generate pulses. With an Arduino I can attach the pins
> to interrupts. With a Pi or BeagleBone Black, I'd need to poll the pins
> in a tight loop looking for changes and could miss pulses, yes?

I agree, polling on a pulse train input probably won't work well.

...
> Also, I might be overreacting, but I figure keeping the low level
> control on an Arduino provides some insulation against Linux system
> crashes. In the case of failure, things will tend to stay in the same state.

IOCs in general are very reliable.  Months of uptime is the norm.  The
limiting factors are usually external conditions (power loss or loose
cable), user interactions (people are random), and bugs in individual
hardware drivers.

Deciding where to put your control loop isn't a decision that epics
makes for you (by design).  Its also fairly easy to change your mind.
Usual making this decision is a trade off between loop time and
developer time.  We like to keep both low.

I'd suggest you start by getting an IOC running with some of your
simpler/slower I/O points.  This will give you something real to use
when exploring the various epics tools.

Michael

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System