Functions / Requirements of Direct-Off-Line SMPS -- FOLDBACK (RE-ENTRANT) OUTPUT CURRENT LIMITING

Home | Articles | Forum | Glossary | Books


Foldback current limiting, sometimes referred to as reentrant current limiting, is similar to constant current limiting, except that as the voltage is reduced as a result of the load resistance moving toward zero, the current is also induced to fall. However, this apparently minor change in the characteristic has such a major impact on the performance that it justifies special attention. To introduce the principle, a linear power supply will be considered.

In linear power supplies, the purpose of foldback current limiting is to prevent damage to the power supply under fault conditions. With foldback limiting, the current is reduced under overload conditions, reducing the power stress on the linear regulator transistors.

Because of the high dissipation that would otherwise occur, some form of foldback current limiting is almost universal in linear power supplies.


FIG. 1 shows a typical reentrant characteristic, as would be developed measured at the output terminals of a foldback-limited power supply.

A purely resistive load will develop a straight load line (for example, the 5-7 load line shown in FIG. 1). A resistive load line has its point of origin at zero, and the current is proportional to voltage.

As a resistive load changes, the straight line (which will start vertically at zero load- i.e., infinite resistance) will swing clockwise around the origin to become horizontal for a short circuit (zero resistance). It should be noted that a straight resistive load line can cross the reentrant characteristic of the power supply at only one point, for example, point P1 in FIG. 1 or 3. Consequently, "lockout" cannot occur with linear resistive loads, even if the shutdown characteristic is reentrant.

In the example shown in FIG. 1, as the load current increases from zero, the voltage initially remains constant at the stabilized 5-V output. However, when the maxi mum limiting current Imax has been reached at P2, any further attempt to increase the load (reduction of load resistance) results in a reduction in both output voltage and cur rent. Hence, under short-circuit conditions, only a small current Isc flows in the output terminals.

FIG. 1 Current overload reentrant shutdown characteristic of fold-back current-limited supply.


In the simple linear regulator shown in FIG. 2a, a typical foldback current limit circuit is shown (in dashed outline). The output parameters are shown in FIG. 1, and the regulator dissipation in FIG. 2b.

This circuit operates as follows: When the main series regulator transistor Q1 is conducting, a voltage proportional to the output current I_load is developed across the current limit resistor R1. This voltage, together with the base-emitter voltage of Q1, is applied to the base of the current limit transistor Q2 through the divider network R2, R3.

Since the base-emitter voltage drop Vbe of Q1 will be approximately the same as that of Q2 at the point of current limit, the voltage across R2 is the same as that across R1 but is level-shifted by +Vbe. At the point of transition into current limit, the current flow in R2 is the same as that in R3 (neglecting small base currents), and Q2 is on the threshold of conduction.

FIG. 2 (a) Foldback current limit circuit. (b) Regulator dissipation with reentrant protection.

Any further increase in the load current at this point will increase the voltage across R1 and hence across R2, and Q2 will be progressively turned on. As Q2 conducts, it diverts the drive current from Q3 away from Q1 into the output load, Q1 starts to turn off, and the output voltage falls. Note: Q3 is a constant-current source.

As the output voltage falls, the voltage across R3 decreases and the current in R3 also decreases, and more current is diverted into the base of Q2. Hence, the current required in R1 to maintain the conduction state of Q2 is also decreased.

Consequently, as the load resistance is reduced, the output voltage and current fall, and the current limit point decreases toward a minimum when the output voltage is zero (output short circuit). At short circuit, the current in R1 is very small, and the voltage across R1 and R2 will also be small.

The short-circuit current is not well defined, as the base current of Q2 depends on its current gain, which will vary between devices; also, the Vbe of Q1 and Q2 are temperature dependent. These variations can be minimized by mounting Q2 on the same heat sink as Q1 and by using relatively low values for R1 and R2 (typically R1 would be of the order of 100 7 in the example shown).

In FIG. 1, it should be noted that a current "foldback" occurs when the current tries to exceed Imax. This characteristic is developed as follows:

If the 5-7 load line is allowed to swing clockwise (resistance being reduced toward zero), the current path shown in FIG. 1 will be traced out. From its initial working cur rent (say 1 A at point P1), the current first rises to its limiting value Imax, then falls toward zero as the load resistance is further reduced. For short-circuit conditions, the current falls to a low value Isc .

Since the header voltage of the linear regulator VH remains relatively constant through out this foldback current limiting action, the power dissipation in the series regulator transistor Q1 will initially increase with increasing currents, as shown in FIG. 2b.

The dissipation is small for the first part of the characteristic, but increases rapidly as the supply moves into current limit. It has a maximum value at a current at which the regulator transistor voltage drop Iload product is a maximum [where (VH Vout ) I_load is maximum] -- in this example, at a current of 2.2 A, where there is a maximum dissipation in Q1 P_max of 6.8 W.

When the load resistance is reduced further (below this critical value), the series regulator dissipation is progressively reduced as a result of the current foldback. It has a minimum value of P(Q1) _ IscVH watts in this example. This results in a dissipation of 1.8 W under short-circuit conditions.

It should be noticed that had the current limit characteristic been a constant-current type (shown by the vertical dashed path B in FIG. 1), the maximum dissipation under short circuit conditions would have been ImaxVH , or 12.8 W in this example. Hence the constant current limit places considerably greater stress on the regulator transistor than the reentrant characteristic in the linear regulator example.

FIG. 3 Overload and start-up characteristics of a foldback, current-limited supply, showing performance for linear and nonlinear load lines.


With the resistive load (the straight-line loads depicted in Figs. 1 and 3), there can only be one stable point of operation, defined by the intersection of the load line for a range of given loads with the power supply characteristic (for example, all points P1).

Hence, the reentrant characteristic shown would be swept out as the load resistance is varied from maximum to zero. This characteristic is swept out without instability or "lockout"; however, this smooth shutdown may not occur with nonlinear loads.

FIG. 3 shows a very nonlinear load line R3 (such as may be encountered with tungsten filament lamps) impressed on the power supply reentrant current limit characteristic.

It should be understood that a tungsten filament lamp has a very low resistance when it is first switched on (because of the low temperature of the filament wire). Consequently, a relatively large current flows at low applied voltages. As the voltage and current increase, the temperature and resistance of the filament increase, and the working point changes to a higher resistance. A nonlinear characteristic is also often found in active semiconductor circuits.

It should be noticed that this nonlinear load line crosses the power supply reentrant current characteristic at three points. Points P1 and P2 are both stable operating points for the power supply. When such a supply-load combination is first switched on, the output voltage is only partially established to point P2, and lockout occurs. (It is interesting to note that if the supply is switched on before the load is applied, it may be expected that the correct working point P1 will be established.) However, point P1 is a stable operating point only for a lamp that was previously working. When the lamp is first switched on, lockout will still occur at point P2, during the lamp power-up phase.

This is caused because the slope resistance of the lamp load line at point P2 is less than the slope of the power supply reentrant characteristic at the same point. Since P2 is a stable point, lockout is maintained, and in this example the lamp would never be fully turned on.

FIG. 4 Nonlinear load line, showing "lockout" and modified characteristics to prevent lockout.

Reentrant lockout may be cured in several ways. The reentrant characteristic of the power supply may be modified to bring it outside the nonlinear load line of the lamp, as shown in plots B and C in FIG. 4. This characteristic now provides only one stable mode of operation at point P1. However, modifying the current limit characteristic means that under short-circuit conditions the current is increased, with a corresponding increase in regulator transistor dissipation. This increase may not be within the design parameters of the power supply. For this reason, one of the more complex current limit circuits may be preferred. These change the shape of the limit characteristic during the turn-on phase, then revert to the normal reentrant shape.

Other methods of curing lockout include modifying the shape of the nonlinear load line of the lamp itself-for example, by introducing a nonlinear resistor in series with the lamp circuit. NTCs (negative coefficient resistors) are particularly suitable, as the resistance of the load will now be high when the lamp is first switched on, and low in the normal operating mode. The NTC characteristic is the inverse of the lamp characteristic, so that the composite characteristic tends to be linear or even overcompensated, as shown in FIG. 4.

However, a slightly higher voltage is now required from the power supply to offset the voltage drop across the NTC.

NTCs are the preferred cure, since they not only cure the "lockout" but also prevent the large inrush current to the lamp which would normally occur when the lamp is switched on. This limiting action can considerably increase the lamp life.

Nonlinear loads come in many forms. In general, any circuit that demands a large inrush current when it is first switched on may be subject to lockout when reentrant current protection is used.


Lockout problems can occur even with linear resistive loads when two or more foldback limited power supplies are connected in series. (This series connection is often used to provide a positive and negative output voltage with respect to a common line.) In some cases series power supplies are used to provide higher output voltages.

FIG. 5a shows a series arrangement of foldback-limited supplies. Here, positive and negative 12-V outputs are provided. The normal resistive loads R1 and R2 would not present a problem on their own, provided that the current is within the reentrant characteristic, as shown by load lines R1 and R2 in FIG. 5b. However, the cross-connected load R3 (which is connected across from the positive to the negative output terminals) can cause lockout depending on the load current magnitude.

FIG. 5b shows the composite characteristic of the two foldback-protected sup plies. The load lines for R1 and R2 start at the origin for each supply and can cross the reentrant characteristics at only one point. However, the cross-connected load R3 can be assumed to have its origin at V+ or V-. Hence, it can provide a composite loading characteristic which is inside or outside of the reentrant area, depending on its value. In the example shown, although the sum of the loads is within the characteristic at point P1, a possible lockout condition occurs at point P2, when the supplies are first switched on. Once again, one cure is to increase the short-circuit current for the two power supplies to a point beyond the composite load line characteristic.

FIG. 5 (a) Bipolar connection with cross-coupled load. (b) Composite characteristic with bipolar load connections.

In FIG. 5a, shunt-connected clamp diodes D1 and D2 must be fitted to prevent one power supply reverse-biasing its complement during the power-up phase. With foldback protection, if a reverse voltage bias is applied to the output terminals of the power supply, the reentrant characteristic is deepened and the current is even lower. This effect is shown in the dashed extension to the reentrant characteristics in FIG. 5b.

In conclusion, it can be seen that there are many possible problems in the application of foldback-limited supplies. Clearly, these problems are best avoided by not using the foldback method if it is not essential.


The previous limitations would also apply to the application of foldback protection in switchmode supplies. However, in switchmode units, the dissipation in the control element is no longer a function of the output voltage and current, and the need for foldback current protection is eliminated.

Consequently, foldback protection should not be specified for switching supplies. It is not necessary for protection of the supply and is prone to serious application problems, such as "lockout." For this reason, constant current limits are preferred in switchmode supplies.

Although the nonlinear reentrant characteristic has nothing to recommend it for switch mode supplies, it is often specified. It is probable that its introduction and continued use stems from the experience with the linear dissipative regulator, where excessive internal dissipation would occur under short-circuit conditions with a constant current limit.

However, this dissipative condition does not occur in switchmode supplies, and since a reentrant characteristic can cause problems for the user, there would seem to be no reason to specify it for switchmode applications. It makes little sense to put extra circuitry into a power unit which only degrades its utility.


1. Explain in simple terms the phenomenon of "lockout" and its cause in foldback current limited supplies.

2. How is it possible to ensure that lockout will not occur with a foldback current limited power supply?

Also see: Our other Switching Power Supply Guide

Top of Page

PREV.   NEXT   Guide Index HOME