LabCD — controller design with receipts
LabCD is the IDO platform for control design. It identifies a plant model from your data, tunes the PID or MPC against it, and returns the margins, step response and validation evidence with the controller, not after it.
- Line Engineer
- Control systems engineer doing identification and loop tuning
- Operators / Junior Engineers
- Trial-and-error tuning on live plant during commissioning
- Tools
- MATLAB / SimulinkControl System ToolboxPID autotunersSpreadsheet step tests
Plant model first, gains second
Most loops are tuned by feel and signed off by whether the trend looks calm. That works until the operating point moves. LabCD fits a plant model to logged step and excitation data, tunes against that model, and reports what the tuning is worth: gain margin, phase margin, sensitivity peak, disturbance rejection.
- System identification from logged step, PRBS or closed-loop data
- PID and MPC tuning against the identified model, with the model error stated
- Gain margin, phase margin and sensitivity peak reported per loop
- Validation run comparing predicted response against measured response
Control with receipts
A receipt is the difference between a loop that is tuned and a loop that is proven. For every controller LabCD hands back the model it was tuned on, the margins it holds, the step and disturbance responses, and where the model stops being trustworthy. If the plant is too non-linear for the fit, that is part of the output rather than a surprise at commissioning.
Where it fits
LabCD is built for robotics and mechatronics engineers and operations-side technical leads who own loops without being control specialists: motion axes, temperature and pressure loops, flow and level control, actuator and drive tuning. It explains the selection logic rather than assuming you live in the control literature.
A temperature loop that is stable in winter and oscillates in summer
A process line holds temperature with a PID that was tuned once during commissioning. It is quiet for months, then starts hunting when ambient conditions and throughput change. Nobody can say how much margin the loop ever had. LabCD fits a plant model from the existing historian data at both operating points, shows that the original tuning sits close to the stability limit at the hotter one, retunes for margin across the range, and hands over the step response and margin figures that justify the change.
Common questions
Do I need to bump the plant to use this?
A deliberate excitation gives the cleanest model, but LabCD can identify a usable model from historian data when the process already moves enough. The output states how confident the fit is either way.
PID or MPC?
PID for most single loops. MPC earns its place when there are real constraints, coupling between loops or a significant dead time. LabCD makes the case from your model rather than defaulting to the fancier option.
Does it write PLC or controller code?
It produces the tuning, the model and the validation evidence. The gains go into your existing controller, and the report is what you hand to whoever signs off on the change.
Talk to an engineer about LabCD
Bring a loop that misbehaves and the data you already log. An IDO engineer will scope what can be identified, tuned and proven.
Contact IDO Engineering