Method for controlling a switching regulator and related switching regulator

[0001] An embodiment of the disclosure relates to techniques for generating a controlled voltage and more particularly to the methods for controlling a switching regulator.

[0002] A block diagram of a voltage regulator that supplies a load L through a cable C is depicted in FIG. 1. A control system keeps the voltage generated by the converter at a constant value when changes of the input voltage Vin and/or the load L occur.

[0003] Optionally, a second control system may be present to regulate the current delivered by the converter.

[0004] The two control systems are mutually exclusive: if the current demanded by the load is lower than the current regulation setpoint, the voltage control system will regulate the output voltage and the current control system will be inoperative; contrarily, the current control system will take over and the voltage loop will be inoperative. Voltage control and, when present, current control use a closed-loop negative feedback: the voltage generated by the converter and current through the load, respectively V_OUT and I_OUT, are fed back to the error amplifiers EAV and EAC and they are compared with their references V_REF and I_REF, respectively.

[0005] The input signals V_CV, V_CC to the controller come from the error amplifiers that sense the difference between reference values (V_REF and I_REF) and the feedback signals (V_OUT and I_OUT). Depending on the input signals, the controller generates a PWM signal that drives power switches. Through a transformer, an output rectifier and a filter, energy is transferred from the supply voltage source V_IN to the load L. The diagram shown in FIG. 1 is quite general and may have several possible alternative embodiments.

[0006] Typically, energy is transferred to the load through a cable C. The voltage control loop keeps the voltage Vout regulated but, depending on the output current, the voltage on the load, V_LOAD, will be affected by a voltage drop along the cable, out of the control loop. Thus if a zero load regulation is to be achieved, it may be necessary to compensate the drop along the cable in some way.

[0007] A simple known way of meeting this potential need is illustrated in FIG. 2 and consists in using an additional sensing wire to sense the voltage V_LOAD. In this way a zero load regulation may be achieved, but an additional wire is needed. A three-wire cable is not as common as a two-wire one and may be more expensive.

[0008] Another solution, that avoids the need of additional wires, is to adjust the voltage loop reference (V_REF) by an amount proportional to the average output current, the value of which can be sensed directly even with a remote load. Cable drop compensation (briefly CDC) can be performed if the value of the cable resistance R_cable is known. This solution is depicted in FIG. 3.

[0009] The transfer function of the CDC block is:
V′_REF=V_REF+k_CDC·I_OUT,
where k_CDC is the cable drop compensation gain and V′_REF is the adjusted reference.

[0010] In the circuit of FIG. 1, during voltage regulation, it is:
V_OUT=k_CV·V_REF and V_LOAD=V_OUT−R_cable·I_OUT,
where k_CV is the voltage loop gain, V_OUT is the regulated voltage and V_LOAD is the real voltage on the load.

[0011] With reference to the diagram of the FIG. 3 the output voltage is:
V′_OUT=k_CV·V′_REF=k_CV·(V_REF+k_CDC·I_OUT)=V_OUT+k_CV·k_CDC·I_OUT.

[0012] As the resistance R_cable is known by the application, the k_CDC value is chosen in order to satisfy the condition V_LOAD=V_OUT, hence:

[0013] $k_{\mathrm{CV}}·k_{\mathrm{CDC}}=R_{\mathrm{cable}}\Rightarrow k_{\mathrm{CDC}}=\frac{R_{\mathrm{cable}}}{k_{\mathrm{CV}}}.$

[0014] Typically, the output current is sensed directly.

[0015] A common way of sensing the output current and adjusting the voltage reference proportionally in a non-isolated step-down switching converter is illustrated in FIG. 4 (from the STMicroelectronics AN1061 applications note, all versions of which are incorporated by reference). In particular, by connecting the resistor R_K as shown in FIG. 4, it is possible to adjust the voltage reference value by shifting the ground voltage of the IC by an amount proportional to the current I_LOAD.

[0016] A similar technique applied to an isolated flyback switching converter is shown in FIG. 5 (from the STMicroelectronics TSM1052 datasheet, all versions of which are incorporated by reference). Only the secondary side is shown; V_OUT and I_OUT are sensed and compared against their respective references; the error signal (of the loop in control) is transferred to the primary side via an optocoupler, where it is properly handled.

[0017] A typical isolated flyback configuration using the optocoupler to transfer the output information from secondary side to the primary one is shown in FIG. 6 (from the STMicroelectronics Viper53 datasheet, all versions of which are incorporated by reference).

[0018] There is a special class of low-cost isolated converters, in which output voltage regulation is quite loosely specified and use a simpler approach, according to which there is no sensing element or any reference on the secondary side and, therefore, no specific means for crossing the isolation barrier to transfer the error signal to the primary side, as depicted in FIG. 7 (from the STMicroelectronics Viper53 datasheet, all versions of which are incorporated by reference). In these systems, the voltage drop along the output cable adds to their inherently poor load regulation and can make unacceptable the use of such low-cost systems. In this case, a cable drop compensation circuit would make the difference. However, there is no known technique to compensate the cable resistance for this type of switching converter.

[0019] It has been found that it is possible to use the technique of adjusting the voltage reference even in flyback switching converters that do not have any voltage or current sensing means on the secondary side, and also do not have means for transferring an error signal from the secondary side to the primary side of the converter.

[0020] It has been demonstrated that the average output current delivered by the converter may be accurately estimated using signals available on the primary side, by providing a dedicated circuit block for estimating such a value.

[0021] More precisely, the average output current I_OUT is proportional to the product of Is and the ratio T_ONSEC/T wherein I_S is the secondary peak current, T_ONSEC is the time during which the secondary current is flowing and T is the switching cycle.

[0022] It has been found that signals accurately proportional to the ratio T_ONSEC/T and to I_S can be extracted from the primary side in any switching converter with primary feedback, thus it is not necessary to use dedicated sensors nor means for crossing the isolation barrier from the secondary side to the primary side.

[0023] For example, a signal accurately proportional to the ratio T_ONSEC/T may be produced in different alternative ways:

• • [0024] measuring, with counters or with any other suitable digital means, the time interval T_ONSEC in which the logic control signal that flags the beginning and the end of demagnetization phases is active and the duration T of the switching period; and
  • [0025] calculating the ratio between the above times for producing a signal the level of which represents the ratio T_ONSEC/T.

[0026] As an alternative, a signal proportional to the ratio T_ONSEC/T may be produced by integrating over each switching period the logic control signal that flags the beginning and the end of demagnetization phases.

[0027] Another signal proportional to the ratio (T_ONSEC/T)⁻¹ may be obtained using the charge voltage of a filter capacitor on the primary side of the switching regulator that is discharged during each demagnetization phase by a resistor and is charged by a constant current in the remaining part of each switching period.

[0028] These signals representative of the current delivered to a load are used for estimating the voltage drop on the cable that connects the regulator to the load. Therefore, it is possible to control the effective voltage on the load instead of the voltage generated on the secondary side by the switching regulator.

[0029] Embodiments of the techniques herein described for estimating the output current of a flyback switching regulator without using sensing elements on the secondary side may be used also for other useful purposes.

[0051] Primary and secondary sample current waveforms of a flyback switching converter working in discontinuous mode are depicted in FIG. 8. It will be assumed that its PWM modulator uses a current mode control. The average output current I_OUT is:

[0052] $I_{\mathrm{OUT}}=\frac{I_S}{2}·\frac{T_{\mathrm{ONSEC}}}{T},$
where, I_S is the secondary peak current, T_ONSEC is the time during which the secondary current is flowing, and T is the switching-cycle period.

[0053] By adding a dedicated circuit, able to estimate the ratio T_ONSEC/T, in the current mode IC controller, it is possible to calculate the I_OUT, value by the above formula. This approach may be applied to any current-mode-controlled switching converter with primary feedback.

[0054] In order to better understand the gist of this technique, the functioning of an off-line all-primary-sensing switching regulator, disclosed in U.S. Pat. Nos. 5,729,443 and 6,590,789 (which are incorporated by reference) will be discussed.

[0055] An equivalent high-level circuit scheme of the switching regulator disclosed in U.S. Pat. No. 6,590,789 for regulating the output voltage is reproduced in FIG. 9. An accurate image of the output voltage is obtained by sampling the voltage on the auxiliary winding immediately at the end of transformer's demagnetization phase, as illustrated in the graph of FIG. 10. The switch Q1 is turned on after the end of the demagnetization phase and then turned off by a comparator that monitors the source current of Q1 using a sense resistor R_S.

[0056] An equivalent high level circuit scheme of the switching regulator disclosed in U.S. Pat. No. 5,729,443 for regulating the output current is reproduced in FIG. 11. The switch Q1 is operated by the PWM signal, set by the end of the demagnetization phase of the transformer, and reset by a comparator that monitors the source current of Q1 through the sense resistor R_S.

[0057] The voltage of an auxiliary winding is used by a demagnetization block DEMAG through a protection resistor. The demagnetization block DEMAG generates a logic flag EOD that is high as long as the transformer delivers current to secondary side. Waveforms of the currents in the primary side and in the secondary side of the regulator, of the logic flag EOD, and of the current I_C through the filter capacitor C during a switching period, are shown in FIG. 12.

[0058] The logic flag EOD is used to turn on and off a MOSFET switch Q2 for discharging/charging the filter capacitor C. A resistor R in series with it absorbs a current U_C/R, where U_C is the voltage across the capacitor C. This capacitor C filters the charge current I_REF and the discharge current (I_REF−U_C/R) so that U_C is practically a DC voltage, that is applied to an input of the current mode comparator.

[0059] At steady state, the average current I_C is zero. If T_ONSEC is the time during which the secondary current I_S is flowing, it is:

[0060] $I_{\mathrm{REF}}·\left(T-T_{\mathrm{ONSEC}}\right)+\left(I_{\mathrm{REF}}-\frac{U_C}{R}\right)·T_{\mathrm{ONSEC}}=0,$
which can be simplified in:

[0061] $\begin{array}{cc}{U}_C=R·I_{\mathrm{REF}}·\frac{T}{T_{\mathrm{ONSEC}}}& \left(1\right)\end{array}$

[0062] The voltage U_C is then used to set the peak primary current I_p:

[0063] $I_P=\frac{U_C}{R_S},$
which defines the peak secondary current I_S:

[0064] $\begin{array}{cc}{I}_S=n·I_P=n·\frac{U_C}{R_S}& \left(2\right)\end{array}$

[0065] The average output current I_OUT can be expressed as:

[0066] $\begin{array}{cc}{I}_{\mathrm{OUT}}=\frac{I_S}{2}·\frac{T_{\mathrm{ONSEC}}}{T}& \left(3\right)\end{array}$

[0067] By combining the previous equations, we obtain:

[0068] $I_{\mathrm{OUT}}=\frac{n}{2}·\frac{R·I_{\mathrm{REF}}}{R_S}.$

[0069] Thus it is possible to set the average output current of the switching regulator by fixing the reference current I_REF and the resistances R and R_S.

[0070] It has been found that a signal proportional to the output current can be generated by using signals already available in the primary side of the converter.

[0071] Indeed, combining equations (1) and (3), leads to the following expression:

[0072] $\begin{array}{cc}{U}_C=\frac{R·I_{\mathrm{REF}}}{2}·\frac{I_S}{I_{\mathrm{OUT}}}& \left(4\right)\end{array}$

[0073] Hence the charge voltage of the filter capacitor contains information concerning the average output current, thus it can be used for compensating the voltage drop on the cable that connects a load to a flyback switching regulator.

[0074] Moreover, during the voltage regulation, the voltage control loop signal establishes the peak primary current I_p:

[0075] $\begin{array}{cc}{I}_P=\frac{V_{\mathrm{CV}}}{R_S}& \left(5\right)\end{array}$
wherein V_CV is the voltage generated by the error amplifier EAV (in the circuit of FIG. 1) proportional to the difference between the reference voltage V_REF and the output voltage V_OUT generated by the controller.

[0076] Therefore, by combining the equations (4) and (5) it results:

[0077] $U_C=\frac{n}{2}·\frac{R·I_{\mathrm{REF}}}{R_S}·\frac{V_{\mathrm{CV}}}{I_{\mathrm{OUT}}}$

[0078] In the above formula all the signals are known except for the I_OUT value.

[0079] In the IC controller is inserted a dedicated CDC block for performing the division between the signals V_CV and U_C in order to obtain a signal proportional to the output current:

[0080] $\begin{array}{cc}\frac{V_{\mathrm{CV}}}{U_C}=\frac{2}{n}·\frac{R_S}{R·I_{\mathrm{REF}}}·I_{\mathrm{OUT}}& \left(6\right)\end{array}$

[0081] In an embodiment, the CDC block is analog, as depicted in FIG. 13, and comprises an analog divider the output of which is multiplied by a constant k, a filter and an analog subtractor of the output of the filter and the reference voltage V_REF.

[0082] As an alternative, the CDC block could be digital, converting the signals V_CV and U_C in digital form, carrying out the division, subtracting the result from the voltage value V_REF, and converting the result back into an analog signal.

[0083] The next step is to adjust the voltage reference V_REF by an amount depending on the output current, as explained previously. In fact, the CDC block is designed to implement the following transfer function:

[0084] $V_{\mathrm{REF}}^{\prime }=V_{\mathrm{REF}}-k·\frac{2}{n}·\frac{R_S}{R·I_{\mathrm{REF}}}·I_{\mathrm{OUT}}.$

[0085] The CDC block, during the output voltage regulation, introduces a positive feedback that may compromise the stability of the primary loop. For this reason a low-pass filter is preferably added, as shown in FIG. 13.

[0086] Looking at FIG. 13 it is possible to notice the analog divider, the output signal of which is multiplied by a constant k, the filter and the analog subtractor.

[0087] FIG. 14 shows the architecture of an embodiment of a voltage mode converter, that includes a CDC block in the primary loop for adjusting the voltage reference value (V_REF) by an amount proportional to the output current. The new voltage loop reference is V_REF′. This allows to compensate the voltage drop along the output cable and, ideally, to achieve a zero load regulation.

[0088] This technique may be applied even by modifying the feedback voltage on the capacitor C* instead of directly acting on V_REF. A sample embodiment of this type is shown in FIG. 15, where the CDC block sinks a current proportional to the output current from the feedback resistor divider in order to modify the sampled value:

[0089] $I_{\mathrm{CDC}}=k·\frac{2}{n}·\frac{R_S}{R·I_{\mathrm{REF}}}·I_{\mathrm{OUT}}.$

[0090] Another way to modify the voltage feedback signal value is to generate a voltage proportional to the output current:

[0091] $V_{\mathrm{CDC}}=V_{\mathrm{REF}}-k·\frac{2}{n}·\frac{R_S}{R·I_{\mathrm{REF}}}·I_{\mathrm{OUT}}$
and to connect a resistor R_CDC as shown in the FIG. 16. The resistor R_CDC is an external component which gives the user the possibility to set the CDC gain depending on the application. Its value is calculated by the following equation:

[0092] $R_{\mathrm{CDC}}=k·\frac{2}{n}·\frac{N_{\mathrm{OUT}}}{N_{\mathrm{AUX}}}·\frac{R_1}{R_{\mathrm{cable}}}·\frac{R_S}{R·I_{\mathrm{REF}}},$
where, n is the ratio between primary and secondary windings, N_OUT is the number of the windings on the secondary, N_AUX is the number of the windings on the auxiliary, R_cable is the cable resistance and R_S is the sensing resistor connected to the power MOSFET source. The use of that resistor is a possible way to set the CDC gain depending on the application. In fact, applying the previous embodiments, without R_CDC, the same objective can be reached by trimming the constant k value.

[0093] A signal proportional to the ratio T_ONSEC/T may be generated by exploiting the logic control signal EOD that flags the beginning and the end of magnetization phases, for example using the embodiment of the circuit depicted in FIG. 17. Two pulse counters COUNTER generate digital signals corresponding to the duration of the time intervals T_ONSI and T−T_ONSEC by counting clock pulses while the signal EOD and the inverted replica thereof are active, respectively, then a calculation block DIGITAL CALCULATOR generates a digital signal that represents the ratio T_ONSEC/T, that is converted in a corresponding analog signal Vratio by a digital-to-analog converter DAC.

[0094] If the CDC block can be input with digital signals, then the converter DAC is not necessary.

[0095] According to an alternative embodiment, a signal proportional to the ratio T_ONSEC/T may be generated by the circuit of FIG. 18, that uses three monostable flip-flops for switching three capacitors C, C1 and C2. In correspondence of the leading edge of the signal EOD, the charge voltage of the capacitor C is sampled and held on the capacitor C1, and the capacitor C is discharged (signal RESET). The capacitor C is charged again by the current generator IREF and its charge voltage is sampled and held on the capacitor C2 when the signal EOD switches low (that is at the end of each demagnetization phase). Therefore, the charge voltages VC1 and VC2 of the capacitors C1 and C2 represent the duration of a period and of the magnetization phase, respectively:

[0096] $V_{C1}=\frac{I_{\mathrm{REF}}}{C}·T,V_{C2}=\frac{I_{\mathrm{REF}}}{C}·T_{\mathrm{ONSEC}}$

[0097] A divider generates the signal Vratio as the ratio V_C2/V_C1.

[0098] The signal RESET used for discharging the capacitor C is substantially a delayed replica of the pulse T, such to zero the charge voltage of the capacitor C substantially immediately after it has been held on the capacitor C1.

[0099] According to an alternative embodiment, the voltage Vratio may be generated by integrating the signal EOD over a switching period T, as schematically depicted in FIG. 19.

[0100] A CDC block suitable for using the voltage Vratio for adjusting the reference voltage VREF′ is depicted in FIG. 20. This CDC block is similar to that depicted in FIG. 13, but it has an input multiplier instead of an input divider.

[0101] An embodiment of a switching regulator that employs the CDC block of FIG. 20 and a circuit for generating a voltage Vratio proportional to the ratio T_ONSEC/T, such as the circuits of FIGS. 17 to 19, is shown in FIG. 21. The functioning of this switching regulator is evident in view of the description made referring to FIGS. 14 to 16. Furthermore, some to all of the components of the switching regulator of FIG. 21 may be disposed on an Integrated Circuit (IC) die, and the regulated output voltage V_OUT may provide power to a circuit, such as a controller processor, that is disposed on the same die or on a different die.

[0102] Naturally, in order to satisfy local and specific requirements, a person skilled in the art may apply to the solution described above many modifications and alterations. Particularly, although the present subject matter has been described with a certain degree of particularity with reference to described embodiment(s) thereof, it should be understood that various omissions, substitutions and changes in the form and details as well as other embodiments are possible. Moreover, it is expressly intended that specific elements and/or method steps described in connection with any disclosed embodiment of the disclosure may be incorporated in any other embodiment as a general matter of design choice.