Class 16 · Capstone and Fault-Finding Exam

Build it, prove it, document it, hand it over. Then find three faults in somebody else's under time pressure, with the clock running and nobody helping.

What the capstone is#

A working piece of show electronics that you designed, built, tested and documented. It must:

• Take an input — a sensor, a cue, a DMX channel, or more than one
• Drive a real load through switching you built and can explain
• Have a defined power-up state and a defined behaviour on loss of control
• Detect at least one fault condition and respond to it in a way you decided in advance
• Show its state on an indicator readable across a room
• Come with one side of A4 that a stranger can operate it from

What the capstone is#

Scale is not the point. A well-executed practical lamp with honest documentation scores far above an ambitious machine that half works and has no sheet, and it does so for the same reason a production would choose the first one.

How it is assessed#

Four parts. Note the weighting, and note that only the first is about the object.

How it is assessed#

The proof, in detail#

You will be asked to demonstrate, with instruments, that your claims are true. Expect to be asked for any of:

• The current your load actually draws, measured, against what you designed for
• The voltage across your switching device while conducting, and the dissipation calculated from it
• That the isolation barrier is intact, measured
• That the power-up state is what you documented, demonstrated three times from cold
• That the loss-of-control behaviour is what you documented, demonstrated by disconnecting the control
• That the fault detection works, demonstrated by causing the fault
• The PWM frequency, or the loop time, or the debounce interval, on a scope

"It works" is not a proof. A measurement, with the meter or scope visible and the number stated, is a proof. This is the Class 2 discipline, assessed.

The documentation, in detail#

One side of A4:

1. What it is and what it is for, in two sentences
2. Ratings: supply voltage, current, maximum load, isolation, duty cycle
3. Pinout, drawn, with connector orientation
4. Power-up behaviour
5. Behaviour on loss of control
6. Fault conditions detected, and the indicator code for each
7. Reset procedure
8. Known limitations
9. Test results, dated and initialled
10. Who to contact

Point 8 is where marks are won and where most people lose them. A limitation you have named is engineering. The same limitation undocumented is a defect. Every device in this room has limitations. The sheets that list three specific ones score above the sheets that list none.

The fault-finding exam#

The fault-finding exam#

Ninety minutes. Three faulted systems, thirty minutes each. Each has been built and then broken by somebody else. You work alone. You may use any instrument in the room and any page on this site.

Each system draws on a different part of the course:

• One electrical. A power, switching or load fault. Class 1, 2, 5, 6.
• One signal or data. An analogue, grounding, DMX or network fault. Class 7, 8, 9, 10.
• One control. A microcontroller, sensor or logic fault. Class 11, 12, 13, 14.

What is marked#

What is marked#

You can score well on a system you did not fix. A clear log showing a correct method, a narrowed boundary and a defensible next test scores above a lucky find with no log. That is not a consolation, it is the actual professional standard: the method transfers to faults nobody has seen, and luck does not.

And one absolute rule. If any system presents a safety issue — an exposed conductor, a defeated interlock, a burning smell, anything that should not be energised and is — stop, isolate, and report it. That is a full-mark response, and continuing to work on it is a zero regardless of what else you find.

The half hour, structured#

A shape that works, and that you should practise:

• 0–5 min. Look, do not touch. What works, what does not, what is warm, what is lit, what smells. Read the documentation if there is any. Write down the boundary.
• 5–10 min. First hypothesis and the test that would disprove it. Not confirm: disprove. A test that can only confirm tells you nothing when it passes.
• 10–25 min. Half-split within the boundary. One change at a time, all of it logged.
• 25–30 min. Prove the fix, or write down where you got to, what you excluded, and what your next test would be. Never leave this blank. It is worth marks and it is what a real handover looks like at the end of a shift.

The documentation sheet, in detail#

Twenty per cent of the mark, and the part most people lose points on. It is one side of A4 and it has ten things on it.

It is not paperwork. It is the answer to the questions somebody will ask at 19:45 when your device has stopped and you are not in the building: what is it, what is it rated for, what does it do when it loses signal, and how do I reset it. Without the sheet, the honest answer is "we would have to take it apart", and the real answer becomes "use something else".

The stranger test is the specification. Somebody who has not seen the device, given only the sheet and no conversation, must be able to power it up, trigger it, cause a fault, identify the fault from the indicator, and reset it. Anything they have to guess is a defect in the sheet rather than a shortcoming in them.

Test it on somebody outside this course before you arrive. Every question they ask is a point you have not written down, and it is far cheaper to find them the week before than in the peer review.

The documentation sheet, in detail#

At the bench#

Three hours and forty minutes.

Block A — Set up and peer review (30 min)#

Set your piece up and get it running. Then swap with a partner: operate theirs from the sheet alone, in silence, and write down every moment you had to guess.

Exchange notes. You have fifteen minutes to fix your sheet before assessment, and everybody uses it.

Block B — Demonstration and viva (40 min)#

Ten minutes each with the assessor. Demonstrate, prove your claims with instruments, and answer questions about the design decisions.

The questions asked every year: why that component, what happens if this wire is cut, what did you choose not to do and why, and what would you change if you built it again. Have answers.

Block C — The fault-finding exam (90 min)#

Three systems, thirty minutes each, alone, logged.

Block D — Debrief (40 min)#

The faults are revealed and each is walked through: what it was, the fastest route to it, and what misled people.

Then the last exercise of the course. Each person names one thing they now know they do badly under pressure. Not a weakness in knowledge, a weakness in method: I start touching before I look, I do not write things down, I stop half-splitting when I get impatient, I trust the report instead of observing.

That sentence is the most useful thing you will take out of this room, and it is the only part of the course with no marks attached to it at all.

After this course#

After this course#

You are not an electrician and you are not an electronic engineer. You are something the industry is short of: a technical director who cannot be told a comfortable lie about electricity, who can read a schematic, measure a claim, spot a defeated interlock, and write a specification that a supplier will take seriously.

Three things worth doing in the next year:

Build something nobody asked for. The skills decay without use, and a self-directed project is where you find out which parts you actually understood.

Get the qualification your jurisdiction requires if you intend to do mains work yourself. This course does not give you that and does not pretend to. What it gives you is the ability to supervise and to check the work of people who have it.

Keep the notebook. Every measurement you took this term is in it. In three years the numbers will be less useful than the record of how you searched, which is the thing that actually got better.

Where to go next has the specifics: courses, standards, books and the kit worth owning.

Numbers from this class#