EC-13.2 The Measurement Bench

What this is and why it exists

This is the first stage of the progression and it deliberately measures nothing interesting.

The subject is the bench itself. Do the instruments agree with each other? Are the connections what you think they are? What does the setup show when nothing is connected?

A reader who can establish those three things has a foundation for every later session. A reader who cannot will produce confident readings all year without knowing which of them mean anything.

The vocabulary

• Bring-up — establishing that a setup works before using it for anything.
• Known reference — something whose value you already know, used to check an instrument.
• Cross-check — measuring one quantity two different ways.
• Noise floor — what your setup shows when there is nothing to measure.
• Coupling — whether an instrument passes the steady part of a signal or removes it.
• Range — the span an instrument is currently set to measure.

The mental model

Setting up a bench has an order and the order exists for a reason.

Supply first, with its current limit set low. Then the generator, with its output switched off. Then the instruments. Then, last, the circuit. Powering a new circuit before the current limit is set is the mistake this order exists to prevent. It is the difference between a supply sitting quietly in limit and a destroyed part.

Now verification, which is the habit this topic exists to build. Measure something you already know, before measuring anything you do not.

A resistor from the drawer whose value is marked. The oscilloscope's own calibrator output. The supply read on the supply's display and on a separate meter. Five minutes of this prevents a whole session of confident wrong readings. It also catches the instrument left on the wrong setting by whoever used it last.

The cross-check is the same idea applied during the work. Measure a voltage on both a meter and an oscilloscope. If they disagree, that disagreement is more informative than either reading alone.

It usually names the problem. A meter reading lower than the oscilloscope suggests loading. An oscilloscope reading lower at high frequency suggests bandwidth. A large disagreement on an alternating signal usually means one instrument is measuring an average and the other a peak.

The noise floor is the number people never establish and constantly need. Record what the instrument shows with the input shorted, and again with nothing connected at all.

Anything smaller than that has not been measured. Knowing the number stops a whole class of imaginary results, and it takes two minutes at the start of a session.

Recording a reading properly means recording more than the number. The value, the units, the uncertainty, the instrument and its range, and the conditions.

A number alone in a notebook is not usable a week later, even by the person who wrote it. This is the same argument the measurement module makes about stored data, and it applies at every scale from one reading to a whole acquisition.

Finally, three mistakes worth producing deliberately once, so that you recognise them later.

Put a ground clip somewhere unintended and watch what happens to the trace. Set an instrument to the wrong range or the wrong coupling and see how plausible the wrong answer looks. Load a generator differently from what its display assumes and confirm the factor of two.

Each takes a minute to reproduce and each saves an hour the first time it happens by accident.

What you should now be able to explain or do

• Set up a bench in an order that protects an untested circuit.
• Verify each instrument against something known before trusting it.
• Cross-check one quantity two ways and read the disagreement as information.
• Establish the noise floor of your own setup and use it as a limit.
• Record a reading with everything a later reader would need.
• Recognise the three commonest bench mistakes from their symptoms.

Check yourself