Errors in converter circuits may be detected at various points. The responses to detected errors, however, may be very different. The term fast protection is used if an error signal is detected in the switch or in close vicinity to the switch and the respective switch is turned off immediately by the driver stage. The total response time of the switch may even be just a few 10 nanoseconds.
For error detection outside the switches, an error signal is initially transmitted to the control board, where a response to the error is triggered. This can be referred to as slow protection. Alternatively, the processes that occur may be assigned to converter control (e.g. the system’s reaction to overload).
Modern converters mostly combine slow and fast protection processes, depending on the specific application and design philosophy.
Detection and reduction of fault currents
Detection of fault currents
Figure 1 shows a voltage source inverter circuit. Here, the measuring points at which fault currents can be detected are marked.
Figure 1. Voltage source inverter (VSI) with detection points for fault currents
Fault currents can be classifi ed as follows:
- Overcurrent: detectable at points 1-7
- Arm short circuit: detectable at points 1-4 and 6-7
- Load short circuit: detectable at points 1-7
- Ground short circuit: detectable at points 1, 3, 5, 6 or by calculating the difference between 1 and 2
In general, to control short circuit currents, fast protection measures with direct control of the driver output stage are needed since the transistor has to be switched off (with reduced switching speed, active control provided) within tsc_max (typically 10 μs). To do so, fault currents may be detected at points 3, 4, 5, 6 and 7.
Measurements at points 1-5 may be taken using measuring shunts or inductive measuring current transformers (frequently used at point 5).
- simple measuring method
- low-resistance (1...100 mW), low-inductance power shunts needed
- measuring signal is highly sensitive to interference
- measuring values are not available with potential isolation
- far more expensive than measuring shunt
- measuring signal is less susceptible to interference than measuring shunt
- measuring values are available with potential isolation
At points 6 and 7, fault currents are detected directly at the IGBT/MOSFET terminals. Here, protection is provided by way of VCEsat or VDS(on) monitoring (indirect measuring method) and current sensing if a sense IGBT/sense MOSFET is used (direct measuring method). Figure 2 shows the principle circuits.
Figure 2. Fault current detection using a) current sensing and b) VCEsat monitoring
Current sensing with sense IGBT
In a sense IGBT, a few cells are combined to create a sense-emitter, generating two parallel current arms. Information is given by the conducted collector current as soon as it passes the measuring resistor RSense. At RSense = 0 the current division ratio between both emitters is ideal, corresponding to the ratio of the number of sense cells to the total number of cells. If RSense is increased, the current conducted in the measuring circuit will be reduced by feedback of the measuring signal. For this reason, resistance RSense should be within a range of 1...5 Ω to obtain a sufficiently accurate measurement result for the collector current.
If the turn-off current threshold value is only slightly higher than the transistor rated current, the current monitoring has to be made ineffective during IGBT turn-on because of the reverse recovery current peak of the free-wheeling diode.
For very high sense-resistances (RSense → ∞), the measuring voltage corresponds to the collector-emitter saturation voltage, meaning that the current sensing acts as VCEsat monitoring.
VCEsat monitoring makes use of the relationship between collector current and collector-emitter voltage (forward voltage, output characteristic) indicated in the transistor datasheets.
The collector-emitter voltage is detected by a fast, high voltage diode and compared with a reference value. If the reference value is exceeded, the error memory is set and the transistor turned off. The fast desaturation process in the transistor means that short circuits are quickly detected. If the transistor is not desaturated in the event of a fault (e.g. if slowly increasing ground fault currents and overcurrents are involved), the use of VCEsat monitoring for fault detection is restricted.
To guarantee safe turn-on of the IGBT during normal operation, VCEsat monitoring has to be blanked out long enough for the collector-emitter voltage to fall below the reference voltage. Since no short circuit protection exists during this period, the blanking time must not exceed tsc_max. Temperature dependency of the output characteristic as well as parameter tolerances have negative effects on VCEsat monitoring. However, the main advantage over current sensing using a sense IGBT is that this protection concept is applicable to every standard IGBT or MOSFET.
Fault current reduction
A better way to protect the transistor switch is to reduce or limit high fault currents, especially with regard to short circuits and low-impedance ground fault circuits.
A short circuit of type II will generate a dynamic short circuit overcurrent due to the increase in the gate-emitter voltage as a result of high dvCE /dt. The amplitude of the short circuit current may be reduced by clamping the gate-emitter voltage. Besides limiting dynamic short circuit overcurrents, static short circuit currents may also be decreased by reducing the gate-emitter voltage.
This will reduce transistor power losses during the short circuit phase. At the same time, overvoltage is decreased since the short circuit current has to be turned off at a lower level. This principle is shown in Figure 3.
Figure 3. Short circuit current limitation by reducing gate-emitter voltage
This protection method limits the static short circuit current to about 2.5...3 times the rated current in given applications in practice.
For more information, please read: