Testing Open- and Closed-Loop Current Sensors

Open-loop current sensors have a definite price advantage over closed-loop counterparts. While you can use them in drive applications where torque accuracy isn't demanding — such as pump and fan markets — they're prone to saturation and temperature drift. So, how suitable is an open-loop sensor for a drive product, and what precautions should you follow to use its low-cost feature? Before answering these questions, let's look at why current sensors are built into drive products in the first place.

For modern ac drives to fulfill various applications, current sensors are built into the products for many functions:

Drive overcurrent protection

Typical scenarios include overload or short-circuited output. The current required from ac drive for delivering to its load exceeds its rating; however it needs fast and accurate regulation.

Drive output current polarity detection

Any inserted dead time between complimentary devices along a phase leg results in output voltage gain or loss requiring accurate compensation to ensure motor performance. This compensation scheme ^[1] needs detection for output current polarity.

Accurate current magnitude measurement

With current measurement from zero to rated magnitude, highly accurate motor states can be determined to facilitate high-performance torque delivery.

Ground current detection

The power converter system may be involved in a ground fault, which needs to be detected for safe shutdown.

As it stands today, determining the suitability of an open-loop sensor for a drive product depends on more system analysis. It's important to follow certain precausions before an oversized sensor can be determined to prevent from saturation during drive output short-circuited fault. Current sensors should maintain their accuracy within their temperature and output frequency (dc to 60 Hz) variation range. Among available current sensing technologies, current shunt and transformer are simple and cost effective. However, a shunt has circuit insertion loss, its impedance varies with frequency, and common mode noise is easily introduced into the measurement.

Current transformers have excellent high-frequency response. However they're not capable of measuring any dc component and generally have poor accuracy for low-frequency components (<5 Hz) ^{[2], [3]}. Hall effect sensors provide good accuracy from dc to their bandwidth of a few hundred kilohertz, and have been favored by ac drive manufacturers ^[4]. Closed loop Hall effect sensors have zero circuit insertion loss.

Laboratory Tests

Fig. 1, on page 44, illustrates the test lab setup. An ac drive is driving a 600 Hp dyne set. Six sensors, listed in Table 1, on page 44, are put into one of the output phase W. This table displays a sample of six sensors with their per unit prices. Sensors CL1, CL2, CL3, OL1, and OL2 have the same ratings. OL3 has different ratings. However, it's included for later short-circuit and temperature tests. Differences between CL2 and CL3, OL1, and OL2 are other sensor performance indices such as output driving capability and thermal drift of offset current. Notice the price ratio between an open- and closed-loop sensor with the same rating is about 1-to-2 and becomes smaller with increase in rating.

Separate power supplies are used for each sensor for maximum measurement isolation and fidelity. Length and position of the 3-phase drive output cables are adjustable.

Waveform reproducibility test

For the sensors to pick up the same output current waveform, they're mounted as close as possible to the output terminal block of the drive. The cables for the other two output phases, V and U, are arranged far apart from all current sensors.

Operational principles for open- and closed-loop Hall effect sensors ^[4] indicate both sensor types should operate better with dc than ac signal. AC signal, which is rich in high-frequency components, gives additional core losses and tests the boundary of the electronic amplifier within the sensor. Thus, we feel sufficient to test waveform fidelity with ac signal — in this case, 60-Hz output from the drive. Choosing this 60-Hz test condition allows us to incorporate a precision current monitor as the reference. We chose a Pearson 301X high bandwidth, highly accurate current monitor from Pearson Electronics ^[2] as the base for comparison.

Fig. 2 shows a collected plot from all current sensors and indicates that they track each other nicely.

Reflective wave test

Often, long cables are required to connect between an ac drive and its motor. The IGBT fast rise and fall times can easily excite the traveling wave effect because the motor is a poor termination for arbitrary length of cables. The traveling wave coming back from the motor and seen by the drive output terminals is a reflective wave. Because drive output voltage is clamped by the stiff dc bus, this reflective wave phenomenon is mainly a current concern on the drive side and a voltage concern on the motor side. Current sensor fidelity for sensing this reflective wave current is required in order for both overcurrent protection and dead-time compensation to function correctly.

Fig. 3 depicts a plot for current measurement from all sensors with 800 ft cable length between the drive and its motor. An output voltage (V_{u_busN}) with respect to the negative bus is also shown. The drive is switching at 2 kHz and outputting 53A_RMS current. In this test condition, the reflective wave has a high frequency component of about 160 kHz, which is above the bandwidth specified by manufacturers of the current sensors (open-loop 50 kHz, closed loop 100 kHz). However, all sensors still track the wave very well. Thus, the drive should have the correct information to operate. One caution is that reflective wave frequency varies with cable material, length, switching frequency, IGBT rise and fall times, etc. There are cases where the frequency is pushed into the megahertz range. Both open- and closed-loop Hall effect-type sensors may not be able to measure correctly in such frequency ranges ^[5].

Proximity effect test

In a drive environment, it's possible for any current sensor to be subjected to current sources with large magnitude and rich in frequency in proximity of the Hall effect gap. It's essential for the current sensor to represent the current passing through and not external to the sensor. Fig. 4 shows all waveforms under 60 Hz, 300A_RMS operation when a cable from another output phase is placed in parallel with all sensors along W phase. For this placement of an adjacent conductor, both types are immune from any proximity effect.

Short-circuit test

It's possible for the output of a drive to be short circuited. Drive manufacturers may use IGBT de-saturation phenomenon to detect short-circuit current. However, careful circuit layout is required for this de-sat protection to be immune from nuisance tripping. The alternate is short-circuit current magnitude detection by current sensors. This short-circuit current magnitude allowed in a drive is a function of IGBT short-circuit capability, feedback to control delay time, and short-circuit cable length.

Fig. 5 shows all sensor currents during a 20 ft cable short-circuit between W and another output phase. A short-circuit fault signal is also included to indicate when the control shut down is initiated. As evident, the short-circuit current reaches about 1250A, far surpassing the three times current rating of two of the 300A open-loop sensors (OL1 and OL2). As the result shows, both sensors saturate before successfully detecting the short-circuit current for the drive.

The higher rated open-loop sensor OL3 doesn't saturate and detects the current correctly. However, according to its data sheet, OL3 is going to saturate around 1800A, which may be reached if the output short-circuit cable is shortened even more. To accommodate short-circuit current capability, oversizing the sensor is a must if an open-loop sensor is to be used.

Fig. 5 also shows that all closed-loop sensors are capable of detecting the short circuit correctly. Although the closed-loop sensor is designed to only handle 1.5 to 2 times continuous rated current, it can detect very high current pulse for a short period of time because of its zero flux nature ^[4]. The limit for this high current measurement is thermal dissipation within the sensor and power driving capability of its power supply. Because this is a short-circuit protection function, the limit is easily met within the drive design.

Temperature effect test

All current sensors are moved into an environmental chamber for temperature cycling. The drive is outputting 300A_RMS current to the motor. Their dc drifts for an ambient temperature variance of 25°C to 85°C are listed in Table 2 against a 300A base. Table 2 indicates that a closed-loop sensor is three to 30 times better than an open-loop type in terms of dc drift.

All sensor ac components are calculated as error in percentage with respect to the reference established by the Pearson 301X current monitor, which remains outside of the environment chamber for the entire length of the thermal test. Fig 6, on page 47, plots the result. Errors for all closed-loop sensors remain within 0.5%, while errors for open-loop sensors range from 2.5% to 7.5%. Again, the closed-loop sensor is 2.5 to 11 times better than open-loop type. It's obvious that when a modern high performance drive demands its torque linearity to be within 5% error, any open-loop current sensor isn't suitable.

Are the Results Open or Closed?

A rigorous industrial environment requires ac motor drives to incorporate current sensors for protection and performance delivery. With limited applications, simple current shunt and current transformer techniques aren't widely used because drive current is rich in frequency spectra. Hall effect sensors are broadly deployed because of their dc current sensing capability, galvanomagnetic isolation, and zero circuit insertion loss.

Laboratory tests illustrate that both open- and closed-loop sensors are capable of reproducing drive current fidelity and are fairly immune to any existing proximity effect. Provided that the frequency of the reflective wave during long cable operation isn't much higher than the bandwidth offered by the current sensors, the reflective current wave can also be reproduced with reasonable fidelity. However, when this reflective wave frequency is above the current sensor bandwidth (i.e., 10 times more), troublesome drive operation can be predicted due to the loss of fidelity from the current sensors. Tabulated information on such boundary for loss of fidelity is usually provided from drive manufacturers to prevent their drives from malfunctioning. Beyond this boundary, means such as output filters or cable termination can be added, either at the drive side or the motor end, to limit such reflective wave phenomenon.

The most severe task for a current sensor in a drive product is when the sensor is subjected to output short circuit. The lab test reveals that the normally provided three times maximum current rating from open-loop sensor manufacturers isn't sufficient. To prevent saturation, oversizing is a must. Complex system analysis is needed before knowing how much oversizing, for an open loop sensor is required. Conversely, due to its zero flux compensation technique, closed-loop sensors don't exhibit such difficulty. Normal drive product design should provide enough driving power to closed loop sensors for high current magnitude compensation.

Temperature cycling test reveals that both dc and ac drifts for open-loop sensors can add up to about 8%, while closed-loop counterparts remain within 0.5% or so. This indicates that closed-loop sensors should be used when there's a tough requirement on torque performance from the drive product.

The price advantage of open sensors may only be realized for applications where temperature variation can be restricted. Although closed-loop sensors are more expensive, they are suitable for both commercial and industrial applications. Furthermore, they don't require being oversized when designed into a drive product.

References

Zhou, D. and Rouaud, D., “Dead-Time Effect and Compensations of Three Level Neutral Point Clamp Inverters for High-Performance Drive Applications,” IEEE Transactions of Power Electronics, July 1999, pp. 782-788.
Pearson Electronics Inc., “Standard Current Monitors,” www.pearsonelectronics.com.
Ghislanzoni, L. and Carrasco, J.A., “A DC Current Transformer for Large Bandwidth and High Common-Mode Rejection,” IEEE Transaction on Industrial Electronics, Vol. 46, No. 3, June 1999, pp. 631-636.
LEM, “Isolated Current and Voltage Transducers,” 2nd edition, www.lem.com.
Pankau, J., Leggate, D., Schlegel, D.W., Kerkman, R.J., and Skibinski, G.L. “High-Frequency Modeling of Current Sensors,” IEEE Transaction on Industry Applications, Vol. 35, No. 6, Nov./Dec. 1999, pp. 1374-1382.
“Current Sensing Solutions for Power Conversion and Intelligent Motion Applications,” Elektron, Vol. 13, No. 10, Oct. 1996, pp. 63-65.