Designing RCD Clamps for TOPSwitch-FX  Power Supplies Application Note AN-28

This Application Note provides quick design curves for ‘RCD clamps’ in TOPSwitch-FX based power supplies. The Designer just needs to input the leakage inductance and the TOPSwitch-FX device Current Limit at low line. Then the solid-line curves (Figures 1-3) provide several possible R-C combinations to ensure safe and efficient operation. The superimposed dashed line curves provide the corresponding dissipation in the clamp with the Power Supply operated at full power at 85VAC line input. The Designer can thus select the most appropriate R-C combination for his application. The curves apply only to power supplies designed for Universal Input (85-275VAC). They assume 3F/Watt of input bulk capacitance and also require the multifunction pin of the TOPSWitch-FX device to be dedicated to applying suitable ‘Current Limit reduction at high line’. By using any of the R-C combinations displayed, and adhering to the design guidelines, a 15% safety margin on the Drain-Source Voltage is usually assured even at an input line condition as high as 275VAC. The Designer can in most cases expect the RCD clamp dissipation to be lower or comparable to a 200V zener clamp, but this is true only at maximum load. At lighter loads, the RCD clamp is more dissipative than a zener clamp and the Power Supply efficiency will be correspondingly worse.

RCDClamp Off-line Flyback.doc

1

In all cases, the entire dissipation in an RCD clamp is assumed to occur in the Resistor alone. Introduction In a typical flyback power supply, both the leakage inductance of the transformer and the Secondary side trace inductance need careful attention. Besides lowering the Efficiency of the Power Supply, they both have the potential of destroying the TOPSwitch instantly, by inducing large Drain to Source voltage spikes at turn-off. An Overvoltage clamp across the Primary winding is the usual remedy and functions by

QUICK START 1) Measure the in-circuit leakage inductance LLK in Henries. 2) Measure the peak of the Drain current waveform ILIM at rated power at 85VAC. 1 2 3) Calculate ELK   LLK  I LIM (J). 2 4) Tentatively select one of the three clamp capacitor values 4.7nF, 10nF or 22 nF. 5) For a VOR of 105V select Fig. 1, for 135V select Fig. 2, and for 150V select Fig. 3. 6) Draw a vertical line from the calculated ELK to intersect with the solid line curve corresponding to the selected C and KRP. 7) Interpolate between the dotted line curves to estimate the clamp Dissipation. 8) If the Dissipation is acceptable, the ycoordinate provides 1/RC. Calculate R.

December, 97

AN28 If the dissipation is too high, reconsider the values for C, VOR, KRP and also try to reduce LLK. Refer to the text of this Application Note to confirm the design and its implementation.

absorbing crossover.

excess

energy

during

the

Overvoltage clamps can be simple zener/TVS clamps or several variations of Resistor-Capacitor-Diode Networks, generically known as ‘RCD’ ‘clamps. The simple 200V zener clamp was commonly used and recommended for most TOPSwitch-based designs, primarily because it was relatively easy to design and provided an almost fixed and unambiguous clamping voltage level irrespective of line, load, transformer characteristics etc. However, an RCD clamp offers an opportunity for improving electrical performance and/or cost. It is suited for TOPSwitch-FX based designs because these devices allow for setting the Primary side current limit to the load requirement of the application, and also allow for decreasing this limit with increasing line voltage. This forms a necessary requirement for the safe use of an RCD clamp.

RCDClamp Off-line Flyback.doc

2

The RCD clamp chosen for discussion in this Application note is shown in Figure 4. It should not be confused with the RCD ‘snubber’ in common use some years ago which looks schematically identical to Figure 4, but differs in the actual values of R and C used, and the resulting purpose. Whereas, an RCD snubber functions as a ‘switching aid’ network to lower switching losses by shaping the dV/dt at turn-off, the RCD clamp serves only to clamp the excess energy arising from the leakage, to a safe level of VDS, at the instant of turn-off. The snubber capacitor is small enough (relative to the R used), to be able to fully discharge every switching cycle, whereas the RCD clamp has comparatively large R and C values that prevent the capacitor from ever fully discharging in steady operation. Therefore, generally speaking, in an RCD clamp, too low an R will cause excessive dissipation, whereas if the C is too low it will charge up excessively, causing the VDS to exceed the breakdown voltage rating ‘BVDSS’ of the TOPSwitch. Choosing the correct values of R and C is not a trivial task since the clamping level is affected by almost every other parameter and operating condition. This Application Note attempts to help the Designer bypass the complexity involved in designing a proper RCD clamp by providing quick design curves to help pick R and C. It also contains relevant design guidelines and tries to highlight some of the key issues involved in the selection and implementation of a good RCD clamp design. It however makes no attempt to quantitatively predict the behavior of the clamp at loads less than the maximum rated load, and the Designer is expected to validate his entire design on the bench.

December, 97

AN28 Zener to RCD Clamp The functioning of RCD clamp may be intuitively understood by comparing it to a zener clamp. The difference is that unlike the fixed voltage clamping action of a zener clamp, an RCD clamp has a varying clamping level. The dissipation of a zener clamp is given by the following equation:

VCLAMP 1 PCLAMP   LLK  I PK 2  f  2 VCLAMP  VOR with

VCLAMP  VZ where, VZ is the zener clamp voltage in Volts, LLK is the total leakage as seen on the Primary side expressed in Henries, IPK is the peak of the drain current waveform in Amps, VOR is the reflected output voltage in Volts and f is the switching frequency in Hz. For an RCD clamp we can approximate the dissipation equation to be the same equation as given above except that we use for VCLAMP

V1  V2 2 which represents the average voltage across the clamp capacitor. Here, the capacitor is considered to be charging from V1 to V2 (expressed in Volts) during the crossover transition and then discharging from V2 to V1 during the remainder of the switching cycle. VCLAMP 

For either the zener clamp, or the RCD clamp the corresponding voltage across the Drain and Source is clamped to

VDS  VIN  VCLAMP

RCDClamp Off-line Flyback.doc

3

where VIN is the input DC voltage to the converter stage. For the RCD clamp we have assumed here that the clamp capacitor is quite large and so V1 and V2 are almost the same. So whereas for the zener clamp, the VDS was almost fixed at high line, assuring unconditionally safe operation, for the RCD clamp V1 and V2, take on values that change with line, load, leakage inductance, VOR and KRP. This necessitates a complicated mathematical model to predict the values V1 and V2, as the one used for generating the curves of this Application Note. RCD under light load and high line It is important to realize that the voltage levels V1 and V2 on the clamp capacitor are determined by the energy in the leakage just prior to turn-off. Therefore the voltage levels will decrease if the peak of the switching current waveform is less, as for example when the load is decreased or the line voltage is increased. Lower clamping voltage levels cause higher dissipation as is evident from the equation for PCLAMP. We can therefore make some intuitive deductions on how the RCD clamp dissipation will be affected by variations of line and load, in comparison to a 200V Zener clamp. 

Though we can design the RCD clamp well enough so as to assure the highest possible clamping voltage level at full load, if the load decreases, the clamping voltage level will start. Though the absolute value of the RCD clamp dissipation at light loads is less than at full load, a zener clamp would always behave better at light loads because of

December, 97

AN28 its ability to maintain the clamping level. 

For the same reason, as we increase the line voltage, peak current will fall, and the RCD clamp will fare poorly in comparison to a zener clamp.

However, in absolute terms, the worst case RCD clamp dissipation still occurs at full load and low line condition. It is therefore, this condition that has formed the basis for generating the dotted Dissipation curves of this Application Note. RCD at full load and low line The RCD clamp can in fact give better performance at full load and low line than a 200 Volt zener clamp. This happens because the varying clamping level of an RCD clamp can be turned into an advantage in some cases. To understand how this may occur, we must first realize that a zener clamp is usually chosen to be a 200V device only because of the need to keep the VDS at high line to a safe level for the TOPSwitch. If it was possible, we could have substituted a 300V zener at low line, reaping an improvement in he clamp dissipation without compromising the VDS safety margin. It is obvious that this is not practicable for a zener clamp, but the RCD clamp has the potential to do this quite automatically on its own. So for example, if the clamping level at high line is set to be comparable to that provided by a 200V zener clamp, then as the line voltage decreases and the peak current increases, the clamping level would be pushed higher. This would provide the sought after improvement as compared to using a fixed 200V zener clamp. Therefore, a properly designed (and correctly implemented) RCD clamp design has the potential to provide higher efficiency under full load and low line condition than a zener clamp.

The reason that this potential may not always be fully realizable, is the need to maintain a higher safety margin on the VDS

RCDClamp Off-line Flyback.doc

4

at high line for an RCD clamp than that provided by a more ‘predictable’ zener clamp. Tolerances, mainly related to the Clamp Capacitor are part of this consideration as also the behavior of the RCD clamp under overloads and the consequent need for reduction in the Primary side Current limit at high line, as explained below. Current Limit Reduction at High Line The implementation of a good RCD design is as important as the design of the clamp itself. The single most important factor in ‘allowing’ the use of an RCD clamp is the requirement that the current limit of the device must be lowered at high line. A typical flyback power supply has a lower peak switching current at high line than at low line. The RCD design curves presented in this Application note account for this, and compute the required R and C values based on the expected energy at high line. If the energy is even briefly much more than the anticipated value, the clamp capacitor would charge up beyond safe limits leading to instantaneous damage to the TOPSwitch. Such a situation can occur during any overload condition for example. Whereas in a zener clamp, this would only cause a momentary increase in clamp dissipation, in an RCD clamp the increased dissipation is also accompanied by dangerous voltage levels, capable of causing instantaneous damage to the TOPSwitch. It is therefore important, when using an RCD clamp, to ensure that the current peak at high line cannot exceed normal operating peak current values by too wide a margin. It is possible to implement this by using a TOPSwitch-FX, and the bench procedure to ensure this is outlined below.

December, 97

AN28 The starting point of the procedure is Figure 18 of the TOPSwitch-FX datasheet, titled ‘Current Limit Reduction with Line Voltage’. To allow for component tolerances, an ‘overload margin’ of 10-15% is planned to be present at both extremes of line voltage. Step 1: Starting at low line (85VAC), only RIL is first introduced and is tentatively selected according to the curve titled ‘Current Limit vs. External Current Limit Resistance’ on Page 30 of the TOPSwitchFX datasheet. This value of resistor is confirmed to be correct if the power supply is able to deliver about 10-15% more than the required maximum load, but not more. Further, this ‘overload margin’ must be confirmed by allowing the power supply to run for some time for thermal settling to have occurred, since the power capability of a TOPSwitch based power supply at low line depends strongly on the temperature of the TOPSwitch device. Step 2: Then, with the power supply continuing to deliver its rated power, the input line is slowly raised by means of a variac or programmable AC source, to 275VAC. RLS is now introduced and if it has been chosen correctly, power supply should start power limiting at only about 10-15% higher than maximum rated load, and not more. If the overload margin is more than this, RLS must be decreased further. Step 3: Thereafter, the Designer must again reduce line voltage to 85VAC and check that the power supply can still deliver just 10-15% higher than maximum rated power. If not, then RIL needs to be decreased slightly and the steps above repeated sequentially.

RCDClamp Off-line Flyback.doc

5

As a final confirmatory test on the effectiveness of the RCD clamp and its implementation, the power supply must be snap turned-on at 275VAC into a constant current load which is just a little less than that which would cause power limiting to occur in normal operation. The VDS is monitored on an oscilloscope and should never be seen to come too close or to exceed the maximum rating of the TOPSwitch ‘BVDSS’ even momentarily. If it does, the clamp resistor may have to be reduced further. Note that since Current Limit reduction and Duty cycle reduction are mutually exclusive functions in TOPSwitch-FX, the use of an RCD clamp is restricted to applications that leave the Multifunction pin free to allow its use for implementing Current Limit reduction at high line. Therefore any TOPSwitch-FX based power supply application that demands duty cycle limiting must use a normal 200V zener clamp, not an RCD clamp. Advantages of the RCD Clamp We now summarize the main advantages of the RCD clamp as compared to a zener clamp: 

The RCD clamp is usually more costeffective, and preferred by designers for this reason alone.



It is possible to design an RCD clamp that has lower dissipation than a zener clamp when operated at full load and low line. Very rarely will the efficiency of an RCD clamp be worse than a zener clamp under these conditions.

Disadvantages of the RCD Clamp

December, 97

AN28 On the other hand the following are the main disadvantages:

inductance at rated load with 90VDC applied to the input of the converter stage.



1 ELK   LLK  I LIM 2





In a practical implementation, an RCD clamp may have slightly higher dissipation at high line than a 200V zener clamp. At light loads, the RCD clamp will have higher dissipation than a 200V zener clamp. Thus an RCD clamp will not be a good choice for solutions requiring very low standby power. Since the clamp capacitor voltage levels depend on so many factors, it is important to be very sure of the worst case leakage inductance in production, as also tolerances of the clamp components, in particular the clamp capacitor. Here it is pointed out that the design curves are given for a fairly good capacitor like a film type. Commercial quality ceramic capacitors for example, have very wide variations on the initial tolerance. In addition, their capacitance is also affected by voltage, temperature and aging characteristics. These must all be considered in assuring the stated clamp capacitance on which the design curves are based.



An RCD clamp must be implemented alongwith Current Limit reduction at high line. For TOPSwitch-FX, this means that the Multifunction Pin is dedicated for this, and cannot provide any other function. Interpreting the RCD Design Curves The x-axis of the R-C selection curves

A simplifying term ‘ELK’ is introduced as the x-axis of all the R-C design curves. It is defined as the energy residing in the leakage

RCDClamp Off-line Flyback.doc

6

2

where LLK is the effective in-circuit leakage inductance in H seen on the Primary side and ILIM is the set Current limit at low line. A full energy flow analysis reveals ELK is not the only energy that finds its way into the clamp (or else all clamps would be identical in terms of dissipation!). But ELK does form the seed from which the full energy loss term is created. The curves are thus based on this term and it is important to know what it is. ELK itself depends on two terms ‘ILIM’ and ‘LLK’, and in a practical case these can be simply measured as described below. Measuring ILIM The trace leading directly to the Drain of the TOPSwitch is cut and a loop inserted to allow insertion of a current probe connected to an oscilloscope. Another oscilloscope probe is connected across the input bulk capacitor. The power supply is made to deliver its rated output power at 85VAC input and the oscilloscope is triggered on the bulk capacitor voltage waveform with the trigger level set at 90VDC. (If necessary, the input line may need to be lowered slightly to ensure stable triggering). The peak of the current waveform at the moment that the input voltage is 90VDC is ‘ILIM’. Note: 90Volts is used as it is a good estimate of the lowest point of the voltage ripple across any input bulk capacitor of a typical power supply operated at 85VAC, provided

December, 97

AN28 the capacitor 3F/Watt.

is

sized

according

to

the lower the dissipation. However the slight improvement in dissipation by using a bigger capacitor may not be worth the cost in many cases and needs to be evaluated.

Measuring LLK The transformer must be first soldered onto the actual printed circuit board. A short is created by soldering a thick short jumper across the output diode. Another short is placed one across the pins of the (nearest) output capacitor. The inductance across the pins of the primary winding then provides the total in-circuit leakage inductance LLK. Note: The leakage term involves both the transformer leakage as well as the reflected secondary trace inductance. The Designer may wish to refer to Example 3 in AN-26 to understand the mathematics behind how the secondary trace inductance reflects into the primary.



Most of the R-C design curves are plotted upto a maximum ELK (where the curves become almost vertical). It is inadvisable to pick a point too close to the limit as the RCD design will become very sensitive to component tolerances.



The curves are given for possible VOR of 105V, 135V or 150V. The applicable VOR is the value inputted by the Designer into the TOPSwitch-FX Transformer design spreadsheet. It is clear from the curves that if the VOR is high, smaller and cheaper clamp capacitors do not suffice. The clamp dissipation is then much higher for a given ELK. In fact high values of VOR can severely degrade the efficiency of the Power Supply in two ways. Firstly, from the equation for PCLAMP, we can see that the denominator becomes very small if the VOR approaches the maximum permissible clamping levels. In addition, as is clear from Example 3 in AN-26, the secondary side trace inductance reflects into the Primary according to the square of the turns ratio. Since VOR is proportional to turns ratio, the effective leakage LLK to be used in the PCLAMP equation will also increase dramatically for a high VOR, especially for low output voltage power supplies. Combined, these two effects cause a very large increase in the clamp dissipation. Therefore, for most TOPSwitch-FX based designs, the recommended VOR is 135V (or lower).



The curves are given for the three KRP values of 0.3, 0.4 and 0.5. The applicable KRP is the value inputted by the Designer

The Y-AXIS of the R-C selection curves

On the y-axis we have plotted the reciprocal of the time constant ‘RC’. This axis is expressed in kHz. Since, the value of C for any given curve is known we can solve for R. The suggested value of R for a given C expresses the resistance required to ensure a safety margin for the VDS at high line. Selecting the most suitable R-C combination

Many values of C are possible. And for each there is a given R to ensure optimum operation. With a view to selecting which particular R-C combination is the most useful, the following guidelines are provided: 

The R-C selection curves are plotted for three pre-selected capacitors 4.7nF, 10nF and 22nF. The larger the clamp capacitor

RCDClamp Off-line Flyback.doc

7

December, 97

AN28 into the TOPSwitch-FX Transformer design spreadsheet. The chosen KRP corresponds to the lowest point of the voltage ripple (90VDC) across the input bulk capacitor, with the Power Supply operating at full load at 85VAC. As expected, a high KRP causes an increase in the clamp dissipation because of the higher peak currents. A KRP less than 0.3 is not recommended for TOPSwitchFX. Refer to the ‘Initial Current Limit’ specification on the datasheet. Having identified the VOR, the KRP, and a suitable C, the Designer finally knows which of the curves to consult for finding R and estimating the Dissipation. The following step-by-step example will make this clear. A Step-by-step Calculation Example This follows the design steps indicated in the ‘Quick Start’ Box on the first page of this Application Note. 1) By placing shorts across the output diode and the output capacitor respectively, we measure the in-circuit leakage. LLK  20 H. 2) At the rated power, with the oscilloscope triggered at 90VDC on the bulk capacitor, the peak of the Drain Current is measured. I LIM  1.2 Amps 3) Therefore the ELK is 1 20 ELK   6  (1.2) 2 Joules 2 10  14.4 J 4) The value of C tentatively selected is 10nF. 5) Our VOR is 135V and KRP is 0.3. We therefore refer to Fig. 2.

RCDClamp Off-line Flyback.doc

8

6) Drawing a vertical line from the x-axis at an ELK of 14.4 we see that we intersect the curve corresponding to KRP=0.3 and C=10nF at 9.5 on the vertical axis. 7) The estimated Dissipation is about 4.6W. 8) This is considered acceptable. R is now calculated from the y-coordinate (9.5 kHz). The C was 10nF.

1 109  R  C  (9.5 103 ) 9.5 10 4  10.5 k In the practical application, R was 10 k/5W, and C was 10nF/400VDC. The selected diode was an ultrafast 1A/600V device. The VDS under steady operation at full load and 275VAC was found to be less than 600V assuring an adequate safety margin (Current Limit reduction at high line was implemented). Note: The Zener Clamp Dissipation equation predicts that a 200V zener would have given a dissipation of 5.8 Watts. The clamp dissipation has therefore been reduced by 20% by the use of an RCD clamp, under these conditions. Conclusions The RCD clamp can offer an advantage in cost as well as in Efficiency over a 200V zener clamp. But its design and implementation are difficult and this Application Note has attempted to ease the task for the Designer who wishes to exploit the many new features of TOPSwitch-FX devices.

December, 97

AN28 4.7nF 40

VOR=105V Krp=0.5

Krp=0.4

Krp=0.3

10nF

Krp=0.5

Krp=0.4 30

Krp=0.3

20

1/RC (kHz) Krp=0.5

22nF Krp=0.4 10 10W

Krp=0.3

5W

5

10

20

15

E

LK

25

30

35

40

45

(uJ)

Figure 1

RCDClamp Off-line Flyback.doc

9

December, 97

50

AN28

40

VOR=135V

4.7nF

Krp=0.5

10nF Krp=0.5

Krp=0.4

Krp=0.4

Krp=0.3

Krp=0.3

30 Krp=0.5

22nF

Krp=0.4 20 Krp=0.3

1/RC (kHz) 15W

10 10W

5W

5

10

20

15

E

LK

25

30

35

40

45

(uJ)

Figure 2

RCDClamp Off-line Flyback.doc

10

December, 97

50

AN28

40

VOR=150V 10nF 4.7nF

Krp= 0.5 0.4

22nF 0.3

Krp=0.5 0.4 0.3 30 Krp=0.5

Krp=0.4

Krp=0.3 20W 20

1/RC (kHz) 15W

10

10W

5W

5

10

20

15

E

LK

25

30

35

40

45

(uJ)

Figure 3

RCDClamp Off-line Flyback.doc

11

December, 97

50

AN28

VIN R C D

FIGURE 4

RCDClamp Off-line Flyback.doc

12

December, 97

AN28

RCDClamp Off-line Flyback.doc

13

December, 97