Application Note, V 1.1, May 2001 ANPS061E Guideline for determining switching losses associated with switching resistive, capacitive and inductive loads. by Christian Arndt and Michael Mueller-Heiss Automotive Power Never stop -1- thinking. ANPS061E Guideline for determining switching losses 1. Abstract 3. General This Application Note is intended to provide help in all cases where FET switching losses due to repetitive (such as during PWM mode) or slow switching processes must be taken into account. It focuses on thermal considerations for semiconductor switches in switched DC applications. A general idea about how switching losses occur and an outline about what amount of power can be dissipated during switching is given. The effect of these power losses for the temperature of the semiconductor switch is covered as well. The Application Note covers single and/or repetitive switching for ohmic, capacitive and inductive loads. If a load is switched in a DC application, the switching device passes through various operating states during the switching operation. During the switch-on (referred to the load): process, these are 1) No-load: A VDS=Vbb IDS=0 IDS 2. Introduction 2) Matching If a semiconductor device is used in a DC application as a switch for a load, losses always occur in that device. Basically these losses can be subdivided into • steady-state losses (losses during the ON state) and • dynamic losses (losses during switching). VDS=Vbb/2 IDS≠0 UDS V UDS V UDS V A Calculating the steady-state losses is relatively easy. They generally result from componentspecific and load-typical parameters. In the case of semiconductor devices based on a field-effect transistor principle, these parameters are the on-resistance RDS(on) and the load current ILoad. However, calculating the dynamic losses occurring during switching processes is more difficult. IDS 3) Rated operation. VDS≈0 IDS=ILoad A IDS During the switch-off process, each switching device passes through these operating states in reverse sequence. In the case of inductive loads, commutation processes may additionally come into play. These will be discussed in further detail later. Let us just say for now that during commutation processes the energies stored in the load inductance are discharged via the switching device at voltages greater than the operating voltage Vbb. Application Note page 1 of 7 V1.1, 2001-05 ANPS061E Guideline for determining switching losses It can be seen that the maximum losses occur at half the nominal voltage and half the nominal current. This corresponds to the matching case. 4. Switching resistive loads The following section deals first with switching losses associated with purely resistive loads. However, the statements made here are also relevant to the subsequent sections. In the 3 diagrams (no-load, matching and rated operation) it may be observed that all the operating states of the switching device lie on a dotted gray line. This is the load line. This means that the switching losses of each switching operation are influenced by the load. The losses occurring at this point are: P̂Switching_max = PMatching = 4.1.2. Vbb I Load Vbb 2 • = 2 2 4 • R Load Calculation p Switching = p Switching = (1) p Switching = To calculate the switching losses, it is now important to know how the current and voltage vary over time. A distinction may basically be drawn between 2 cases. Linear voltage and current characteristic Assumption 4.1.1. Assuming that the current and voltage vary linearly over time, the switching losses occurring at each instant of the switching process are as follows: p Switching PMatching = TSwitching 1 T Switching 0 æ u i ö ç dt çV • I nom è bb 11 x(1 − x )dx 10 1 0 4.1. (2) If the averaged losses for linear current and voltage responses are now normalised to the reference value, these losses will be as follows: To examine this further, it is useful to introduce a reference variable which is exclusively a function of the load. A good reference variable is the power dissipation in the device during matching. This is given by formula (1) as: PMatching = Vbb 2 4 • R Load (x − x )dx = x2 2 2 − x3 3 1 = 0 1 6 (3) 1 4 2 • = = 67% 6 1 3 The averaged switching losses for a resistive load and linear current and voltage characteristics are therefore approx. 70% of the losses occurring in the matching case, irrespective of the switching time. In reality, the 67% calculated may still increase slightly as a result of the Ron losses additionally produced close to rated operation, so that 70% is a wholly realistic figure. 4.1.3. Validity and typical applications Switch-on process 1 0.9 VDS/Vbb 0.8 0.7 IDS/ILoad 0.6 V DS / V bb I DS / I Last 0.5 Pv 0.4 0.3 0.2 PV 0.1 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Figure 1 – Normalized switching losses for linear current and voltage characteristics Application Note This applies to switch-on and switch-off processes, provided no parasitic inductances must be taken into account. Linear current and voltage characteristics may be observed during slow switching processes. This may occur, for example, if a slowly rising input signal is applied to the device by an external triangular wave generator. This is ® possible e.g. with discrete FETs, TEMPFET ® and HITFET devices. As will be shown later, the linear current and voltage response constitutes the less favourable case in thermal terms, since the highest losses occur here. If the true current and voltage page 2 of 7 V1.1, 2001-05 ANPS061E Guideline for determining switching losses response is not known, this response should be taken as the basis for safety reasons. 4.1.4. Example By way of example, let us now consider the turn-on process of a BTS149. The BTS149 has been turned on by an external triangular signal. With a supply voltage of Vbb=10 volts, it drives a load of 2.18 ohms. For current (Ch2: 1A/div), voltage (Ch3) and losses (Math1), the following waveforms were observed: voltage across the device has fallen to approx. 20% of the nominal voltage at switch-on, and the current has increased to approx. 80% of the nominal current, the switching losses at each moment of the switching process are as follows: Switch-on process 1 0.9 VDS/Vbb 0.8 IDS/ILoad 0.7 0.6 V ds I ds Pv 0.5 0.4 PV 0.3 0.2 0.1 PV VDS 0 0 IDS 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Figure 3 – Normalized switching losses for piecewise linear current and voltage characteristics 4.2.2. Calculation The switching losses are now: p Switching = Figure 2 – Slow switch-on process of a BTS149 For a switching time of approx. 3.5 ms, this produces average switching losses of: 1 p Switching PMatching p Switching PMatching p Switching PMatching = T Switching = p Switching = Vbb • I nom /4 p Switching p Schalt = 59% (4) p Schalt 4.2.1. Piecewise linear current and voltage characteristic Assumption If the current and voltage are assumed to exhibit a piecewise linear response, whereby after approx. 30% of the total switching time the Application Note 0 æ u i ö ç ÷dt çV • I ÷ nom è bb 0,3 æ 2 ö 2 ö 0,7 æ ç 8x − 64x ÷dx + ç 4 − 6x − 4x dx ç 3 ç 9 ÷ 35 49 0 è 25 è (5) 4x 2 64x 3 = − 3 27 0,3 0 4x 3x 2 4x 3 + − − 25 35 147 0,7 0 3 8 14 147 7 49 = − + − − = ≈ 0,116 25 125 125 3500 750 420 PAnpassung 4.2. T Switching 0 236µ36µ 1 • *W 3500µ5 0.01VVs = 11.5W TSwitching 0,7 æ 1 2x 1 é0,3 æ 8x 8x ö 4 2x ö ù ê ò ç (1 − ) • ( )÷dx + ò ç ( − )•( + )÷dx 1ê 0 è 3 3 ø 5 7 ø 0 è 5 7 ë u • i • dt Tswitching 1 = 49 4 • = 46, 6 % 420 1 In the case of a piecewise linear current and voltage response, the averaged switching losses for a resistive load irrespective of the switching time are approx. 50% of the losses which can occur in the matching case. In reality, the 47% calculated may increase slightly as a result of the Ron losses additionally page 3 of 7 V1.1, 2001-05 ANPS061E Guideline for determining switching losses produced close to rated operation, so that 50% is a wholly realistic figure. 4.2.3. PMatching Validity and typical applications To a first approximation, a piecewise linear current and voltage characteristic may be observed primarily with fast switching processes. It is attributable mainly to the twostage gate charging process. This is due to the change in the input capacitance during the switching operation. With highside switches ® such as the PROFET devices, the effect of the charge pump is additionally present. This applies to switch-on and switch-off processes provided no stray inductances must be taken into account. Fast switching processes exist if the device is driven via a low impedance directly by the driver. This is possible in the case of discrete ® ® ® FETs, TEMPFET , HITFET and PROFET devices. 4.2.4. p Switching 1.83µ.83 1 • *W 34µ4 0.005VVs = = 43% 25W 5. Switching capacitive loads The following section deals with the switching losses associated with capacitive loads. These include for example RC networks or lamps whose time constants are much greater than the switching time of the device itself. With capacitive loads, the losses produced in the device during each switching operation are again influenced exclusively by the load. However, in the case of switch-on processes the power dissipation in the device during matching under rated conditions is less critical than the power dissipation during matching under inrush conditions. This is given by formula (7) as Example By way of example, let us now consider the turn-on process of a BTS149. The BTS149 has been driven low impedance by a square wave signal. With a supply voltage of Vbb=10 volts, it drives a load of 1 ohm. For current (Ch2: 2A/div), voltage (Ch3) and losses (Math1), the following waveforms were observed: VIN PV P' Matching = Vbb I inrush Vbb 2 • = 2 2 4 • Rinrush This will be designated “transient matching power” P’Matching in the following. Iinrush is the maximum inrush current flowing and Rinrush the respective total resistance at the moment of switch-on. P' Matching = 6..10 • PMatching Figure 4 - Fast turn-on process of a BTS149 R1 For a switching time of approx. 35 µs, this results in average switching losses of: PMatching = T Switching u • i • dt Tswitching Vbb • I nom /4 Application Note (8) In the case of RC networks, Rinrush results to a first approximation from all the resistances not “short circuited” by parallel capacitances. The following simple example illustrates this. From the resistances shown in Figure 5, VDS p Switching (7) In the case of lamps, Iinrush may well be 6 to 10 times Inom. Consequently, for lamps the formula is usually: IDS 1 (6) R2 C1 Figure 5 – Simple RC circuit R1 results as the effective inrush resistance Rinrush. page 4 of 7 V1.1, 2001-05 ANPS061E Guideline for determining switching losses 5.1. Linear current and voltage characteristic Linear current and voltage characteristics may be observed during slow switching processes where the switching time is, however, still well below the smallest time constant of the load. This may occur if a slowly rising input signal is applied to the device by an external triangular wave generator. This is possible, for example, in ® ® the case of FETs, TEMPFET and HITFET devices. Analogously to Figure 1 and formula 3, the averaged switching losses are given by: p Switching P' Matching p Switching P' Matching = 1 4 2 • = = 67% 6 1 3 ≈ 6..10 • 6. Switching inductive loads p Switching PMatching (9) load − dependent For slow switch-on processes the averaged power dissipations are therefore approx.70% of the “transient matching power”. For slow switch-off processes a distinction must be drawn between how quickly the device is turned off again after switch-on. If switch-off takes place immediately after switch-on, the averaged switching losses according to formula 9 are approx. 70% of the “transient matching power”. If switch-off occurs after rated operation has been attained, the averaged switching losses according to formula 3 are approx. 70% of the matching power. 5.2. For fast switch-on processes the effective power dissipations are therefore approx. 50% of the “transient matching power”. For fast switch-off processes a distinction must be drawn between how quickly the device is turned off again after switch-on. If switch-off takes place immediately after switch-on, the effective switching losses according to formula 9 are approx. 50% of the “transient matching power”. If switch-off occurs after rated operation has been attained, the effective switching losses according to formula 3 are approx. 50% of the matching power. The following section deals with the switching losses associated with inductive loads. These include such loads as coils and valves which have no additional active freewheel circuitry (such as a freewheeling diode). In the case of inductive loads, the losses produced in the device during each switching operation are influenced by the load and by the active Zener or avalanche voltage of the device. If the switch-on time is much shorter than the time constant of the load, no switch-on losses must be taken into account. This is mostly the case. However, if the device switches off the inductive load from rated operation, the energy stored in the load inductance will be dissipated via the switching device. This process is known as commutation. Piecewise linear current and voltage characteristic Piecewise linear current and voltage characteristics may be observed with fast switching processes. This is the case when the device is driven via a low impedance directly by a driver. This is possible with discrete FETs, ® ® ® TEMPFET , HITFET and PROFET devices. Analogously to Figure 3 and formula 5, the averaged switching losses are given by: p Switching P' Matching p Switching P' Matching = 0.116 • ≈ 6..10 • Application Note 4 = 46.6% 1 p Switching PMatching Vbb induct. Load IDS Switch VAZ GND (10) load − dependent page 5 of 7 V1.1, 2001-05 ANPS061E Guideline for determining switching losses To a first approximation the switch-off losses are given by: Wtot = W L + WVbb Wtot = 1 •L•I 2 + 2 t Switching 0 (Vbb • i(t)) • dt (11) 1 1 • L • I 2 + • Vbb • I • t switching 2 2 di I = L• dt t Switching (12) If a load is switched on or off on single occasions, a distinction can be drawn between VAZ the average switching losses produced as shown in the following table: we get Wtot = 1 L 1 • L • I 2 + • Vbb • I 2 • 2 2 V AZ − Vbb Wtot = æ ö Vbb 1 • L • I 2 çç 1 + 2 è V AZ − Vbb (13) For each switch-off process the switching losses are therefore given by t Switching ON (14) inductive Wtot Switching process OFF ON æ V AZ ö 1 • L • I 2 çç 2 è V AZ − Vbb PSwitching = Load capacitive Wtot = Non-repetitive switching 7.1. With V L = V AZ − Vbb = L • The previous sections explained in greater detail how to determine the equivalent switching losses for various loads to a first approximation. VL This section will now attempt to clarify precisely what this means in terms of thermal loading of the device. In practice there are 2 cases which will be examined in detail below. resistive Wtot = 7. Thermal considerations OFF (immediat ely after switch-on) OFF (during rated operation) OFF Switching time TSwitching << tSwitching >> tSwitching << tSwitching >> tSwitching << tSwitching >> tSwitching << TSwitching >> Formula tSwitching << tSwitching >> 5 3 / 14 5 3 5 3 10 9 10 9 Table 1 – Overview for determining the switching losses Further examination of the thermal implications now requires precise knowledge of the thermal conditions of the application as a whole. For short-duration switching operations up to a few milliseconds, the Zthjc values specified in the data sheets can be used to a first approximation. However, for longer switching operations, precise knowledge of the overall thermal conditions is necessary. For some products, the data sheet may also indicate how the thermal linkage behaves on a 6cm² copper PCB. If this coincides with the application, this can be taken as the basis. Application Note page 6 of 7 V1.1, 2001-05 ANPS061E Guideline for determining switching losses To determine the temperature swing of a switching operation, the following steps must be followed: measure the switching time tswitching determine the thermal impedance Zth(tswitching) from the data sheet calculate the junction temperature of the device using formula 15: ϑ j = PSwitching • Z th (t Switching ) + ϑ amb 7.2. (15) Repetitive switching (PWM operation) If a load is repeatedly switched on or off, in most cases it must be ensured that the junction temperature is not exceeded during quasisteady-state operation. However, the on-state losses (PV(on)) and the switching losses must be taken into account here. For a device operated in PWM mode at a frequency fSwitching, we get: Pv = PSwitching(application) + PR(ON) (t ON ) 7.3. Typical application To illustrate the approximating calculation methods described before, one examples is given below. Temperature swing of a BTS142D for a nonrepetitive switch-on process Given: A BTS142D mounted on a 6cm² copper PCB is controlled externally by a triangular wave generator. The switching time is extended from the original typical switching time of 60µs to tSwitching=1ms. The BTS142D drives a 1 ohm resistance as a load. The supply voltage Vbb=14 volts. 7.3.1. Required: Of interest is the final temperature which the device attains after the switching process for an ambient temperature of 50°C. 7.3.2. Solution: (16) Using formula 2 we have The total switching losses Pswitching (application) are given by: ( P̂Switching_max = PMatching = ) PSchalt(Applikation) = W Schalt(ON) + W Schalt(OFF) • f Schalt ( ) Vbb 2 (14V)² = = 49W 4 • R Load 4 • 1Ω (20) = PSchalt(ON) • t (ON) + PSchalt(OFF) • t (OFF) • f Schalt (17) Pswitching(ON) being the switching losses for each switch-on process and Pswitching(OFF) being the losses for each switch-off process. These may differ depending on the application and are therefore detailed separately here. Assuming a linear current and voltage characteristic here, the averaged power dissipation according to formula 3 is given by: p Switching = 0.67 • PMatching ≈ 33W (21) For a single pulse, the BTS142D data sheet gives an approximate ZthJA(@6cm²)=0.2K/W. The on-state losses, PR(ON) (tON), are given by: In accordance with formula 15, the final temperature is now therefore: PR(ON) (t ON ) = P(ON) • duty cycle 2 = I load • RON (150°C) • t ON • f Switching ϑ j = PSwitching • Z th (t Switching ) + ϑ amb (18) The average junction temperature of the device is therefore expressed by formula 19 ϑ j = 33W • 0.2K/W + 50°C = 56.6°C ϑ j = PSwitching • (Rthjc + Rthca ) + ϑ amb The device thus undergoes a temperature swing of approximately 6.6K and therefore attains a junction temperature of 56.6°C after the switchon process. Application Note (19) page 7 of 7 (22) V1.1, 2001-05 ANPS061E Guideline for determining switching losses 8. Disclaimer 9. Infineon goes business Excellence Edition 2000-07-14 Published by Infineon Technologies AG, St.-Martin-Strasse 53, D-81541 München, Germany © Infineon Technologies AG 8/d/yy. All Rights Reserved. Attention please! The information herein is given to describe certain components and shall not be considered as warranted characteristics. “Business excellence means intelligent approaches and clearly defined processes, which are both constantly under review and ultimately lead to good operating results. Better operating results and business excellence mean less idleness and wastefulness for all of us, more professional success, more accurate information, a better overview and, thereby, less frustration and more satisfaction.” Terms of delivery and rights to technical change reserved. We hereby disclaim any and all warranties, including but not limited to warranties of non-infringement, regarding circuits, descriptions and charts stated herein. Dr. Ulrich Schuhmacher Infineon Technologies is an approved CECC manufacturer. Information For further information on technology, delivery terms and conditions and prices please contact your nearest Infineon Technologies Office in Germany or our Infineon Technologies Representatives worldwide (see address list). Warnings Due to technical requirements components may contain dangerous substances. For information on the types in question please contact your nearest Infineon Technologies Office. Infineon Technologies Components may only be used in life-support devices or systems with the express written approval of Infineon Technologies, if a failure of such components can reasonably be expected to cause the failure of that life-support device or system, or to affect the safety or effectiveness of that device or system. Life support devices or systems are intended to be implanted in the human body, or to support and/or maintain and sustain and/or protect human life. If they fail, it is reasonable to assume that the health of the user or other persons may be endangered. http://www.infineon.com Application Note page 8 of 7 Published by Infineon Technologies AG V1.1, 2001-05