In a Two-Degree-of-Freedom PID (2DOF PID) Controller, the Beta (β) Parameter: Technical Guide for Coffee Roasting Applications

[Terracotta-6]

🔧 I. Understand the Roles of β and γ

Parameter	Role in Formula	Practical Impact
β (Beta)	Weight of SV in the proportional term: $u_P = K_p[\beta\cdot SV-PV]$	Controls the aggressiveness of response to target temperature changes: • β↑ → More intense heating, prone to overshoot • β↓ → Mild heating, suppresses overshoot
γ (Gamma)	Weight of SV in the derivative term: $u_D = K_d[\gamma\cdot\frac{dSV}{dt}-\frac{dPV}{dt}]$	Controls whether "derivative kick" is triggered by sudden setpoint changes: • γ=1 → Derivative term responds to SV jumps (prone to jitter) • γ=0 (Recommended) → Only based on PV change rate (Derivative on Measurement, DoM), more stable

✅ For the vast majority of coffee roasting scenarios, it is recommended to fix γ = 0 to avoid severe heater fluctuations caused by setpoint steps (e.g., clicking to switch to a new roasting phase).

[Terracotta-6]

2DOF PID + Gain Scheduling: An Advanced Temperature Control Strategy for Coffee Roasting

In coffee roasting—a highly nonlinear, time-varying process extremely sensitive to thermal dynamics—the precision and stability of temperature control directly determine flavor development. While traditional fixed-parameter PID control is widely used, it often struggles to handle the drastically different thermodynamic behaviors across roasting phases (e.g., drying, Maillard reaction, first crack, development). To address this, combining **Two-Degree-of-Freedom PID **(2DOF PID) with Gain Scheduling offers a sophisticated, adaptive solution for intelligent roast control.

---

1. Limitations of Conventional PID in Roasting

• "One-size-fits-all" parameters: A single set of (K_p, K_i, K_d) cannot simultaneously satisfy rapid heating early on and fine-tuned control later.
• Setpoint changes cause overshoot: Jumping from, say, “yellowing point” to “first-crack target temperature” creates a large error, leading to aggressive heater output and bean temperature spikes.
• Inability to distinguish between "command changes" and "disturbances": The controller reacts passively without anticipating operator intent.

---

2. 2DOF PID: Decoupling "Tracking" from "Disturbance Rejection"

2DOF PID introduces **setpoint weighting **((b)) and **derivative feedforward **((c)) to split control into two independent paths:

• (b < 1): Reduces the impact of setpoint jumps on the proportional term, minimizing overshoot.
• (c = 0): Disables derivative action on setpoint changes, avoiding "derivative kick".
• Result: Smooth tracking of roast profiles while robustly rejecting disturbances (e.g., batch weight variance, ambient temperature shifts).

✅ Example: As first crack approaches, the system automatically lowers (b), preventing the controller from overreacting to exothermic heat release.

---

3. Gain Scheduling: Letting PID "Adapt with the Roast"

Gain scheduling dynamically adjusts PID parameters based on real-time variables—such as bean temperature (BT), environmental temperature (ET), time, or roast phase:

• Define multiple parameter sets:
  • **Drying phase **(<160°C): High (K_p) for rapid heat absorption.
  • **Pre-first-crack **(195–205°C): Low (K_p), high (K_d) to suppress exothermic overshoot.
  • **Development **(>205°C): Emphasize integral action for stable rate-of-rise (RoR).
• Flexible scheduling methods:
  • Lookup tables
  • Linear or quadratic interpolation (Artisan already supports gain_scheduling_quadratic)

✅ Benefit: Seamless transitions between thermodynamic regimes without manual intervention or control disruption.

---

4. 2DOF + Gain Scheduling: Synergy Beyond Summation

Integrating both means each scheduled parameter set is a full 2DOF configuration ((K_p, K_i, K_d, b, c)):

Roast Phase	(K_p)	(K_i)	(K_d)	(b)	Control Objective
Charge – Drying	High	Medium	Low	1.0	Rapid establishment of thermal energy
Yellowing – Pre-FC	Medium	Medium	Medium	0.7	Stable ramp, avoid premature cracking
First Crack	Low	Low	High	0.3	Suppress exothermic overshoot
Development – Drop	Very Low	High	Medium	0.0	Precise development control, lock flavor

Outcomes:

• Roast curves closely follow target profiles;
• Tighter first-crack clustering and uniform development;
• Significantly improved batch-to-batch repeatability;
• Reduced need for manual intervention—“set and trust.”

---

5. Implementation Potential in Artisan

Artisan already provides a strong foundation:

• Support for PV/SV-based gain scheduling;
• Event/alarm-triggered parameter updates;
• Rich device interfaces and PID tuning tools.

Next steps could include:

1. Introducing a 2DOF PID structure (extend PID settings UI with (b) and (c) parameters);
2. Allowing users to define complete 2DOF parameter sets per temperature segment;
3. Integrating automatic roast-phase detection (e.g., via dBT/dt for first crack).

This would elevate Artisan from a “data logging and basic control platform” to a true intelligent roasting engine, empowering users to achieve scientific, precise, and reproducible professional roasting.

---

Conclusion

2DOF PID + Gain Scheduling is not just an elegant application of control theory—it’s a technical embodiment of the essence of coffee roasting: finding balance within dynamic change. It enables machines not merely to execute commands, but to understand the process, anticipate transitions, and respond gracefully. Ultimately, this frees roasters to focus on flavor craftsmanship—not firefighting. This is the future of intelligent coffee roasting.

[MAKOMO]

On first read this all makes sense, however, I have to verify this in detail once I find the time. Thanks for sharing!