Time in states

After a gas accident in the lab, a button is to suck the room empty. The vent sucks until the pressure is below 5 kPa and then keeps running for another 5 seconds so the rest ends up in the pipe as well.

Your machine does not answer two questions yet: How long has the run-on been going? And what if the pressure never drops below 5 kPa – because a door is open or the tank behind the vent is full? Then the vent sucks forever, and nobody notices.

Counting ticks#

A tick lasts 0.5 s. A register ticks counts up by one every tick:

define RUNTIME 10       # ticks = 5 s run-on
define MAXTIME 40       # ticks = 20 s, then fault
alias ticks r2
add ticks ticks 1       # one more tick in this state

Write the times as ticks into the define and the seconds into the comment. Whoever changes the value later sees straight away what is meant.

Waiting time per state#

The run-on simply waits until the counter is big enough:

s_runon:
sge tmp ticks RUNTIME
select next tmp READY next
j output

Detecting a timeout#

Pump-down has two exits: success (pressure below 5 kPa) and timeout. Check the time first and then the success. The last select overwrites next, so success wins within the same tick:

s_pumpdown:
sgt tmp ticks MAXTIME
select next tmp FAULT next
slt tmp pressure EMPTY
select next tmp RUNON next
j output

Resetting the counter on a transition#

Every new state has to start at zero. The safest place for that is one spot, namely where the new state is taken over:

output:
seq tmp next state         # state unchanged?
select ticks tmp ticks 0   # otherwise counter to 0
move state next

That way no transition can forget the reset, not even one you add next month.