07 - Pulses and diagnostics

Two blocks for the things a pump site always ends up needing: a flow meter's
pulses turned into a total and a rate, and a look at the controller itself when
something is odd on site.

A flow meter: PULSE_COUNT

Pins first

Wiring it

VAR RETAIN
  flowTotalL : LREAL;          (* litres, kept over a power cut *)
END_VAR

flow( TOTAL := flowTotalL, CHANNEL := 3, K_FACTOR := 100.0, RATE_PER := T#1m );
(* flow.RATE: litres a minute *)

The meter's plate gives its K-factor: pulses per litre (or per m³, per
gallon). Put that in K_FACTOR and TOTAL and RATE come out in that unit.
A meter that gives one pulse per 10 litres has a K-factor of 0.1.

What it does

• The chip counts, not the scan. The ESP32-S3's pulse counter sees every
  rising edge, so a fast meter loses nothing however long the scan is. Four
  inputs can be counted at once (the chip has four counters); a fifth reports
  UIO_TOO_LARGE. The input still reads normally on DIGITAL_IN.
• The total is yours. Each scan's new pulses (divided by K_FACTOR) are
  added to TOTAL. Make it RETAIN and it carries over a power cut; write 0
  to reset it. It is an LREAL because a REAL total of a meter that has run for
  years loses its small increments. Pulses that arrive while the PLC is off are
  not counted.
• The rate suits fast and slow meters. Fast pulses are counted over at
  least WINDOW, from one pulse to another, so the answer does not jitter
  with the scan. Slow pulses use the time between them: one pulse every ten
  seconds reads a steady 0.1 a second, not 1 then 0. When the pulses stop, the
  rate falls away (it can be no more than one pulse over the time since the
  last) and is 0 after ZERO_AFTER. Set ZERO_AFTER above the longest gap
  you expect at the lowest real flow.
• Contact (reed) meters bounce for milliseconds. The hardware filter tops
  out at about 12 µs, so it cannot remove that. Use a meter with an
  open-collector (solid-state) output, or an RC filter on the input.

The controller: SYS_INFO

Pins first

info();
STS.heapMin := info.HEAP_MIN;
STS.resetReason := info.RESET_REASON;

No inputs: call it once, anywhere. It reads the heap once a second (the
figures walk the heap, so not every scan).

Reading it

STACK_MIN covers every IEC task (each is a FreeRTOS task of that name when
the board runs in RTOS mode) and the firmware's own service tasks
(plc_dispatch, plc_modbus, plc_protocols). Without RTOS mode it is the
task SYS_INFO runs in.

PROGRAM_ID is the MD5 the editor's debugger checks: it tells which program
is on a board.

Alarm on RESET_REASON being a panic or a watchdog: the board restarted by
itself, and nobody would otherwise know.