28 Aug 2014
Current Sense Transformer Performance
Paul Lee, Director of Business Development, Murata Power Solutions UK, describes some techniques to employ to achieve best performance for current transformers used for monitoring, control and protection
Current transformers are widely used in many areas associated with electronics for monitoring, control and protection.
However the use of current transformers is not well understood generally, and this can lead to poor performance.
Current transformer basics
An ideal current sense transformer has infinite inductance, zero winding resistances, perfect coupling and no parasitics.
Secondary current is simply primary current divided by turns ratio.
A real CT shown in Figure 1 has magnetising inductance Lm and parasitic components of leakage inductance Llk, winding resistances Rp and Rs, coupling capacitance Cc, winding self-capacitance Cs and core loss Rloss. Rb is the external ‘burden’ resistor and Rf is secondary impedance reflected to the primary. T1 is an ‘ideal’ transformer. The core can also magnetically saturate. These characteristics, interacting with external components, limit the performance of the CT over frequency and signal amplitude.
Figure 1 CT with parasitics and termination resistor
In operation, sensed current It divides between Rf, and Lm with only the partial current through Rf transforming to current through Rb, indicating a value less than expected. As frequency and Lm impedance reduces, the error increases, counteracted by reducing Rb. Figure 2 shows normalised accuracy for the Murata Power Solutions 53200C 200:1 CT which has Lsec = 8 mH, Cs = 30 pF, Llk = 50 µH and Rs = 34 ohms plotted against Rb. Low frequency bandwidth improvement with decreasing Rb is clear. Another low frequency limit is core saturation, occurring at about 450 milliteslas at 25 Celsius for ferrites. Figure 3 shows that saturation occurs well below the rated bandwidth of 50kHz, again improving with reduced Rb. The maximum current of 10 A within the rated bandwidth is a thermal limit. Note that the ‘diverted’ current into the magnetising inductance causes saturation not the sensed current directly.
At high frequencies, Cs and Rloss shunt away current from Rb and the accuracy falls, lower values of Rb again reducing the effect as shown in Figure 2. The rated bandwidth of this part, 500 kHz, is very conservative with a useful response up to several megahertz with low values for Rb.
The disadvantage of reducing Rb is that less voltage is developed across it for the same sensed current perhaps making the circuit more prone to noise pick-up and in some situations CT reset time is increased, described later.
Figure 2 Bandwidth Murata 53200CT
Figure 3 Saturation current limit Murata 53200CT
With a square wave sensed current, the ‘diverted’ current into the magnetising inductance increases from zero with time according to Im = Et/Lm where E is the decreasing voltage dropped across Rf as current through it decreases with time. Practically, Vb has an exponential ‘droop’, with reducing Rb or increasing Lsec lessening the effect. Using our Murata CT with a 10 µs pulse, the recommended value for Rb of 200 ohms gives 22.6% droop while Rb = 50 ohms gives just 9.5%.
A common sensed waveform has a rising current superimposed on the pulse from magnetising current in a flyback converter transformer or reflected output inductor current in a forward converter. Adding the droop effect to this waveform could leave Vb with a net fall making over-current detection problematic. Also, in current-mode control of switched-mode converters, Vb must have a positive slope to initiate switch turn-off.
Note that the sensed current slope and droop do not arithmetically add because the effects interact giving a net slope less than might be expected. If a net positive slope is required, the effect should be allowed for with a higher value for Lm, lower Rb or larger slope to the sensed current. Remember that the droop is exponential so the slope of Vb tends to a constant, higher value over time.
Pulsed current waveforms
A more realistic waveform for a pulsed current has edge skew and ringing. Predictably, reducing Rb decreases edge skew, the time it takes for the current Is through Lsec to reach its settling value, It/n .
Is is driven by the voltage E across Lsec during the skew time according to: dIs/dt=E/Lsec
E, varying and not very determinate, is the secondary voltage during the skew which depends on the primary compliance voltage and its source impedance. Cs and Lsec have an effect on ringing which is damped by Rb and core losses.
Cc rings with Lsec if the primary and secondary of the transformer have a common ground or have grounds with significant mutual capacitance. Toroidal CTs such as the Murata Power Solutions 5600C series can have lower values of Cc than bobbin-wound types.
Maximum width of pulsed currents is dependent on what droop can be tolerated and eventual saturation of the transformer core. With the 53200C CT, for a sensed current of 10 A, Tonmax is 23, 40 and 63.5 µs for Rb = 200, 100 and 50 ohms respectively before saturation. A rising slope to the sensed current reduces the allowed value. For example with 0.1 A/µs slope on the initial value of 10 A, Tonmax is just 21.1 µs for Rb = 200ohms.
Flux in a CT must be allowed to ‘reset’ between successive pulses otherwise ‘staircase saturation’ of the core can occur. Flux is proportional to current so for maximum speed of reset the magnetising current should decrease as quickly as possible driven by the reverse winding voltage from di/dt = – (E/L).
Rb however limits this voltage to – (Is . Rb). In practice unipolar current waveforms are sensed with the circuit of Figure 4. Here D1 blocks the reset current through Rb so Rr can be a high value giving fast, high voltage reset as long as there is also no reset current path around the transformer primary. However, the reset voltage increases in proportion to Rr and must be kept within the breakdown limit of D1.
Figure 4 Sensing unipolar current waveforms
In real circuits, secondary winding capacitance sinks current and limits the voltage and speed of reset. If all the magnetising energy from the transformer resonantly transfers to Cs, the maximum possible peak voltage Vr is:
For our 5300C CT with for example a primary magnetising current of 2 A peak, maximum secondary reset voltage Vr is 163 V
Rr could be dispensed with and D1 rated for 163 V plus any cathode voltage. Remember however that this voltage also transforms to the primary winding, in this case, (-) 163/n=-0.815 V
If for example, the CT were in the emitter of a bipolar transistor or source of a low threshold MOSFET, held off by 0V on its base or gate, this voltage could turn on the transistor, possibly with disastrous consequences.
D1 adds to Rb voltage drop reflecting to the primary causing additional magnetising current and reset voltage compared with the bipolar sensing arrangement of Figure 1. A Schottky diode is preferred for D1 for this reason. A fast recovery diode should anyway be used as any recovery current delays reset. Figure 5 shows the simulated resonant reset voltage of the circuit of Figure 4 with Rr = 10 kΩ, a 10 V/ 5 us pulse and with 1N4148 and 1N4003 diodes showing the effect of the long recovery of the 1N4003.
Figure 5 Reset with fast and slow diodes
In summary, effects of the termination impedance and associated components can be tailored to optimise performance for best accuracy over the widest frequency or pulse width range often exceeding the headline specifications of the part.
Page 1 of 1
About the author
Paul Lee is Director of Business Development for Murata Power Solutions in Milton Keynes UK. He has been with the company for more than 22 years, previously as Director of Engineering. He has an honours degree in electronics from the University of Kent and is a Chartered Engineer.
Murata is a leading manufacturer of electronic components, modules, and devices. The complete range of this Technology house includes ceramic capacitors, resistors/thermistors, inductors/chokes, timing devices, buzzers, sensors and EMI suppression filters. Whilst the company gets the majority of its revenue from its ceramic capacitor products, it is also the world leader in Bluetooth® & WiFi™ Modules, the world's no.1 manufacturer of board-mount DC-DC converters and is a key manufacturer of standard and custom AC-DC power supplies. Established in 1944, Murata is headquartered in Japan and has European offices in Finland, France, Germany, Hungary, Italy, the Netherlands, Spain, Switzerland and the UK.
Most popular articles in Test & measurement
Share this page
Want more like this? Register for our newsletter