Open-loop speed drop under load, worked out
In the lab you pin a held-voltage run and a feedback run and watch them part at 4 s. Here the same comparison is done on paper first: one steady-state line predicts both the open-loop speed after the load and the voltage feedback must find, and a third case shows the limit feedback cannot cross.
Companion to the lab experiment Open loop versus closed loop. Run it there; this note predicts what you will see and why.
The model in play
The same two-state motor (R = 2 Ω, L = 0.12 H, Kₑ = Kₜ = 0.05, J = 0.002 kg·m², b = 0.0005 N·m·s) starts from rest with no load. A persistent 0.04 N·m load step arrives at 4 s. In open loop the command is a held 7 V and the 100 rad/s target is only a comparison line; in feedback the default PID (kp = 0.22, ki = 0.6, kd = 0.008) acts on the speed error every 20 ms.
Simplifications that matter here
- The steady-state line ignores the transient. The motor's slow mode has a time constant of 1.10 s, so after a change it needs several seconds to reach the predicted speed.
- The load enters as a torque on the rotor, not as a change of the speed measurement.
Predict it by hand
Setting both derivatives to zero gives one line for any steady state: V = 0.07ω + 40τ. Read it two ways. With V held at 7 V it gives the speed, before and after the load. With the target speed fixed it gives the voltage feedback must reach.
Check against the engine
The feedback run ends at 100.00 rad/s with 8.60 V, both as predicted. The open-loop run reads 77.16 rad/s at 12 s, 0.02 rad/s above the predicted 77.14: eight seconds after the load the slow mode still has e^(−8/1.10) of its 22.86 rad/s drop left to go. Feedback is not instant either: the closed loop dips to 96.22 rad/s before the integral has raised the voltage above its pre-load 7 V.
What to change next
- Hold 8.6 V in open loop instead of 7 V.
It is the right voltage for the loaded motor, but only for that load. Before 4 s the unloaded motor heads for 8.6/0.07 = 122.86 rad/s (the engine is at 119.45 rad/s when the load arrives), and afterwards it settles at 100.01 rad/s by 12 s. An open-loop command is calibrated for one disturbance.
- In feedback, make the load step 0.15 N·m.
Holding 100 rad/s now needs 7 + 40 × 0.15 = 13 V, more than the limit, so even feedback stops at (12 − 40 × 0.15)/0.07 = 85.71 rad/s at most. The engine ends at 84.96 rad/s with 11.95 V: the anti-windup rule refuses any integral step that would push the request past 12 V, so the request parks just below the limit and the speed stays 0.75 rad/s under the ceiling.
Common mistakes when reading this run
- Treating the target line as an open-loop command. The held voltage never reads it; it is there to measure the 22.86 rad/s shortfall.
- Saying feedback makes the motor stiffer. The motor's own response to the load is unchanged; feedback recovers by raising the voltage, which you see in Control effort.
- Assuming feedback can absorb any load. Every load above (12 − 7)/40 = 0.125 N·m costs speed at this target.
Where to go next
Run this experiment in the lab, pin a run and try the two changes above.