Lesson 3 — Loops and Acting Direction

Lesson 3 · Objective

Two More Ideas Before We Start Tuning

Before we get into actually tuning P, I, and D settings, there are two ideas you need first. One is a distinction you've already been using without a name attached to it. The other is something that trips up even experienced technicians, because it's easy to treat as a memorized rule when it actually has to be worked out fresh for every single loop.

Part 1

Open Loop vs. Closed Loop

Does the control system actually measure the result and correct for it — or does it just run and hope? Every PID system in this course is closed loop. Open loop is the thing it's not, and it's worth seeing clearly what that alternative looks like.

Part 2

Direct-Acting vs. Reverse-Acting Control

Should the controller's output go up or down when the process variable rises? The answer isn't a property of "the type of process" — it depends on the physical arrangement of that specific loop, and getting it backwards is a real, safety-relevant mistake.

By the end of this lesson you should be able to look at an unfamiliar loop — a valve, a pump, a tank — and reason your way to the correct controller action from scratch, rather than recalling it from memory.

Part 1 · Open Loop

Open-Loop Control: No Feedback

Open-loop control sets an output and lets it run — with nothing measuring the actual result and adjusting for it. The controller (or a person) decides what the output should be ahead of time, and it just executes that decision, blind to whatever actually happens to the process variable.

You've probably used one of these already. A microwave set to "3 minutes" runs the magnetron for exactly 3 minutes regardless of the food's actual starting or finishing temperature — it never checks whether the food's already hot. A sprinkler on a timer works the same way — you've probably seen one of these too — running for 20 minutes because that's what the timer was set to, never checking whether the lawn is already soaked from last night's rain or bone dry from a two-week drought. Both devices commit to an output ahead of time and never look back to see if it actually worked.

Dashed red line = the connection that doesn't exist in an open-loop system.

Open loop doesn't fail because it's bad engineering. It simply cannot correct for anything it doesn't measure. If the lawn was already soaked, the sprinkler runs its 20 minutes anyway — there was never a sensor in the loop that could have told it otherwise.

Part 1 · Closed Loop

Closed-Loop Control: Feedback Changes Everything

Closed-loop control measures the actual process variable and continuously compares it to setpoint, adjusting the control output to correct any error. This is what every control system in this course has actually been — including the ones that didn't look like it yet.

Go back to Lesson 1: on/off control (the thermostat) measures room temperature and reacts to it — that's feedback, that's closed loop. Continuous PID control also measures PV and reacts to it — also closed loop. On/off and PID differ in how they react (slamming between two extremes vs. easing smoothly), not in whether they react. Both of Lesson 1's control methods live on the closed-loop side of this new line we're drawing today.

Green feedback line = the connection that makes this a closed loop. Compare directly to the broken loop on the previous page.

Closed loop is generally what real process control needs — a plant can't just hope a tank level or a reactor temperature stays where it should. But open loop still has its place: it's simpler and cheaper, and it's perfectly fine when precision genuinely doesn't matter, or when feeding a measurement back isn't practical or worth the cost. From here forward, though, every lesson in this course — including the sign-of-the-controller-output question on the next page — lives entirely inside a closed loop.

Foundational Concept

Manual vs. Automatic Control Mode

Before you ever tune a single value, there's a switch on every PID controller you need to know about — Auto and Manual. Every trainer you'll use in this lesson runs in Auto, where the controller's own math sets the output. But that math can be switched off entirely, handing direct control of the output straight to a person.

Auto

CO = controller math

The PID algorithm calculates CO every cycle from the error between SP and PV. This is the normal running state — what you've seen in every trainer so far.

Manual

CO = operator

A person sets CO directly by hand — a slider, a number entry. The controller's calculated output is ignored until it's switched back to Auto.

This switch exists on every real PID controller — a DCS faceplate on a control-room screen, a PLC's HMI, or a panel-mounted single-loop controller bolted to a wall with nothing but its own buttons. That's a thing: an operator or technician can walk up to any one of them and take direct control of the output by hand.

Why would anyone do that? Real, practical reasons — during startup, before the process is stable enough for automatic control to make sense; during an abnormal situation; when an instrument is acting up and the reading can't be trusted; for maintenance or troubleshooting; or simply because an operator needs to intervene directly for a moment.

Bumpless transfer: switching between Auto and Manual should never cause a sudden jump — a "bump" — in the actual output signal. A well-designed transfer keeps CO continuous across the switch, in both directions. A bump isn't just untidy — it can slam a valve open or closed, or spike the process, right at the moment someone is trying to take careful control. Keep this term in mind; it comes back again later.

Try it below: watch the trainer run in Auto, switch it to Manual and move the output yourself, then switch back to Auto and watch the transfer happen cleanly.

Trainer — Level Control, Inlet Valve (Manual vs. Automatic Mode) Same inlet-fill-valve tank used by the trainers ahead — this one is only about the Auto/Manual switch.

Part 2 · Definition

Controller Action: Direct-Acting vs. Reverse-Acting

Inside a closed loop, the controller constantly asks: "PV just moved — which way should I move CO to fix it?" The answer to that question is the controller's action, and it comes in exactly two flavors:

Direct-Acting

PV ↑ → CO ↑

As PV goes up, CO goes up too. They move the same direction.

Reverse-Acting

PV ↑ → CO ↓

As PV goes up, CO goes down. They move opposite directions.

That's the whole definition. The hard part isn't remembering the definition — it's figuring out, for one specific real loop, which of the two actually applies.

The critical, easily-missed point: which one is correct is not a fixed property of "the type of process." A statement like "level control is always direct-acting" is simply wrong. The correct answer depends on the actual physical arrangement of that specific loop — which final control element is used, and exactly where it sits in the process. The very same control objective — holding a tank's level, say — can require opposite controller-action settings depending on that arrangement. There is no shortcut rule to memorize here. You work it out, every time, from the physical arrangement — the next few pages show you the method.

Part 2 · Worked Example 1 of 3

Level Control With an Inlet Fill Valve

A tank is fed through a control valve on its inlet line. The tank also has some steady outflow (a fixed drain, or a downstream demand) that isn't being throttled. The controller's job: hold the tank's level at setpoint by adjusting that inlet valve.

Reasoning It Through
Step	Reasoning
1	What happens to the level if the inlet valve opens further? More flow in, nothing else changed → level rises.
2	So (valve % open) → (level) move the same direction — opening the valve raises PV.
3	Assume the ordinary case for now: more controller output (CO) drives the valve more open. So CO → valve-open is also the same direction. Combined: CO up → PV up.
4	If PV (level) drifts too high, what must CO do to correct it? It must go down — close the valve to stop overfilling.
5	PV up requires CO down → opposite directions → this loop needs a REVERSE-ACTING controller.

Answer: Inlet fill valve, ordinary valve wiring → reverse-acting controller.

Part 2 · Worked Example 2 of 3

Same Tank, Same Objective — Now a Discharge Valve

Same control objective — hold this tank's level at setpoint — but now the control valve sits on the discharge (outlet) line instead, with steady, unthrottled inflow feeding the tank from somewhere upstream.

Reasoning It Through
Step	Reasoning
1	What happens to the level if the discharge valve opens further? More flow out, nothing else changed → level falls.
2	So (valve % open) → (level) now move opposite directions — opening this valve lowers PV, the sign flipped from Example 1.
3	Same ordinary valve wiring as before: CO up → valve more open. Combined: CO up → PV down.
4	If PV drifts too high, what must CO do now? It must go up — open the discharge valve further to drain the excess.
5	PV up requires CO up → same direction → this loop needs a DIRECT-ACTING controller.

Same tank, same control objective, opposite controller action — because the valve is now doing the opposite job. This is exactly why "level control is always direct" or "always reverse" is the wrong way to think about it. The valve's placement decided the answer, not the fact that it's "level control."

Try It Yourself

Reverse vs. Direct Acting Trainers

Try the different setups below and see what happens when you get the direction wrong — reverse when it should be direct, or direct when it should be reverse.

Trainer — Level Control, Inlet Valve Matches Example 1 earlier in this lesson (inlet valve).
Trainer — Level Control, Discharge Valve Matches Example 2 earlier in this lesson (discharge valve).
Trainer — Level Control, Variable-Speed Pump Same tank, fed by a variable-speed pump instead of a valve.

Try It Yourself · Saturation

When the Output Runs Out of Room: Saturation

A controller's output (CO) is just a number the algorithm calculates — but it has to drive a real, physical actuator, and that actuator can only go so far. A valve can't open past 100%, and it can't close past 0%. Saturation is what happens when the controller's math wants to push CO further than that, but there's nowhere further for it to go: CO clamps — "pins" — at its limit, even while the calculated demand behind it keeps growing.

You've already seen the ingredients for this without the name attached. On the previous page, flipping a trainer's Action to the wrong direction made the loop fight itself — every correction pushed CO the wrong way, so the error never stopped growing. Saturation is what that runaway output looks like once it actually reaches the end of its range and gets stuck there. It matters here for one reason worth remembering: a controller that's saturated and accumulating error is exactly the setup for a problem called integral windup, which a later lesson covers in depth. For now, just learn to recognize saturation itself.

What to watch for in the demo below: the CO readout pinning at 0% or 100% and staying there, and the valve graphic visibly reaching its fully-closed (or fully-open) limit and sitting there — even though the controller keeps "asking" for more, the physical valve simply has nothing left to give.

Trainer — Level Control, Inlet Valve (Saturation Demo) Kp preset to 3, Action defaults to Direct (backwards) — watch CO pin and the valve bottom out.

Review

Next up: back to the building blocks — Proportional Control, now that you know both which loops are closed and which direction a controller needs to push.