Measure Before You Replace

The pump will not start. The technician replaces the starter, then the overload, then the relay, then the input module, and after two hours and four parts it starts. Nobody knows why, the four old parts go into a box marked suspect, and the station fails the same way a month later. This is not a story about a bad technician. It is a story about a method that most of the trade learned by watching, and it costs utilities more than any other single habit.

Why swapping fails

Replacing a part answers one question: is this specific part the cause. It does not rule out anything else, it disturbs wiring that was fine, it consumes a spare that may be the last one, and when the fault is intermittent it produces false conclusions, because the fault goes away for a while whether or not the part was the cause. Worse, it teaches nothing. A fault found by measurement leaves the technician knowing the system better; a fault fixed by swapping leaves a box of suspect parts and a station that will do it again.

The method

1. 1

   State the symptom exactly

   Not the pump will not start, but the run command is on at the controller output and the starter coil is not pulling in. The exact symptom already excludes half the system.

2. 2

   Ask what changed

   Work done recently, weather, a power event, a new device. Most faults have a cause that arrived in the last week.

3. 3

   Read what the system already knows

   Indicators on the panel, fault codes on the drive, the controller fault record, the alarm log, the trend of the value over the last day. This costs nothing and often ends the search.

4. 4

   Pick the measurement that splits the problem

   Between the controller output and the starter coil there are a relay, a fuse, a terminal, and a wire. Measure at the middle: voltage at the relay contact. Present on both sides, the fault is downstream; present on one, the relay; absent on both, upstream.

5. 5

   Measure, then move

   Each measurement moves the boundary. Keep going until the two sides of one interface disagree: voltage into a terminal and none out, current into a module and no count in the tag.

6. 6

   Confirm the cause

   Before replacing, make the cause explain everything: the symptom, the intermittence, the timing, what changed. A cause that explains half is a coincidence.

7. 7

   Fix, verify, record

   Replace or repair the one thing, prove the symptom is gone by the same measurement that found it, and write down what was measured and what was found.

Splitting in half

The power of the method is in the choice of measurement. A control circuit from an output module to a starter coil has perhaps eight places it can be broken. Measuring at each in order takes eight steps; measuring at the middle, then the middle of the half that failed, takes three. The habit is to look at the path, find its midpoint, and put the meter there. It applies to everything: a 4 to 20 milliamp loop is split at the panel terminal into field and panel; a communication failure is split at the switch into the controller side and the device side; a wrong reading on the screen is split at the controller tag into the field side and the SCADA side.

Reading before measuring

Modern equipment reports on itself, and reading the report is the fastest measurement there is. A drive that tripped says why on its display. A controller keeps a fault record with the routine and the rung. A managed switch counts errors per port. A transmitter reports open loop and out of range. An hour of a technician tracing wires is often preceded by a fault code on a display that nobody read because the display was behind a door. Reading is measuring with the instrument the manufacturer built in.

When swapping is right

Substitution is a measurement when it is done as one: a known-good radio in place of a suspect one, with the signal reading before and after, answers a specific question. Replacing a fuse to see if it blows again, with the circuit sectioned first, answers a question. The difference between substitution and swapping is whether a question was asked. Swap a part to test a hypothesis you formed from a measurement; do not swap parts to form one.

The intermittent fault

The fault that comes and goes is where swapping does the most damage, because anything replaced appears to fix it for a while. The method for an intermittent is to catch it in the act: a recording meter on the suspect supply, a trend on the tag with a fast sample, the port error counters cleared and watched, a wiggle test with the meter on the terminal. The measurement that shows the fault happening is the only one that counts, and it usually takes patience rather than parts.