Isolated Voltage Sensing for Safety Control

single image

The second leading cause of electrical fires is improperly maintained or unprotected electric motors. It’s not surprising, then, that safety is a top priority when designing an industrial motor drive with high voltages and currents. Most drives, in particular, include an inverter circuit that converts alternating current (ac) into a high-voltage direct current (dc) signal, known as a dc link, which serves as a power supply for circuitry that generates drive signals to power the motor.

The dc-link voltage must be controlled continuously. Under certain conditions, a motor can act as a generator, delivering a high voltage back into the dc link via the inverter’s power device and/or recovery diodes. This high voltage increases the dc-link voltage, and the IGBTs (insulated-gate bipolar transistors) that drive the motor can be stressed by a high — and potentially damaging — surge voltage.

Such overvoltages are easily detectable by modern optical couplers used to monitor the dc link. The same holds true for detecting undervoltage conditions caused by a power grid overload or the loss of a power phase, both of which are common. Isolated voltage sensing is required for safe, robust system control and protection in a variety of applications, not just motor drives. Solar inverters and uninterruptible power supplies are two examples of other applications (UPS).

In all of these cases, a sensor must accurately measure the high dc-link voltage and provide galvanic isolation between the potentially dangerous high-voltage side and the low-voltage controller side, which is located near system operators.

In addition to isolated voltage sensors, isolated temperature sensors and 0-to-10-V analog control protocols can benefit from isolated voltage sensors. In this case, the voltage sensor must measure the temperature or control signal linearly and accurately before sending it across the power isolation  barrier, providing insulation as a safety measure and eliminating ground loops.

The optical signal path providing the electrical insulation barrier is depicted in the internal block diagram of a modern isolation amplifier (ACPL-C87X). Because the signal is optical rather than electrical, it is not affected by magnetic fields or electrical noise.

Optically isolated amplifiers are specifically designed for shunt-based current-sense applications that require an input voltage range of 200 mV (full scale 320 mV). However, because these amplifiers have a limited dynamic range, it is difficult for them to detect low-level signals in the presence of high voltages. As a result, there has been a demand for power isolation amplifiers with a wider dynamic range.

To handle voltage-sensing applications, optical isolation amplifiers with a nominal input range of 0 to 2 V have recently been designed. Aside from an increased input voltage range, the new isolation amplifiers have a 1-GΩ input impedance versus the 500-kΩ input impedance of previous models, a 2,000 improvement. The high input impedance reduces the error caused by signal source loading significantly.

How Optical Isolation Amplifiers Work

A look at the functional blocks of an isolation amplifier reveals that the amplifier first senses the input voltage (single-ended analog signal) and converts it to a digital bit stream. The bit stream is then sent across an optical coupling pair made up of an LED and a photodetector. The electrical insulation barrier is provided by this optical signal path. Because the signal is optical rather than electrical, it is not affected by magnetic fields or electrical noise.

The photodetector recovers the optical signal and converts it to an electrical signal, which is then decoded and filtered to generate an analog output signal. The output voltage is proportional to the input voltage with a gain of one and is provided in differential mode for better common-mode noise rejection.

The isolation amplifier comes with three gain accuracy options: ±0.5,  ±1, and ±3%. It’s also housed in a stretched SO-8 package that’s 30% smaller than the standard DIP-8 package. According to the UL 1577 safety standard, all amplifiers have a 5,000 VRMS/1 min double protection rating. The IEC/EN/DIN EN 60747-5-5 maximum working voltage specification of 1,414 VPeak ensures that circuits on the low voltage side are not damaged by dangerous high voltages.

When isolation amps are used to sense the dc link voltage in inverters, a voltage divider (here formed with resistors R1 and R2) is typically formed to scale the dc bus voltage down to the isolation amp’s 0 to 2-V input range (ACPL-C87X). The overvoltage range threshold can be set to 2.4 V, which is 20% greater than the nominal voltage.

Designers who use isolation amplifiers as isolated voltage sensors first choose resistors to create a voltage divider, which scales down the voltage to a level within the sensor input range. The application circuit is significantly simpler than alternative approaches that use separate sensing and  power isolation devices.

Isolated voltage sensing in ac and servo motor drives, isolated dc-link voltage sensing in solar and wind-turbine inverters, and signal isolation in data-acquisition systems are examples of typical applications.

The dc-link voltage must also be continuously controlled in motor drives. Under certain conditions, a motor can act as a generator, delivering a high voltage back into the dc link via the inverter’s power devices and/or the recovery diodes surrounding the power devices. This high voltage increases the dc-link voltage, and the IGBTs are stressed by a high — potentially damaging — surge voltage.

Undervoltage is a common condition caused by a power grid overload or the loss of a power phase. Isolation amplifier circuits can detect such undervoltages due to their wide input range. A low linearity error of as little as 0.05 percent is also useful for this application. As an example, as an undervoltage condition, a designer could set a threshold that is 20% lower than the nominal dc-link voltage. As a result, the lower limit is set to 1.6 V.

The accompanying diagram shows an example of a detailed circuit with an isolation amplifier with a wide input range. Given that the voltage sensor nominal input voltage for VIN on the isolation amp (ACPL-C87x) is 2 V, the designer must select resistor R1 using Equation 1:

R1 = (VL1–VIN)/VIN × R2   (1)

For instance, if VL1 is 600 V and R2 is 10 k, then R1 is 2,990 kΩ.

There are several approaches to selecting resistor values. One method is to combine several resistors in series to achieve the desired value; for example2 MΩ, 430 kΩ, and 560-kΩ resistors in series produce 2,990 kΩ exactly. A VIN of 2 V corresponds to a VL1 of 600 V; however, specific resistance values may be difficult to find if VL1 is not 600 V.

To simplify resistor selection, another method is to round the target resistance up to a convenient value, such as 3 MΩ . The scaling relationship may need to be fine-tuned in such cases. VIN is 1.993 V in the same example with a VL1 of 600 V, R1 of 3 MΩ , and R2 of 10 k.

A simplified version and an example of measuring high voltage converted to an isolated ground referenced output.

The isolation amplifier senses the down-scaled input voltage after it has been filtered by the anti-aliasing filter formed by R2 and C1, with a corner frequency of 159 kHz (The value of R1 is usually much larger than R2, so it is ignored in this calculation.). The differential output voltage (VOUT+ – VOUT–) of a galvanically isolated differential amplifier is proportional to the input voltage. The differential signal is converted to a single-ended output by the OPA237, which is configured as a difference amplifier. This stage can also be used to amplify the signal and, if necessary, to low-pass filter it to limit its bandwidth.

The difference amplifier in this circuit has a gain of one and a low-pass filter corner frequency of 15.9 kHz. The gain can be adjusted by changing the resistors R5 and R6. By increasing the capacitance of C4 and C5, the bandwidth can be reduced. The isolated output voltage VOUT can be safely connected to the system microcontroller because it is linearly proportional to the line voltage on the high voltage side.

You may like