“Induced currents in brushless motor drives are essential, even if only for current limiting. Current readings must be consistent if performance and stability are to be optimized over the entire operating range. There are many ways to switch a three-phase bridge, which complicates current sensing. It’s not always obvious which way current will flow through the bridge, especially when considering the various braking and drive modes.

“

Induced currents in brushless motor drives are essential, even if only for current limiting. Current readings must be consistent if performance and stability are to be optimized over the entire operating range. There are many ways to switch a three-phase bridge, which complicates current sensing. It’s not always obvious which way current will flow through the bridge, especially when considering the various braking and drive modes.

The simplest and most common format for low-current drivers is to place a shunt on the common negative return line, as shown in Figures 1 and 2.

Figure 1: Single shunt in common negative

Figure 2: Single shunt in common negative with current indication

For single-sided switching in analog drives, this format is very unsatisfactory because the average shunt current is not the average motor current. There is a flywheel current inside the bridge that does not pass through the shunt resistor. Peak current detection is required, but this immediately introduces noise problems.

Bridge current for one-sided switch

Using a digital driver, the shunt current can be sampled during the on-time of the relevant phase, but this is not without problems. When the phases are close to commutation, the time available to sense the current is very short because the duty cycle is very short at low speed. To further exacerbate the problem, the parallel current has many high frequency components that must be detected by an amplifier with a very high slew rate (perhaps around 20V/µs). The common-mode voltage swing across the shunt must also be considered (Figure 3).

Figure 3: Signal layout during on-time

Influence of PWM on Single-channel Parallel Current Detection

With dynamic braking, the shunt may not see any current, so the bridge is vulnerable to damage. The lower FET legs will need to be turned on periodically so that the current can be sampled in the shunt (Figure 4).

Figure 4: Effect of PWM on Single Parallel Current Sensing

Dynamic braking current

Double-sided switches can be avoided if good control of braking is required without a specific braking mode. For zero voltage drive, 50% PWM is required. This does mean more time to measure current at low speed rather than high speed. The shunt sees a current reversal every PWM cycle (Figure 5).

Figure 5: Dynamic Brake Current

double-sided switch

A possible disadvantage of a double-sided switch with a low-inductance motor is that the ripple current can be high, which increases switching losses in the FET and I2R losses in the motor and shunt.

This time there is a further complication regarding sinusoidal commutation. With trapezoidal commutation, only two branches of the bridge are switched at any one time, but when we turn to sinusoidal commutation, all three branches are switched simultaneously. In Figure 6, we can see that a single shunt will see phase C current, but that shunt will not indicate phase A or phase B current. This just means that a single shunt is not suitable for sinusoidal commutation.

Figure 6: Sine Commutation

Only single parallel connection can accurately indicate C-phase current

To overcome this problem, two shunts can be used, but this still does not solve the problem of detection error due to low PWM ratio, or the problem of no dynamic braking current or circulating current being detected. To this end, we will now examine the difference between the actual measured shunt current and armature current in a typical brushless driver.

The investigated system is a modified ladder drive with a double-sided switch with 3 shunts as shown in Figure 3. A Hall effect current sensor has been added to measure the actual armature current.

Figure 7: Ladder drive for double-sided switch with 3 shunts

Three parallel drivers

The current readings from the shunt and current sensor can then be compared.

The screenshot in Figure 8 depicts the shunt voltage (yellow) and current sensor output (blue).

Figure 8: Shunt Voltage and Current Sensor Output

Phase current waveform

Observation results:

Shunt current is a rectified form of armature current. For the same current, the sensor output amplitude shown is 50% greater than the parallel output.

In this drive, there is a considerable periodic difference between the shunt current and the true armature current.

At the current maximum, the relationship is more or less consistent, but as the current approaches zero and rises, the consistency is poor. There is no consistency when the current drops.

The reason for the inconsistency is that there are circulating currents in the bridge that do not flow through shunts or overlapping currents like sinusoidal commutation.

If we zoom in on time base inconsistencies, it is further described in Figure 9.

Figure 9: Phase Current Waveforms

Again, there is a reasonable relationship between the shunt current and the armature current during the current rise, but at the commutation point the shunt current drops to zero immediately and then starts to rise again while the armature current gradually drops. Because of the inductance, the armature current cannot be changed immediately. Most likely the induced current is not passing through the specific shunt, causing the difference.

At different times of the cycle, we can observe Figure 10.

Figure 10: Phase current waveform

The armature current is close to zero, but the parallel current is not. At the commutation point, both currents are zero, which is correct, but the shunt current rises rapidly, whereas the armature current does not.

We can clearly conclude that using a shunt in the negative line is a very poor way to measure current in many brushless drivers. The current flow in a three-phase bridge is obviously very complex, especially when considering multiple modes of operation, including dynamic braking, regenerative braking, forward and reverse drive, and FET branch switching options.

Symptoms of incorrect current measurement may be motor vibration and/or load-related instability. Coping with this instability can mean additional damping, which often results in poorer transient response.

An alternative is to use a shunt to measure the phase currents, but this immediately presents the challenge of having a large common-mode voltage swing across the shunt, which requires a very accurate resistor divider network and precision amplifiers. Also, heat in the shunt starts to become a problem when the current is greater than about 30 amps.

The logical choice is to use a galvanically isolated current sensor to measure the true armature current. These don’t have to be bulky or expensive. Figure 11 shows an example of a low-cost, compact installation of a Raztec sensor.

Raztec current sensor installed in 90A motor driver

The ring current sensor is only connected around the motor. Galvanic isolation completely eliminates common mode effects and allows accurate measurement of true armature currents. The output of the sensor is close to rail-to-rail, so no amplification is required for the C current signal to be fed directly to the A/D. Raztec RAZC series sensors are capable of measuring currents from up to 40A to up to 250A with a maximum outer diameter of 10mm.