Electronic switch for decoupling capacitor

[0002] This application is a divisional application Ser. No. 08/418,971 filed on Apr. 7, 1995, now U.S. Pat. No. 5,506,457.

[0003] 1. Field of the Invention

[0004] The present invention generally relates high speed integrated circuit (IC) chips which experience large changes in current as circuits switch and, more particularly, to the addition of decoupling capacitors to limit ΔI noise without the attendant reduction in chip yield normally encountered with the addition of such decoupling capacitors.

[0005] 2. Background Description

[0006] In high performance chips where there are large changes in current as circuits switch, it has become necessary to add capacitance between the power supply and ground leads to limit ΔI noise. This capacitance can be formed from normal device thin oxide. However, any defects in this added thin oxide can result in high leakage currents or shorts due to defective capacitors, lowering chip yield and greatly increasing the cost of production of high performance chips.

[0007] It is therefore an object of the present invention to provide decoupling capacitors for high performance IC chips to limit ΔI noise without the attendant reduction in chip yield normally encountered with the addition of such decoupling capacitors.

[0008] According to the invention, there is provided an electronic switch consisting of a device in series with a capacitor so that only good capacitors are connected. The bad capacitors are switched out automatically by means of feedback at each capacitor. The circuit comprises a control device that operates in the subthreshold or off device state to detect leakage in a decoupling capacitor. This control device operates in a low impedance state if the capacitor is good and in a high impedance sate if the capacitor is bad. A feedback circuit is connected from an internal node of the capacitor to a gate of the control device so that once a state of the capacitor is detected it can be stored on the gate of the control device. A single external signal source shared by a group of capacitors activates the control device to detect leakage in the capacitor. The circuit operates to switch out capacitors that fail during normal operation.

[0009] The foregoing and other objects, aspects and advantages will be better understood from the following detailed description of a preferred embodiment of the invention with reference to the drawings, in which:

[0010] FIG. 1 is a schematic diagram showing the equivalent circuit of the chip structure according to a preferred embodiment of the invention;

[0011] FIG. 2 is a cross-sectional representation of the chip structure for the capacitor and the control device;

[0012] FIG. 3 is a schematic diagram showing the equivalent circuit of a second embodiment of the is invention incorporating a NOR gate; and

[0013] FIG. 4 is a schematic and logic diagram the circuit of FIG. 3.

[0014] Referring now to the drawings, and more particularly to FIG. 1, there is shown the equivalent circuit of the structure according to a preferred embodiment of the invention. The circuit consists of the decoupling capacitor 10, an n-channel control device 11 in series with the capacitor and ground, an inverter composed of p-channel device 12 and n-channel device 13 connected between the shared node 14 of the capacitor and control device 11 and node 15 connected to the gate of the control device, and a further n-channel device 16 connected between node 15 and ground with its gate connected to a common line 17 that is connected to each capacitor circuit.

[0015] During chip power up, a Capacitor Power On Reset (CPOR) signal is applied to the common line 17 connected to the gate of device 16. CPOR is switched high, causing node 15 to go low and thereby turning off control device 11 except for a small amount of subthreshold conduction current. This current is in the range of 0.1 pA to 0.1 nA per micron of channel width (variation around minimum channel length at 25° C.). If the capacitor 10 is good, the leakage through the oxide will be extremely small (<<fA) and node 14 will leak to a low level through the control device 11 in about 0.002 to 1 second, depending on the channel length of the control device. If the capacitor is bad or leaky (>1 nA to short), node 14 will quickly rise to a high level. The only requirement on device sizes it that n-channel device 16 must be strong enough to pull node 15 low when node 15 is low and p-channel device 12 is attempting to pull node 14 high. Making devices 12 and 13 minimum width devices and device 16 twice minimum width should work well and provide adequate margin. This should give a four fold difference in current between device 12 and device 16 (i.e., twice the difference from the width and twice the difference from the transconductance) insuring that node 15 goes low. At the cost of an additional p-channel device, this requirement could be eliminated. The added p-channel device 21 could be in series with device 12 and with its gate connected to the global CPOR line 17 as shown in FIG. 3.

[0016] For the case of a good capacitor, node 14 will be low and when the CPOR signal goes low, node 15 will be pulled high through the inverter p-channel device 12, causing the control device 11 to turn on. Thus, the good capacitor is switched on. If the capacitor 10 is bad, node 14 will be high and when the CPOR signal goes low, node 15 is held low through device 13 of the inverter and the control device 11, and hence the capacitor, is off. For the case where the leakage of the capacitor 10 is about the same as the subthreshold current of the control device 11, node 14 could be in an intermediate state when the CPOR signal goes low. If the intermediate voltage on node 14 is above the switching voltage of the inverter and is rising with time, the control device 11 will remain off and the capacitor 10 is switched off. Conversely, if the intermediate voltage on node 14 is above the switching voltage of the inverter but falling with time, the inverter will switch when the voltage on node 14 reaches the switching voltage of the inverter. When this happens, node 15 will rise and the control device 11 will turn on, switching on the capacitor 10. The capacitor 10 will also be switched on if the voltage on node 14 is at an intermediate level below the switching voltage of the inverter when the CPOR signal goes low.

[0017] Only those capacitors which have lower leakage than the subthreshold conduction of the control device 11 will be switched on after the CPOR signal goes low. If during normal operation a capacitor fails, it will be switched out if its leakage is large enough to cause node 14 to rise above the inverter switching voltage. This provides protection against any high leakage fails that occur after the initial CPOR cycle. Of course, the procedure of pulsing CPOR high could be repeated at any time after the chip is powered up providing it does not interfere with normal chip operation.

[0018] In order for the electronic switch to work, it has been assumed that if the capacitor 10 is leaky or shorted that the only path for current to flow is through the control device 11. High leakage to the substrate from the capacitor plate 20 connected to the power supply is a concern and a potential problem. This can be avoided by placing the capacitor 10 in an n-well 18 as shown in the cross-section of the capacitor and control device 11 shown in FIG. 2. Assuming the substrate is at ground, the only paths from the capacitor 10 to ground are through the off device current of the control device 11 and through the reversed biased n-well 18 and drain junction 19 of the control device. Thus, by placing the capacitor in the n-well, the potential of a high leakage path to substrate is eliminated, and the circuit operates as described. Any additional leakage from the n-well and device drain diffusion will add to the control device subthreshold leakage and will help speed up the discharge of node 14 if the capacitor 10 is good.

[0019] In an alternative embodiment shown in FIG. 3 and FIG. 4, a classic NOR gate is formed by the addition of a p-channel device 21. In FIG. 3, the same reference numerals are used to reference the same components as in FIG. 1. This embodiment adds a second p-channel device 21 in series with p-channel device 12. The gate of p-channel device 21 is connected to line 17. Thus, the line 17 carrying the CPOR signal comprises one input of the NOR gate 22 shown in FIG. 4, while the junction of the capacitor 10 and the control device 11 comprises the second input to the NOR gate. Without the additional p-channel device 21, it is possible to pass a significant amount of current during initial turn on through devices 12 and 16 of many parallel circuits of this type. The NOR gate configuration significantly reduces this current by eliminating the path for current.

[0020] The foregoing description assumes a capacitor formed by planar thin oxide. The same technique is applicable to capacitors formed by other means, including between metal layers or using trenches.

[0021] While the invention has been described in terms of a single preferred embodiment, those skilled in the art will recognize that the invention can be practiced with modification within the spirit and scope of the appended claims.