AVR-Chat

On Saturday 21 May 2005 14:39, John Samperi wrote:

> From what I understand so far the valve controls gas to a burner
> which needs to keep a temperature of about 450 degrees celsius
> (flaming hot!!) perhaps +- 30 degrees. A Mega16 (possibly) will
> be monitoring the thermocouples and will adjust the opening of the
> valve to maintain the temperature. So, yes I have some feedback
> on what's happening with the valve perhaps 1 or several seconds later.

You should still be controlling the output, not the input. :-)

You need to understand the process a bit better, I think. The delays
in the system will otherwise create instability. Such instability
may be invisible during normal operations but typically has
actuators moving unnecessarily as they hunt around the set-point.
Nobody wants that because it wears everything out very rapidly.

Look at the thermal constants, etc. Maybe you can measure the
desired temperature (not the flame) directly using a pyrometer (they
can measure changes quickly < 0.10 seconds).

You can also, by knowing the thermal constants, do "predictive"
control based on the previously-measured response of the system and
the present state of flow; especially if the thermal response is
slow.

If you're doing something like (hypothetically of course)
controlling the temperature of a hot-dip galvanizing bath, then the
main perturbation to the system is the immersion of cold steel which
will result in an observed "plummet" in the bath temperature;
especially if the bath contains "minimal" zinc. The rate at which
the temperature initially falls is an indicator of the thermal
capacity of the object being inserted and therefore the amount of
energy that the increase in energy to be provided by the gas flame
increase so that the minimum tolerable temperature is maintained.

It is of course impossible to maintain an exact temperature. Any
control system requires an error signal to attempt to maintain the
required output.

> The valve will be moving a maximum of 1/2 turn, so no high speed
> sensors are needed. An Absolute Position sensor would be nice
> to have to find if the valve has moved or if the valve motor
> has moved in the right direction or not moved at all i.e. faulty
> motor, broken wire, jammed/locked valve etc. as a safety precaution.

Monitoring motor current draw may be sufficient to detect a jammed
valve. Relative positioning is OK as long as you can map flame
temperature against step size (I'm assuming a well-regulated gas
pressure and sufficient oxygen). The "current flame" could probably
be used as a reference point against which to apply the delta.

-- 
/"\ Bernd Felsche - Innovative Reckoning, Perth, Western Australia
\ /  ASCII ribbon campaign | I'm a .signature virus!
 X   against HTML mail     | Copy me into your ~/.signature
/ \  and postings          | to help me spread!