1. Field of the Invention
The present invention refers to a control device for a switching voltage regulator.
2. Description of the Related Art
Switching voltage regulators are known in the prior art. A switching voltage regulator of the buck/boost type is shown in
The triangular wave generator 41 determines the frequency at which the system works. If we indicate with duty cycle D the percentage of period in which the high side of a half-bridge is on, the regulated output voltage Vo is related to the input voltage by the relation Vo/Vi=Di/Do where Di is the duty cycle of the half-bridge 1 and Do is the duty cycle of the half-bridge 2.
The average current in the inductance II is given by II=Iload/Do where Iload is the current on the load and the duty cycles Do and Di are obtained by comparing the integrated error e1 with the triangular wave coming from the generator 41; the latter component provides a triangular wave proportional to the input voltage Vi.
A simple way to drive the half-bridges 1 and 2 is to make Do=1−Di, in other words to drive the half-bridge 2 with a signal complementary to the half-bridge 1. This gives the following relations:
In this case there is considerable loss on the series resistance of the inductance L which is proportional to the average current flowing through it. Maximizing the duty cycle of the half-bridge 2 would allow one to reduce this current. The optimum solution is obtained by letting Do=1 during the buck state and letting Di=1 during the boost state. This situation also makes it possible to halve the switching losses of the switches of the half-bridges 1 and 2.
While running during the buck state we have Vo=Di*Vi; if we let Vh be the amplitude of the triangular wave we have Di=e1/Vh where e1 is the integrated error in input to the PWM block 40. If the amplitude Vh is a function of the input voltage Vi, we have Vh=Vi/K and Vo=K*e1. During the boost state we have Vo=Vi/Do=Vh*Vi/e1=Vi2/(K*e1), therefore the output voltage Vo varies instantaneously with every variation of the square of the input voltage Vi.
The PWM block 40 transforms the integrated error e1 into a pair of duty cycles Di, Do to drive the half-bridges 1 and 2. The duty cycle Di is generally produced by comparing the integrated error e1 directly with a saw-tooth wave. The duty cycle Do instead is constructed by comparing a signal e2, which is a function of the integrated error e1, with the triangular wave produced by the generator 41. The function from which this signal is generated depends on the type of driving chosen. In the case of the optimum efficiency condition it is very important that when the signal e1 exceeds the amplitude of the triangular wave, the signal e2 starts to cross it. There should be only one point at which the duty cycles D1, Do are at 100%; if this does not happen a case may arise in which for a certain range of the ratio Vo/Vi the half-bridges 1 and 2 either switch simultaneously, reducing the efficiency of the regulator, or neither one of the half-bridges switches and this could cause undesired oscillations in the output voltage Vo.
and which serves to produce the voltage Vo. To make the system linear while running we let Vo/Vi=Vi/(K*e1) and thus obtain
In the buck state the transfer function Fe=K/Vi; thus the voltages Vi and Vo do not enter any block of the diagram and therefore the gain of the loop is independent of the input and output values.
In the boost state the function Fe is no longer linear and can be expressed as follows:
Therefore the gain of the loop varies with the square of the ratio Vo/Vi and is greater than the gain in the buck state. The double pole of the filter LC is shifted to lower frequencies. The increase in the gain and the shift to the lower frequencies of the double pole are factors that contribute to worsening the stability of the system. This translates in practical terms as having to create compensation networks that take these variations into account. In particular in a compensation system with a dominant pole it is necessary to ensure the stability in the worst conditions or have a high output voltage and low input voltage which however makes the system over-compensated in the other conditions and therefore slower.
One embodiment of the present invention provides a control device for a switching regulator that can overcome the disadvantages explained above.
One embodiment of the present invention is a switching regulator control device, said regulator comprising at least a first and second half-bridge connected to each other and each comprising at least one switch, said first half-bridge being located between an input voltage to the regulator and a reference voltage and said second half-bridge being located between an output voltage from the regulator and said reference voltage, said regulator comprising detecting means suitable for detecting the error between a voltage representative of the output voltage and another reference voltage and integration means suitable for integrating said error, the purpose of said control device being to provide a first and second duty cycle to drive the switches belonging to said first and said second half-bridges, said control device having said regulator input voltage in input, wherein said control device is suitable for providing a first duty cycle proportional to the value of the integrated error divided by the value of the regulator input voltage and a second duty cycle proportional to the value of the regulator input voltage divided by the value of the integrated error.
The characteristics and advantages of the present invention will appear evident from the following detailed description of an embodiment thereof illustrated as non-limiting example in the enclosed drawings, in which:
The control device 100 comprises a triangular wave generator 101 suitable for producing a triangular wave TR of variable amplitude as a function of the integrated error e1, as shown in
if e1<Vi/K then Vh=Vi/K
if e1>Vi/K then Vh=e1.
The control device 100 comprises comparison circuitry 102 suitable for providing the duty cycles Di and Do. The duty cycle Di of the half-bridge 1 is normally obtained by comparing the integrated error e1 with the triangular wave TR while the duty cycle Do of the half-bridge 2 is obtained by comparing Vi/K with the triangular wave TR; we have:
In the buck state things remain substantially identical to the system described previously. Do will still be at 100% because Vi/K is greater than the triangular wave for reasons of construction while Di will increase as the integrated error increases. In the boost state Di will be constantly at 100% because the integrated error is greater than the triangular wave for reasons of construction. Do instead will decrease as the integrated error increases.
In a system of this type there is linear proportionality between the value of the integrated error and the regulated output voltage since Vo=k·e1.
Applying this driving implementation of the half-bridges 1 and 2 the problems arising from possible errors introduced by the PWM block of the regulator in
In this way in running condition the error will assume a value around that of Vo/k both in the buck and boost states. A variation in the voltage Vi does not affect the integrated error e1 and consequently the output. This system therefore uses a feed forward on the feed in every buck or boost condition.
Substituting Fe=K/Vi we have, in the block diagram of
In some applications it is necessary to limit the duty cycle of the output half-bridge because if this should reach 0% for some reason (high side still off) the control loop could open consequently losing control of the output voltage. A collateral advantage of this application is that the limitation of the duty cycle takes place automatically. In fact to have Do=0% it would be necessary to have Vi zero or the error e1 infinite which are impossible conditions.
The control device 100 can be implemented as shown in
The comparators Ci and Co have non-inverting input terminals connected respectively to the error e1 and the voltage Vi/K and provide the duty cycles Di and Do comparing the triangular wave TR with the error e1 through Ci and with Vi/K through Co.
To reduce further the mismatches between the two comparators Co and Ci it is possible to use a single circuit comparator Cun comprising pairs of NMOS and PMOS transistors, as shown in
From the foregoing it will be appreciated that, although specific embodiments of the invention have been described herein for purposes of illustration, various modifications may be made without deviating from the spirit and scope of the invention. Accordingly, the invention is not limited except as by the appended claims.
Number | Date | Country | Kind |
---|---|---|---|
06425201 | Mar 2006 | EP | regional |
Number | Name | Date | Kind |
---|---|---|---|
5402060 | Erisman | Mar 1995 | A |
5982156 | Weimer et al. | Nov 1999 | A |
6989660 | Mauthe | Jan 2006 | B2 |
20030052654 | Kanakubo et al. | Mar 2003 | A1 |
20050007089 | Niiyama et al. | Jan 2005 | A1 |
Number | Date | Country |
---|---|---|
1361651 | Nov 2003 | EP |
Number | Date | Country | |
---|---|---|---|
20070236977 A1 | Oct 2007 | US |