When the Instrument Is Lying to You

From Tips of the Week, 29 February 2020, 7 November 2020, 5 December 2020 and 17 April 2021 — written across three outages. Here is a lesson I have now learned four separate times, in four different ways, and it took me those four times to notice it was the same lesson.

Before you believe a measurement, prove that the thing producing it is telling you the truth. Not the circuit — the measurement. The isolation you assumed. The lead you’re holding. The link you looked at.

Chase the reading before you check the instrument and you will spend a shift solving a problem that does not exist.

1. The isolation you assumed

You hear the question asked constantly in our industry: is the solution as proposed really going to solve the problem? It would be embarrassing, costly, and a waste of time to implement something that doesn’t — especially if we could have recognized it from the beginning.

But sometimes the question shows up in a subtler form.

Troubleshooting a circuit while looking for stray grounds, the idea arose to disconnect the ground connection by opening a link — breaking continuity between the two sides of a terminal block. The goal was to separate the circuits and prove which one the extra ground was in.

Opening that one link wouldn’t accomplish it.

The circuits in question were connected in a big loop. The ground connection would be broken at the initial point, but carefully tracing the circuit showed that it eventually fed back in at a downstream connection. It was not fully isolated, as had been thought.

The troubleshooting plan relied on that separation. The results assumed that separation. As initially proposed, a measurement would have produced results that could easily have been misinterpreted — sending the team hunting for grounds in a circuit where none existed, but where they only appeared to.

The conclusion was that several other links, for several other wires, needed to be opened as well to achieve the isolation we wanted.

It can be hard to see this kind of thing from wiring diagrams. It took looking at a schematic or elementary to spot it. It also took a team effort. You need the big picture, and multiple sets of eyes won’t hurt (especially at 3:30 am).

When proposing a solution, go the extra mile to make sure it really will do what you think it will. You may be tempted to rely on intuition. It’s better to perform a total analysis, trace the entire circuit, and let the analysis bring you to the conclusion.

Requesting a sanity check from a coworker is always a good idea. Without counsel plans fail, but with many advisors they succeed.

2. The lead you’re holding

Testing is when you iron out the wrinkles, squash the bugs, and prove the design works as intended.

But weird things can start happening during testing that will make your head spin. Voltage measurements produce strange, unexpected, unexplainable results. Simple continuity checks report open circuits where there shouldn’t be any.

Here’s my tip: when that happens, stop and get a check on the test equipment.

It is entirely possible for the test leads on a multimeter to go bad between tests (Murphy’s Law). That will cause you to draw wrong conclusions from faulty premises. And then you will truly go down the rabbit hole, into a strange and absurd universe where you begin questioning the basic truths of electrical engineering, then Ohm’s law, then eventually the foundations of reality itself.

It can happen.

Don’t chase white rabbits or imaginary problems. Confirm the test equipment is working before getting bogged down in strange results. Dispel illusions first.

3. The link that looks closed

It’s our custom to land field cables on one side of a terminal block and internal wiring on the other. For that reason it makes a great deal of sense to specify terminal blocks with sliding links — the STATES ZWM line is one example.

They’re especially helpful during testing, because they make it easy to isolate circuits. Which makes them helpful during troubleshooting, too. The ZWM line is qualified for nuclear use, which helps eliminate questions about suitability in some applications.

One warning, and it belongs in this article more than anywhere else.

When specifying any sliding-link terminal block, add special work instructions to check that each link is closed prior to the start of testing — and specify that closure be confirmed by a continuity check, not just a visual one.

Sometimes the blocks can be open even though they look closed.

That single sentence is the whole theme of this article in miniature. The link looks closed. Your eyes are the instrument, and your eyes are wrong.

I came up with a bit of poetry one night: “Loose links make outage kinks.” Many people didn’t think it was funny at the time. But I’ve never forgotten it.

4. The knowledge you assume you have

One more, of a different kind.

If asked today for troubleshooting guidance, could you remember off the top of your head how to use a clamp-on ammeter to measure three-phase current? Do you check one line at a time, all three, or both? What quantities would each report?

Thankfully there is usually a video that will get you to the answer quickly — here is one demonstrating exactly this, in under three minutes. Just what the engineer needs.

These days, you could probably get an answer faster from AI. Those were still the dark ages, BAI: before AI. It’s still hard to argue with a good video, though.

Sometimes it’s good to run a test on a bench simply to confirm your theoretical knowledge. We don’t always have access to the equipment — but people who do have probably already made a video.

The common thread

Four situations, one discipline:

• The isolation you believe you have — trace it.
• The meter you’re reading — check it.
• The link that looks closed — buzz it.
• The procedure you’re sure you remember — verify it.

Every one of these cost somebody time before it became a lesson. None of them are difficult. They’re just the things you skip when the outage clock is running, which is precisely when you can least afford to.