Cmos latch-up recovery circuit

This invention relates in general to power supply circuits, and more particularly to an overcurrent shut down circuit for momentarily shutting off a switching regulator power supply in the event of CMOS latch-up or other high current drawing fault conditions in circuits connected to the supply.

In the event excessively large input signals are applied to an amplifier, active devices (eg. transistors) can be driven into saturation or cut-off conditions. Under saturation conditions, local feedback loops may be set up due to parasitic transistors that are inadvertently established between p and n-type regions of an active device and the substrate. This local positive feedback results in what is known as latch-up.

Thus, in the event a large input signal is applied to an active device (due to static shock, etc.,) simultaneously with a power supply delivering current to the device, latch-up may occur. In general, recovery from latch-up is only possible by manually turning off or unplugging the power supply for a short period of time.

Considerable research has been conducted in the field of latch-up prevention. U.S. patent 4,209,713 issued June 24, 1980,to Tokyo Shibaura Electric Company Limited describes a circuit for eliminating the effects of parasitic conduction. A noise absorption means, typically in the form of a resistor, is connected between a source electrode of a CMOS circuit and a positive terminal of an external power source.

However, in a CMOS integrated circuit, resistors of appropriate values are difficult to fabricate. In addition, resistors restrict the full power supply voltage from being applied to the CMOS devices in the circuit.

Many circuits have been developed for - 1 1287103 controlling the timing of application of high and low voltage sources to a CMOS circuit in an effort to prevent the establishment of parasitic transistors leading to latch-up. For instance, U.S. patent 4,441,035, dated April 3rd, 1984 and issued to Mitel Corporation discloses circuitry for controlling the timing of the application of high and low, similar polarity DC power supplies to a single CMOS circuit. According to the Mitel circuit, the application of the lower voltage supply is prevented until the higher voltage supply has been applied in order to substantially eliminate the formation of parasitic transistors .

U.S. patent 4,353,105 dated October 5th, 1982 and issued to National Semiconductor Corporation describes a protection circuit for bulk silicon CMOS circuits which detects latch-up, and in response starves the CMOS circuit of power for a predetermined length of time and re-enables normal circuit operation once the latch-up condition has been eliminated.

The devices described in the Mitel and National Semiconductor Corporation patents are typically disposed within the CMOS circuitry itself, requiring complex and expensive structure and fabrication techniques. In addition, the National Semiconductor protection circuit provides a control signal to the CMOS circuit which places the CMOS circuit in a low current stand-by mode. Accordingly, the CMOS circuit includes additional circuitry for interpreting the control signal, typically contributing to circuit complexity and expensive fabrication.

Numerous power supply circuits have been designed for providing correct application sequencing of high and low voltage sources to an active device. Such circuits are described for example in U.K.

patents 1,331,962 published the 26th of September 1973 of Matsushita Electric Industrial Co., Ltd., and 1,423,149 published the 28th of January 1976 of Philips Electronic and Associated Industries Limited. These prior art power supply circuits utilize complex timing and sequencing circuitry in an effort to prevent CMOS latch-up and other high current drawing fault conditions yet provide no means for recovering from such conditions in the event that they occur.

In particular, according to the Philips patent, a power supply is provided with circuit means which in the event of operation in a fault condition, reduces the output voltage by approximately half its nominal value, but not to zero volts. Thus, in the event of latch-up, the CMOS circuits would remain latched-up in spite of the reduced voltage.

Similarly, according to Matsushita patent, a power supply system is described wherein circuitry is provided for interrupting the supply of power in the event of an excessively large input signal or in the event of a short circuited load. This well known over voltage protection technique suffers from the disadvantage of requiring a switch for restoring thyristor circuitry to its original state to reactivate the power supply, since the thyristor circuitry is incapable of self-restoration even when the overcurrent disappears. Thus, the Matsushita patent suffers from the disadvantage of requiring manual resetting of the power supply.

According to the present invention, an overcurrent shut down circuit is included in a power supply, which monitors input current, and in the event the current exceeds a predetermined threshold level, shuts off the power supply for a predetermined length of time after which the supply is automatically reset. The shut down circuit is disposed within the power supply, thereby overcoming the disadvantages

- 3 1287103 associated with prior art complex and expensive CMOS protection circuits wherein the CMOS circuits themselves were required to include complex on board circuitry.

The circuit according to the present invention actually provides for recovery of the power supply in the event of a fault condition, and does not merely attempt to prevent such a condition by means of complex timing and delay circuitry as used in prior art sequencing power supply circuits.

An embodiment of the invention is a switching regulator including an input circuit for receiving an input signal, apparatus for pulse width modulating the input signal thereby generating a modulated input signal, apparatus for coupling the modulated input signal to an output circuit, thereby forming an output signal having a D.C. voltage level proportional to the duty cycle of the modulated input signal, apparatus for monitoring the output signal and generating an error signal for application to a D.C. feedback circuit in response thereto, and apparatus for receiving the error signal from the feedback circuit and varying the duty cycle of the modulated input signal in proportion to variations in the D.C. level of the error signal; an over-current shut down and recovery circuit, comprised of apparatus for detecting the average D.C. level of the modulated input signal and generating an average D.C. input signal in response thereto, apparatus for comparing the average D.C. input signal with a predetermined threshold signal and connecting the feedback circuit to ground potential in the event the average D.C. input signal is greater than the threshold signal, whereby the duty cycle of the modulated input signal is reduced to zero thereby ι

21 22 reducing the D.C. voltage level of the output signal to zero volts, and time delay apparatus for removing the feedback circuit from ground potential after a predetermined delay time, thereby reapplying the error signal to the apparatus for varying the duty cycle of the modulated input signal, and restoring the D.C. voltage level of the output signal to a normal level.

A better understanding of the invention will be obtained with reference to the detailed description below in conjunction with the following, in which;

Figure 1 is a schematic diagram of an overcurrent shut down circuit for use in a switching power supply regulator according to the present invention.

With reference to Figure 1, an unregulated input voltage V_IN is applied to one lead (TIP) of a balanced lead pair TIP and RING, commonly found in telephone systems such as PABXs, the RING lead being connected to ground.

A controller supply 1 receives the unregulated input voltage (typically in the vicinity of 25-60 volts) and generates an output voltage +V, in the range of from 4 to 6 volts. Control supply 1 is in the form of a series-pass regulator for providing the reference voltage +V in a well known manner.

An input current I_IN carried by the TIP lead passes through a primary winding 3A of a transformer 3 and is pulse width modulated via FET power transistor 5, under control of pulse width modulating circuitry including a triangle wave oscillator 7 and comparator 9, as described in greater detail below.

During each cycle that transistor 5 is

Ï5

4a conducting, input current transformer 3 stores energy received directly from the balanced TIP and RING leads. During each cycle that transistor 5 is not conducting, the primary coil 3A releases the stored energy to a secondary coil 3B according to well known principles of electromagnetic coupling. In response, a voltage develops across coil 3B which is rectified via a diode 11.

The rectified voltage output from diode 11 is filtered via a capacitor 15 (which in the successful prototype was 100 Zy Farad), connected between the output of diode 11 and isolated ground.

The resulting filtered output voltage signal Vq_UT is applied to an output terminal 17 for application to external circuitry, such as CMOS circuits, etc.

Voltage-series negative feedback of the output voltage signal V_OUT Provided for regulating the power supply in a well known manner.

In particular, a triangle wave oscillator 7 generates a triangle wave signal (which in a successful prototype was at 28 kHz frequency) for application to an inverting input of comparator amplifier 9. A representation of the output voltage signal V_Oy_T, denoted as ERROR, is transmitted via optical isolation circuitry including photodiode 21 and photo-transistor 23, for application to a non-inverting input of comparator 9. The D.C. level of the ERROR signal is compared in comparator 9 to the generated triangle wave signal such that comparator 9 generates a pulse control signal for application to transistor 5, having a duty cycle which varies linearly with the D.C. level of the ERROR signal.

The open collector output of comparator 9 is connected via pull-up resistor 25 to the reference voltage +V, and also directly to a gate terminal of FET transistor 5 for causing pulse width modulation of the input current Ijj$ through the source-drain circuit transistor 5, as described above.

In operation, once the D.C. level of the output voltage signal Vqut reaches a desired output level (for example +5 volts) precision comparator 31 is activated via detection of a predetermined voltage (eg. 2.5 volts) appearing at the node connecting voltage divider resistors 27 and 29, which results in current flowing to isolated ground through photodiode 21, resistor 32, and comparator 31. Photodiode 21 and photo-transistor 23 are preferably configured as an opto-isolator circuit according to well known techniques. Photo-transistor 23 operates as a common-emitter circuit for generating the aforementioned ERROR signal representing the output voltage signal Vqut · Thus, as the current drawn through photodiode 21 decreases, the D.C. level of the ERROR signal carried by the collector terminal of photo-transistor 23 increases. This, in turn causes the duty cycle of the pulse width modulated control signal from amplifier 9 to increase, resulting in more input current flowing through primary coil 3A via transistor 5, thereby delivering greater power to the secondary coil 3B and consequently to the output terminal 17.

The ERROR signal from transistor 23 is voltage limited via resistors 33 and resistor 35, and filtered via capacitor 37 connected to ground. This provides for duty cycle limiting of comparator 9 in the event of power up/down and low line voltage conditions .

The supply is protected from excessive load currents, typically accidental CMOS latch-up, by an overcurrent shut down circuit for monitoring the average D.C. input current Ij_N flowing through primary coil 3A and power transistor 5, and interrupting the power supply output in the event of excess current drawn for a predetermined length of time after which the supply is reactivated.

The input current Ijn translated to a voltage across a sensing resistor 41 (which in a successful prototype was approximately 10 ohms), and subsequently filtered to an average D.C. level via resistor 43 connected to capacitor 45, and the average D.C. signal is applied to an inverting input of comparator 39. A non-inverting input of comparator 39 is connected via a resistor 40 to a predetermined threshold voltage V2 for establishing a threshold level for the input current Ijjj above which the power supply is interrupted.

The output of comparator 39 is connected via a pull up resistor 47 to the reference voltage +V and to ground via a timing circuit comprised of resistor 49 and capacitor 51. The node connecting the output of comparator 39, resistors 47 and 49, and capacitor 51 is also connected to an non-inverting input of a further comparator 53, having an inverting input thereof connected to a threshold voltage source VI, (eg. 1.5 volts).

In the event the average D.C. input current Ijjq exceeds the predetermined threshold voltage, the output of comparator 39 goes to a low voltage level causing the output voltage on comparator 53 to go to ground level, thereby grounding the D.C. ERROR signal applied to the non-inverting input of comparator 9.

The D.C. ERROR signal is grounded for a predetermined length of time governed by the time constant of resistor 49 and capacitor 51, which in a successful prototype was approximately 50 milliseconds, allowing sufficient time for the CMOS circuitry to recover from latch-up.

In response to grounding of the D.C. ERROR signal, the output of comparator 9 goes to a logic low level, thereby shutting off transistor 5 and interrupting the power supply.

- 7 1287103

Once the current in the primary coil 3A stops flowing, capacitor 51 begins recharging and the power supply circuit is restored to normal operation within the predetermined time limit (50 milliseconds).

The output of comparator 53 is an open collector such that during normal operation, the overcurrent shut down circuitry appears as a high impedance or open circuit to the non-inverting input of comparator 9.

In summary, a switch mode fly back regulator power supply is disclosed incorporating an overcurrent shut down or latch-up recovery circuit for providing supply cycling in the event of CMOS latch-up or accidental power supply fault conditions, such as short circuits. By incorporating the recovery circuitry within the power supply itself, expensive and complicated prior art CMOS protection circuits are avoided.

A person understanding the present invention may conceive of other embodiments thereof. All such modifications are believed to be within the. sphere )and scope of the present invention as defined by'tiie claims appended hereto.