High-tech precision manufacturing facilities, particularly semiconductor photolithography processes and advanced display production lines, require HVAC control precision dimensionally different from general building HVAC. Environments housing exposure equipment (Scanner/Stepper) must constrain room temperature within ±0.1 °C to prevent overlay defects caused by wafer thermal expansion.
In these environments demanding extreme precision, attempting to control OACs (Outside Air Conditioners) or precision AHUs — which process large outdoor air volumes — with a single PID loop inevitably encounters severe temperature hunting and control failure.
“For a comprehensive overview of the broader electrical control architecture in these environments, refer to our guide on cleanroom HVAC electrical control.“
This article addresses an engineering case where temperature control delay and overshoot problems in a precision cleanroom air handler were fundamentally resolved by introducing a cleanroom cascade control algorithm.
1. Incident Overview: Symptoms and Operating Conditions
1.1 Equipment and Control Conditions
The precision cleanroom (ISO Class 3) zone air handler was operating under the following conditions:
| Parameter | Design Setpoint | Allowable Tolerance |
|---|---|---|
| Controlled space | Photolithography (Yellow Room) cleanroom | — |
| Room temperature (master target) | 23.0 °C | ±0.1 °C (precision control) |
| Existing control architecture | Single loop based on room (return) temperature | DDC directly modulates chilled water valve based on room temperature |
| Outside air intake conditions | Large seasonal load variation | Chilled water and hot water coil mixed operation |
1.2 Symptom Profile
During AHU commissioning and initial operation, the following instability symptoms appeared in BMS (Building Management System) trend logs:
- Severe temperature hunting: Room temperature (PV) oscillated dramatically between 22.5 °C and 23.5 °C around the 23.0 °C setpoint with a 15~20 minute period.
- Delayed disturbance response: When external disturbances such as chilled water header pressure variation or outdoor air temperature changes occurred, system recovery took excessive time (over 30 minutes).
- Extreme supply air temperature fluctuation: Every time the valve opened, 12 °C extremely cold air was dumped; when closed, air over 20 °C was supplied, delivering massive thermal shock to the room.
2. Root Cause Analysis: Physical Limits of Single-Loop Control (Thermal Inertia and Transport Delay)
Field analysis revealed this was not a mechanical valve fault or chilled water supply problem. The cause lay in the control architecture itself, which ignored the spatial thermal inertia and dead time of air transport.
2.1 Integral Windup Due to Transport Delay
When the AHU’s chilled water valve opens, cooling the coil and the cooled air travels through fans and ducts into the cleanroom, ultimately reaching the indoor temperature sensor (return sensor), there is physical time delay (minutes to tens of minutes).
A single PID controller does not understand this “physical delay time.” Seeing that room temperature hasn’t yet dropped, the controller misinterprets “the valve hasn’t opened enough” and continues opening the valve to 100% (integral windup). When the cold air finally reaches the indoor sensor, an enormous amount of cold air has already been pouring into the duct, so temperature plunges past the 23.0 °C target down to 22.5 °C.
2.2 Inability to Respond to Localized Disturbances
Even if chilled water temperature from the chiller plant fluctuates slightly or piping pressure changes affect coil cooling capacity suddenly, a single-loop control only recognizes and begins responding “after” the disturbance has changed room temperature. In other words, it could only perform reactive control — locking the stable door after the horse has bolted.
3. Relevant Industrial Standards and Design Guidelines
The following standards governing precision manufacturing environments were cross-referenced for problem resolution:
| Standard | Application Range in This Case |
|---|---|
| SEMI F47 | Semiconductor equipment voltage and environmental stability — variation tolerance limits |
| SEMI E72 | Equipment environmental specifications — explicit temperature variation limits |
| ISO 14644-1 Class 3 | Photolithography cleanroom particle cleanliness classification |
| ASHRAE TC 9.9 | Precision environmental control guidelines |
| IEC 60751 Class A | Precision Pt100 RTD tolerance specifications |
SEMI E72 clearly defines temperature variation limits for exposure equipment environments, indicating that multi-loop control architecture rather than a single PID loop is practically essential for ±0.1 °C class control.
4. Engineering Solution Applied: Cascade Control Architecture Introduction
For processes with large Lag time, the fundamental solution is to introduce Cascade Control, dividing the control loop into two nested loops. The PLC/DDC algorithm was redesigned as follows:
4.1 Dual Loop Composition
Cascade control consists of a “slow master loop” and a “fast slave loop.”
Master Loop — Room Temperature Control:
- Input (PV1): Room temperature sensor (target 23.0 °C)
- Output (CV1): Does not directly control the valve, but calculates and commands the slave loop’s “supply air temperature setpoint (SP2)” (e.g., “The room is slightly warm now, target supply air at 17.5 °C”)
Slave Loop — Supply Air Temperature Control:
- Input (PV2): Supply air temperature sensor at AHU outlet
- Setpoint (SP2): Temperature commanded by master loop (17.5 °C)
- Output (CV2): Chilled water valve position modulation (0~100%)
4.2 Control Mechanism Differences and Tuning Principles
In a cascade system, when localized disturbances such as chilled water temperature fluctuations cause supply air temperature (PV2) to vary, the slave loop immediately manipulates the valve to defend supply air temperature to the master’s commanded value — before it can impact room temperature (PV1).
For successful cascade control, the following tuning principles were strictly applied:
- Response Speed Differentiation: The slave loop (supply air control) must be tuned (P-Gain increased) to respond at least 3-5x faster than the master loop (room control).
- Master Loop Flexibility: The master controller was designed with wide proportional band (small P-Gain) and long integral time, so it doesn’t violently shake the supply air temperature setpoint.
5. Post-Correction Performance Verification
After converting the algorithm from single PID to cascade structure and completing tuning, system load testing showed:
| Parameter | Before (Single Return Air Temperature Control) | After (Cascade Control) |
|---|---|---|
| Control valve operation pattern | Extreme open/close (0% ↔ 100% oscillation) | Fine modulation defending supply air temperature |
| Supply air temperature stability | 12 °C ~ 22 °C wild swing | Immediate tracking of master’s commanded SP |
| Room temperature (PV) deviation | ±0.5 °C or greater (out of spec) | Precision maintained within ±0.05 °C |
| Disturbance recovery time | Over 30 minutes with hunting recurrence | Converged within 3 minutes (room temperature change negligible) |
The room temperature graph drew a perfect straight line, and thanks to the slave controller actively controlling supply air temperature in the background, photolithography process requirements were met with margin to spare.
6. Conclusion and Implications
In environments demanding extreme precision and stability such as high-tech cleanrooms and data centers, temperature control failures are often mistakenly attributed to “chilled water valve defects” or “PID tuning problems.” However, in facilities with large controlled space volumes and long air transport paths creating dead time, no amount of PID coefficient tuning can overcome physical limits with a single loop.
This case demonstrates that engineering problem-solving’s essence is redesigning the control architecture itself to suit the thermodynamic characteristics of the process (cascade control). Electrical and control engineers designing and operating BMS/DDC must possess design insight to go beyond simple loop wiring and apply multi-loop control to appropriate locations by understanding mechanical thermal inertia and the nature of disturbances.