1. Field of the Invention
This invention relates to the field of ESD protection circuits for switching power converters.
2. Description of the Related Art
Common to all switching power converters is a switching element which, when driven to turn on, conducts a current to a “switching” node (SW). For several common configurations, a high-side switching element is connected between a voltage source (power supply) and the SW node, and a low-side switching element is connected between the SW node and circuit common or ground; during normal operation, the high- and low-side switching elements are alternately turned on and off as needed to charge and discharge an output inductor and capacitor and thereby provide a regulated output voltage.
All electronic circuits are susceptible to electro-static discharge (ESD) which can damage the circuit's components; this includes the switching elements of a switching power converter. This is particularly true when the low-side switching element is an n-channel MOSFET (NMOS FET), and as such, measures are often taken to protect the NMOS FET from being damaged by ESD. One way in which the NMOS FET can be protected from ESD is to design it with a larger-than-normal Drain-Contact-to-Gate-Spacing (DCGS). While this method works well for NMOS FETs having a channel width of a few thousands μm or less, it becomes impractical for switching power converters that utilize NMOS FETs having much greater channel widths—e.g., >10,000 μm. In this case, increasing the DCGS would require increasing the silicon die size to an economically uncompetitive level, and the FET's drain-source on-resistance (Rdson) would become unacceptably high.
With a large NMOS FET, “active” ESD protection—i.e., circuitry which detects ESD events and triggers an appropriate protective response—is often employed. One common method is to use an active ESD protection circuit to force the NMOS FET to turn on, so that an ESD discharge path is provided between the SW node and ground by the NMOS FET's channel. However, making such a circuit stable and reliable can be a daunting task, for several reasons.
Such circuitry is typically connected directly to the SW node, to enable an ESD event to be detected; this arrangement is referred to as ‘direct sensing’. When a very large NMOS FET is forced to turn on by the active ESD circuit, the SW node will be pulled to ground potential nearly instantaneously, since the NMOS FET's Rdson is usually on the order of milliohms. This results in the ESD circuit losing its power, which causes the NMOS FET to turn off. However, if the ESD event has not finished when this occurs, the remainder of the ESD charge will force the voltage on the SW node to shoot up again, which triggers another cycle as described above. Also, if the active ESD circuit does not respond quickly enough, or does not retrigger following a previous trigger event for some reason, the NMOS FET can still be damaged. This results in unreliable ESD protection or unstable oscillatory behavior.
A direct sensing arrangement can also cause the ESD protection circuit to falsely trigger during normal operation. This can occur because very fast and very large switching spikes—comparable to those associated with ESD events—are constantly present on the SW node.
As noted above, a common protective method is to force the NMOS FET to turn on. This is typically achieved by having the ESD protection circuit generate an active-high ‘trigger signal’ when an ESD event is detected, which is directly applied to the gate of the NMOS FET; this arrangement is referred to as ‘direct coupling’. When so arranged, the ESD trigger signal and the control signal which operates the NMOS FET under normal operating conditions will physically share the same node. However, during an ESD event, the state of the normal operations signal is unknown, and it is possible that the two signals will conflict and render the ESD protection unreliable.
Another problem can occur if the low-side switching element is turned on to provide a conductive path between the SW node and ground during an ESD event, without regard to the status of the high-side switching element. If the high-side switching element happens to be on while the low-side switching element is on, a ‘shoot-through’ condition occurs which can cause severe damage to the switching elements of the power converter, especially when false triggering occurs during normal operations.
An ESD protection circuit for a switching power converter is presented which overcomes the problems noted above.
The present protection circuit is suitable for use with a switching power converter which includes a high-side switching element connected between a voltage source (power supply) node and the switching node, and a low-side switching element connected between the switching node and a common node (ground). The circuit includes an ESD sense node—preferably the converter's power supply node—and a current conduction path which couples an ESD event that occurs on the switching node to the ESD sense node. When the high-side switching element is a PMOS FET, the conduction path can be provided, for example, by the body diode of the PMOS transistor. Alternatively, if the high-side switching element does not inherently provide such a conduction path, a dedicated device such as a PMOS FET, an isolated NMOS FET or a forward diode connected between the SW node and the ESD sense node can be used.
The ESD protection circuit also includes an ESD sensing circuit, coupled to the ESD sense node and arranged to generate an output signal when an ESD event is sensed on the ESD sense node. The protection circuit employs a logic gate, which is arranged to keep the high-side switching element off when the output of the ESD sensing circuit indicates that an ESD event has been detected, and another logic gate to cause the low-side switching element to turn on when an ESD event is sensed such that the low-side switching element provides a conductive discharge path between the switching node and common node.
Further features and advantages of the invention will be apparent to those skilled in the art from the following detailed description, taken together with the accompanying drawings.
a is a schematic diagram illustrating an embodiment of an ESD protection circuit per the present invention, which employs an active-high trigger signal when an ESD event is sensed.
b is a schematic diagram illustrating another embodiment of an ESD protection circuit per the present invention, which employs an active-low trigger signal when an ESD event is sensed.
Schematic diagrams illustrating the basic principles of an ESD protection circuit per the present invention are shown in
Note that a complete switching power converter would include a number of elements not shown in
The present ESD protection circuit requires an ESD sense node 14 which is different from the SW node, and a current conduction path 16 which couples an ESD event that occurs on the SW node to the ESD sense node. An ESD sensing circuit 18 is coupled to the ESD sense node and arranged to generate a trigger signal of finite duration when an ESD event is sensed on the ESD sense node; the trigger signal is inactive during normal operation. For the example shown in
In
a illustrates a generalized implementation scheme for an active-high trigger signal 20, in which logic gates 22 and 24 are m-input OR gates (where m=n+1 and n=1, 2, 3 . . . ). Logic gate 22 receives trigger signal 20 and normal operation control signals S11, S12 . . . , S1n at respective inputs, and logic gate 24 receives trigger signal 20 and control signals S21, S22 . . . , S2n at respective inputs. In practice, each of gates 22 and 24 can be an actual m-input OR gate or a Boolean equivalent. Gate 22 is arranged such that its output goes high and turns off high-side switching element 10 when trigger signal 20 indicates the sensing of an ESD event; when trigger signal 20 is inactive, gate 22 allows normal operation control signals S11, S12 . . . , S1n to propagate to high-side switching element 10 without alteration—i.e., without affecting the polarity of the control signals. Similarly, gate 24 is arranged such that its output goes high and turns on low-side switching element 12 when trigger signal 20 indicates the sensing of an ESD event; when trigger signal 20 is inactive, gate 24 allows normal operation control signals S21, S22 . . . , S2n to propagate without alteration to low-side switching element 12.
b illustrates a generalized implementation scheme for an active-low trigger signal 52, in which logic gates 54 and 56 are m-input NAND gates (where m=n+1 and n=1, 2, 3 . . . ). Here, logic gate 54 receives trigger signal 52 and normal operation control signals S11, S12 . . . , S1n at respective inputs and logic gate 56 receives trigger signal 52 and control signals S21, S22 . . . , S2n at respective inputs. In practice, each of gates 54 and 56 can be an actual m-input NAND gate or a Boolean equivalent. Gate 54 is arranged such that its output goes high and turns off high-side switching element 10 when trigger signal 52 indicates the sensing of an ESD event; when trigger signal 52 is inactive, gate 54 allows normal operation control signals S11, S12 . . . , S1n to propagate without alteration to high-side switching element 10. Similarly, gate 56 is arranged such that its output goes high and turns on low-side switching element 12 when trigger signal 52 indicates the sensing of an ESD event; when trigger signal 52 is inactive, gate 56 allows normal operation control signals S21, S22 . . . , S2n to propagate without alteration to low-side switching element 12.
Providing an ESD sense node 14 which is different from the SW node, and coupling the ESD sensing circuit to that node to detect ESD events, is referred to herein as ‘indirect sensing’. This arrangement serves to avoid the problems associated with the ‘direct sensing’ approach described above.
The ESD sense node is preferably the power supply that is associated with the SW node—i.e., supply voltage node V+—as shown in
Switching elements 10 and 12 can be implemented with, for example, PNP and NPN bipolar transistors or, as shown in
The use of logic gates 22, 24 and 54, 56 enables the ESD trigger signal to be indirectly coupled to the switching elements; this approach is referred to herein as ‘indirect coupling’. In
The gates and control signals are arranged such that, when trigger signal 20, 52 indicates the sensing of an ESD event, logic gate 22, 54 respectively keeps switching element 10 off and logic gate 24, 56 respectively causes switching element 12 to turn on and thereby provide a conductive discharge path between the SW and common nodes. The gates and control signals must also be arranged such that the control signals operate switching elements 10 and 12 in normal fashion under normal operating conditions.
Keeping switching element 10 off while switching element 12 is forced on during an ESD event provides an “anti-shoot-through” mechanism A “shoot-through” condition occurs when the high-side and low-side switching elements are both turned on, which results in supply node 14 being shorted to ground. Should shoot-through occur during normal switching operations, catastrophic consequences to the switching elements are usually the end result. Adding the anti-shoot-through mechanism serves two purposes: first, it protects the switching elements from damage due to a false ESD trigger that occurs during normal switching operations, although the chance of such an occurrence is remote when the present ESD protection circuit is implemented. Second, it prevents the supply node from being dragged down to ground during the ESD event, thereby enabling the supply node voltage to stay steady and continue to provide power to the ESD protection circuit to maintain its operation.
The present ESD protection circuit is particularly well-suited for use with switching power converters that employ respective pre-driver circuits to drive the high-side and low-side switching elements;
The pre-drivers might be designed to include the NOR logic gates in the desired locations, or existing pre-drivers might be modified as needed to incorporate the NOR logic gates.
In general, when the ESD sensing circuit produces an active-high trigger signal when an ESD event is detected, the NOR logic gate which combines the ESD trigger signal and normal pre-driver signal should be located at stage 2N+1 of the pre-driver chain, where N=0, 1, 2, 3 . . . with the stage nearest the switching element being stage 0. Thus, in
The sequence of events that occurs when an ESD event strikes the SW node of the circuit in
when an ESD event strikes on the SW node, conduction path 16 conveys the event to the V+ node (14).
the trigger signal going high forces the output of NOR gates 22 and 24 to go low, which propagates down the pre-driver chain and arrives at switching elements 10 and 12 as a logic high, thereby pulling up the gates of both the high-side and low-side FETs connected to SW node.
as a result, the high-side PMOS FET is turned off while the low-side NMOS FET is turned on. This enables the ESD event to be discharged via the NMOS FET, while shoot-through is avoided by keeping the PMOS FET off. This sequence of events will be carried out as long as ESD sensing circuit 18 outputs a logic-high trigger signal when ESD strikes on the SW node, regardless of the logic states of S1 and S2.
An implementation that includes pre-driver circuits and which is suitable for use with an active-low trigger signal is shown in
In general, when the ESD sensing circuit produces an active-low trigger signal when an ESD event is detected, the NAND gate which combines the ESD trigger signal and normal pre-driver signal should be located at stage 2N of the pre-driver chain, where N=0, 1, 2, 3 . . . . Thus, in
The sequence events that occurs when an ESD event strikes the SW node of the circuit in
Note that the embodiments shown in
The ESD sensing circuits can be implemented in many different ways. One possible embodiment, based on a design described in U.S. Pat. No. 5,838,146 to Singer and arranged to generate an active-high trigger signal of fixed duration, is shown in
During normal operation, the ESD sensing circuit can be disabled by applying a logic ‘high’ on the ‘disable’ node; this turns on a FET MN2, which pulls down on the input to the inverters and keeps OUT low, thereby preventing false triggering.
Another possible embodiment, arranged to generate an active-low trigger signal, is shown in
A PMOS FET MP2 is connected to pull up on node 92 when on. MP2 is driven with a NAND gate 96 connected at one input to node 92 and at its other input to the output of an inverter 98 which is driven by the output of inverter 90.
When an ESD event strikes the SW node, the V+ node rises rapidly, but C3 holds the input of inverter 90 low for a short time. During this time, the output of inverter 90 is high and turns on MN3, pulling down node 92 and causing OUT to go low. While the output of inverter 90 is high, the output of NAND gate 96 will be high and MP2 will be off.
After a short period, R2 charges C3 enough to cause the output of inverter 90 to go low, causing MN3 to turn off. Node 92 is still low at this point, causing NAND gate 96 to keep MP2 off. This allows a PMOS FET MP3 connected between V+ and node 92 to begin charging a capacitance C4 connected between node 92 and circuit common. When the voltage on C4 is sufficiently high, the output of NAND gate 96 goes low and MP2 is turned on, forcing node 92 and OUT to a high state, in which it is locked to prevent false triggering. The duration of the active-low signal is determined by the RC time constant of C4 and the on-resistance of MP2. A resistor R3 may be included to provide ESD protection for the gate of MP3.
Note that the ESD sensing circuits shown in
While particular embodiments of the invention have been shown and described, numerous variations and alternate embodiments will occur to those skilled in the art. Accordingly, it is intended that the invention be limited only in terms of the appended claims.
Number | Name | Date | Kind |
---|---|---|---|
6369994 | Voldman | Apr 2002 | B1 |
7236339 | Miller et al. | Jun 2007 | B2 |
7538999 | Watanabe | May 2009 | B2 |
20080024169 | Ng | Jan 2008 | A1 |
20080049365 | Worley et al. | Feb 2008 | A1 |
20080123228 | Hung | May 2008 | A1 |
20110080678 | Zhao et al. | Apr 2011 | A1 |
Number | Date | Country | |
---|---|---|---|
20110080678 A1 | Apr 2011 | US |