Full cohort, 1.40 Pugsley, 12:00-13:00.
HARD STOP 12:50. FDAC control lecture 1 is 13:00-15:00 in this room, and the
TSRs set the Quanser rig up during this hour. 25 minutes each half.
(That FDAC slot is two 50-minute lectures with a break, not one 110-minute
block, so if the rig setup needs longer there is a second break inside it to
borrow from. Steve's call, and not a reason to plan on overrunning here.)
Top-level bullets are fragmented - they appear one at a time on a click.
CHECK BEFORE TUESDAY: the four supplied photographs were assigned in the order
they were sent - Burns, Ward, Graham, Maes. Swap the files in
docs/assets/people/ if any of them is wrong.
Robin Carter starts his PhD on 1 Oct and has flown these autopilots on the MSc
Aerial Robotics project - worth saying so when you introduce him.
MazeMap share links, also kept in teaching/rooms.md so other units can reuse
them. The Hangar (L.080) is in Queens Building too.
Video starts at 23s, muted. In the PDF this is a still image linked to the
video, because Marp's PDF export cannot carry an iframe.
The Cube Orange+ runs ArduPilot and is the Flight Lab's standard research
autopilot - flown for volcano monitoring, conservation and search and rescue.
In AVDASI 2 it is bolted into a wind tunnel with no GPS, and you command
surfaces directly. That is deliberately off-design: much of the avionics work
is getting a flight controller to behave usefully on the ground.
Requirements Spec section 2.2, Electronics. Point at the word "shall": this is
a customer document, not advice. Almost none of the hardware is a choice, and
almost all of the engineering is in what joins it together.
Sections 2.2, 2.3, 2.4, 3.3 and 4.4. The logging rate is the one students
underestimate: it rules out the slow-and-simple approaches.
Section 2.2 for the gate, 2.3 and 2.4 for the servo-tester route. Say plainly
that TB1 ends with two halves that each work on a bench, and that putting them
together is the harder job which starts in TB2 - that is the thing groups
leave too late.
Servo stall means the servo is driving against a stop or a jam and drawing its
maximum current - amps, from a rail sized for milliamps. Nothing to do with the
wing stalling, and the double meaning catches people every year.
Handover point. Steve on theory and design tools; Mark and Vince on
implementation. Sessions start in week 3.
Drawing is DRG A2 from the requirements spec. Tolerances: 2 degrees for
elevator and rudder, 5 for flap and aileron.
The cutaway: demand in on the signal wire, the pot measures the output angle,
the board drives the motor until the two agree. The loop is inside the servo -
which is why the cable, not the shaft, is where we draw the interface.
Photo credits: standard servos by Kborer (public domain); servo internals by
Jstapko, CC BY-SA 3.0; micro servo by Suyash Dwivedi, CC BY-SA 4.0 - all from
Wikimedia Commons. The tester is drawn.
No openly licensed photograph of an HJ or Parallax servo, or of a tester,
exists - the tester is drawn. A lab photograph would be better than both.
Worth saying out loud: a hobby servo is itself electromechanical, with its own
closed loop inside - a potentiometer on the output shaft, compared against the
commanded pulse width, driving the motor until the error is zero. So the real
interface arguably sits *inside* the servo, and the cable is just where we
agreed to draw the line. Every interface is a choice like that.
TODO(Steve): session content and who leads each - to be confirmed with Mark
and Vince.
Mark delivers this one. Seven clicks, one per topic, in the order of his own
slide: kinematics, torque and motion, pivots and joints, standard parts,
machining and metrology, tolerancing, then the examples.
The torque and motion sketch is redrawn from Mark's original: a flap carried
on a support at each end, driven from one end only, so the drive has to reach
across the whole span. The frowning face is his.
Source: Mark's "Mechanisms - Design realisation" slide, September 2026.