Dynamism and cost reduction: the advantages of predictive control
Antoine Gardiol & Xavier Greppin
In applications using DC motors, common problems engineers frequently encounter are a lack of dynamism and/or the cost of the motor itself relative to the rest of the system. To address these problems, FiveCo's team of specialists has developed a high-performance controller using both corrective and predictive models (2-DOF: PID feedback + feed forward dynamic model).
The following table summarises the improvements brought by this dual control approach.
| Classic PID control based on error only | FiveCo PID and predictive controllers | |
|---|---|---|
| Encoder pulses | As many as possible to enable stable control | A precise encoder only improves braking stability |
| Inertia | Increases delay and can make control impossible | Compensated |
| High internal electrical resistance | Increases delay and can make control impossible | Compensated |
| Setpoint tracking | With delay | Without delay and even with anticipation if necessary |
Reducing motor cost
When control quality must be flawless, it is standard practice to use motors with very good performance and encoders with a high number of pulses per revolution. This type of hardware is expensive, and if the application does not require high dynamics, the static predictive model can very often be sufficient to allow the use of lower-grade motors and significantly less precise encoders.
For example, the following figures show the speed tracking of the same motor, with an encoder of 12 pulses per revolution and a particularly high inertia, without predictive control (left) and with it (right).
The difference is clearly visible — the actual curve in blue follows the setpoint in red much more closely with the predictive controller coupled to the error-based controller. The latter only needs to manage small variations.
Dynamism and precision
If system cost is not the primary criterion, responsiveness and precision of speed tracking often are. Without a predictive method, the speed curve frequently lags behind the setpoint, as error correction systems only act when a deviation occurs. With a predictive method, since inertia is known, it is possible to be far more responsive by anticipating the future error, thereby significantly improving speed tracking. The classic method — PID for example — corrects small deviations, particularly during braking.
All of this is clearly visible in the following graphs showing speed tracking (actual speed in blue and setpoint in red). The classic control on the left systematically shows a delay, while the dual control on the right compensates for it.
This precise tracking makes it possible, for example, to synchronise movements in multi-axis machines without requiring feedback between control boards. Only simultaneous commands issued by the control system (PC or other) are necessary.
Conclusion
The dual classic PID and predictive control developed by the FiveCo team simultaneously improves the dynamics and precision of DC motor drives, and reduces system costs for applications that do not require high precision but do require good stability. This technology is now available across our entire range of DC motor controllers.
