|Home | Articles | Forum | Glossary | Books|
Although analog circuits are relatively inflexible, they can furnish process-control systems with operational features comparable to those attainable using digital methods. A stepper-motor pump-drive application illustrates the techniques involved.
For many process-control applications, analog control circuits prove a better choice than their digital counterparts, especially when you expect low product volumes and when fast design time and high noise immunity are design priorities. In fact, if you're working with well-defined operational specifications and don't anticipate having to make major modifications, analog methods serve as viable alternatives to intelligent but dedicated and expensive hardware/ software approaches.
Controlling a pump's speed
To demonstrate, this article describes the design of an analog pump controller that manipulates computer generated command pulses to regulate stepper-motor driven pumps in a critical chemical-mixing process. The controller/pump system furnishes precise fluid delivery at both fast and slow rates, a requirement often arising in chemical and biological process-control systems, which demand high pumping rates for flushing or process startup and slow but accurate flow rates for mixing precise amounts of liquid. Although DC motors can deliver adequate high-speed performance, they often need complex and expensive digital control to perform well at very slow speeds. In contrast, exponentially driven stepper motors can easily handle a pump's conflicting high- and low-speed drive requirements.
Fig 1 diagrams a computer-driven system that governs several pumps feeding an intricate chemical process. The computer controls each pump's speed by periodically sending a pulse-width-modulated control command. Because the computer runs in a time-shared manner, each pump controller must retain the last received pulse width's value.
In this application, each pump gets speed-updated every 30 sec by a 50- to 1000-msec pulse. The pump drive must provide optimum speed-setting resolution for the low-speed ranges to provide increasingly slower flow rates as the system approaches crucial mixing conditions. And the controller must possess a high degree of noise immunity to prevent spurious noise induced responses from degrading process quality.
Fig 2 illustrates a uP-based-controller scheme. In this arrangement, the computer delivers an input pulse that gates a clock. The clock in turn serially loads a bank of parallel counters that determine the input pulse width. The counters address a processor section that converts input data to a frequency output, using an exponential transfer function-a nonlinear response that achieves the required high resolution (precise liquid delivery) at slow pump speeds. Finally, the frequency output activates a stepping-motor driver that runs the pump.
On the surface, this digital controller's operation appears relatively simple. However, the application masks some tricky design problems. For example, the lengthy period between speed updates, coupled with the need to avoid erroneous pump responses, mandates careful power-supply design, including provision of such functions as RFI filtering, memory battery backup and self-checking software.
In addition, the need for a high-resolution, smoothly varying frequency-output function demands careful design attention to how the processor synthesizes its output. Although these problems are amenable to solution, they complicate the controller's design and entail lengthy development time and high cost.
Analog functions prove adequate in simple process-control tasks
Take the analog route
Considering the task's conceptual simplicity, however, reveals a clear edge for an analog-control approach to satisfying this application's critical requirements. A turnkey system, it needs little intelligence or flexibility and can employ a straightforward data-retention structure. And although the digital ~J-P-based approach can also meet these requirements, it involves substantial hardware and software overhead to overcome noise immunity and frequency-shift-resolution problems.
The analog-based design surmounts these obstacles, providing inherent noise immunity and superior frequency-vernier capability. More important, though, an analog approach eliminates the intensive software effort required by uP-based methods. As a matter of record, the analog pump-controller design was conceived, breadboarded and released for production in just 4 wks-and at a cost competitive with an alternative uP-based method.
Fig 3 depicts the analog system. In this scheme, a capacitor furnishes memory storage. An exponentially responding voltage-to-frequency converter (VFC) fulfills the function of Fig 2's processor. In operation, the computer's command pulse gates a current source that linearly charges the storage capacitor. While the capacitor is charging, the sample/hold stage enters Hold mode, blocking the capacitor's ramping action from the VFC.
When the command pulse just ceases, the capacitor achieves a voltage level that the sample/hold accepts and feeds to the VFC. By issuing an extremely wide pulse, the computer actuates the turn-off stage, which deactivates the stepper-motor drive.
Optoisolation eliminates noise
Fig 4 shows the analog pump controller's schematic diagram. To initiate circuit action, the computer sends an input pulse to the 4N28 optoisolator, which eliminates noise-pickup-induced ground-loop and data-line problems. Appearing at its emitter, the opto-isolator's output (Fig 5, waveform A) goes to IC1A and IC1B. IC1A's differentiator setup-a 0.001uF/33-k-Ohm combination generates a short pulse (Fig 5, waveform B) that biases Q7. This transistor in turn resets its associated 1-uF capacitor (Fig 5, waveform C).
Note that Qs's emitter supplies the current to base-bias Q7 ON because IC1A is an open-collector device. In turn, Q8 receives its base bias from the optoisolator, which provides a drive output only when a command pulse appears at the controller circuit's input.
Consequently, in the highly unlikely event that a severe noise disturbance causes IC1A's output to rise, Q1 still doesn't receive a drive pulse, and its 1-uF capacitor does not get reset.
The 1-M-Ohm/4.7-uF filter, which feeds IC1A's minus input, provides additional noise immunity by ensuring a stable trip point during noise disturbances. The optoisolator's output also goes to IC1B, which gates the Q1 current source. When Q7 turns off, its 1-uF capacitor immediately starts to ramp up (Fig 5, waveform C). (Circuit-operation speed in Fig 5 has been increased to provide optimum waveform photographs.) Then, the Am follower unloads the capacitor.
Diode/capacitor decoupling of Q1 assures high noise rejection, even for supply dropouts, during the capacitor's ramp time. During ramping, IC2A's output stays LOW and shuts off S1. This switch maintains A1A's output at a dc level. When the controller's input pulse ceases, IC1B's output goes LOW and disables Q1. The integrating 1-uF capacitor therefore stops charging.
Concurrently, IC2A's output goes HIGH and closes S1. As a result, A1 A's output changes to the capacitor's newly acquired level. Located in A1A's input section, the 3-M-ohm/0.47-uF filter provides a time constant that limits the stepper motor's acceleration rate, thereby preventing stalling.
A voltage-to-frequency converter controls stepper-motor drive
Try an exponentiator
Op amp A1A's output feeds the A2-A3 configuration, which forms an exponentially responding VFC that controls the input current to the A3A-A3B integrator comparator-type oscillator stage. To accomplish this function, A28 and the LM394's dual transistors constitute a voltage-input, current-output exponentiator in accordance with transistor V BE-VS-Ic characteristics.
The 1-kO temperature-compensating resistor connected to the LM394 thermally compensates for the KT/Q drift factor. Similarly, the LM394's dual transistors suppress V sE's contribution to temperature error.
A2A biases the exponential converter's input range by combining A1A's output with the necessary offset term for proper exponentiator operation. Trimmers allow you to adjust the 1200- and 0.2-Hz endpoints.
A3s's pulse-train output contains frequency components that relate exponentially to the controller circuit's most recently received input-pulse width. It drives the CD4022 counter chip, which generates four properly phased signals (Fig 6) for driving the stepper.
Driving the pump
The additional sections of IC1 and IC2 allow the computer's command pulse to shut down the pump. For the normal range of input widths, the 1-uF capacitor at IC1C's plus input (Fig 5, waveform E) never charges above 10V. Under these conditions, IC1C's output always stays LOW. The only source available to charge the 1-uF capacitor tied to IC2B's minus input thus comes through the 18-M-ohm resistor.
However, during normal operation, A1A's output remains positive, ensuring that IC2B's negative input stays that way. This condition forces IC2B's open-collector output to float. If the controller circuit receives an input pulse substantially wider than the normal maximum, therefore, IC1c's input charges above 10v. This action quickly dumps a large charge into IC2B's 1-uF capacitor, forcing its voltage level to rise to the negative rail. This value pulls A1A's input negative, turns on Q2 and cuts off all drive signals to the output transistors (Q3 to Q6).
Use an optoisolator to eliminate noise effects
A2A's negative output also feeds back to IC2C, driving that device's output positive. This output supplies a continuous topping-off current to IC2B's input capacitor.
The connection completes a positive feedback latch, which prevents the pump from operating until the counter receives a pulse width within the controller circuit's normal range. IC_1D functions to clear out the IC_1B’s capacitor's charging action (Fig 5, waveform D) as each new command pulse arrives.
The time constant associated with A1A's input section lets the controller circuit examine each received pulse and never disables this clamping performance unless the pulse width resides within established limits.
Although the latch's positive feedback doesn't require the computer to send successive shutdown instructions to the pump, the controller circuit ensures that the pump's motor can't be energized, even briefly, if successive turn-off-length pulses appear.