CMOS apparatus for driving transmission lines

[0002] The application is a Divisional of co-pending application Ser. No. 09/476,425, filed Dec. 30, 1999 by applicants Jed Griffin and Ernest Khaw, entitled “A Constant CMOS Driver,” of which is a continuation-in-part of U.S. patent application, Ser. No. 09/108,606, filed July 1, 1998, now U.S. Pat. No. 6,137,317.

[0003] 1. Field of the Invention

[0004] Embodiments of the present invention relate to driver circuits for driving transmission lines, and more particularly to complementary metal oxide semiconductor (CMOS) driver circuits.

[0005] 2. Background

[0006] A dominant limitation of conventional manufactured driver circuits is their artificially low transmission rates due to widely varying operating conditions, such as voltage, temperature, and process variation. Due to varied operating conditions, the propagation delay and the output impedance of drivers varies widely, thus, hampering impedance matching.

[0007] Propagation delay can vary typically by a factor of two to three across two extreme operating conditions. This variation of propagation delay seriously impacts system timing at higher frequencies. Without a constant delay across all operating conditions, system timing is adversely impaired such that timing margins have to be introduced to handle any delay time variations due to varying operating conditions.

[0008] A most common and useful communication topology is peer-to-peer connections with full duplex transmission. To achieve optimal impedance matching in this type of topology, the output impedance of the transmitting side must match the characteristic impedance of the transmission line. Impedance matching at the transmitting end has traditionally been accomplished by placing a series resistor between the output driver and the transmission line. For this method to work, the output impedance of the output driver must be kept much lower than the characteristic impedance of the transmission line. This results in a much higher cost in area and power than required for merely transmitting a signal. Moreover, impedance matching is degraded due to varying resistance across operating conditions and the non-linearity of the driver. Another method is to use the nonlinear transistors of the output driver to approximate the linear characteristic impedance of the transmission line. This attempt, however, results in even worse impedance matching than a series resistor placed at the transmitting end.

[0009] FIG. 1 illustrates an I/O (input/output) driver 102 communicating with a receiver 104 via a transmission line 106. The transmission line 106 has a characteristic impedance Z_o, and may be the physical layer of a bus. The driver 102 and the receiver 104 are complementary metal oxide semiconductor (CMOS) circuits. For purposes of mathematical analysis, the input impedance (Z_in) of the receiver 104 is approximated as being infinite relative to other impedances in the circuit. The receiver 104 may be one or more CMOS logic gates, or a differential amplifier.

[0010] The driver 102 is transmitting an electromagnetic wave travelling in the transmit direction 108. If Z_inof the receiver 104 is not equal to Z_o, then a reflected wave will propagate in the receiver direction 110. If the impedance of the driver 102 is not matched to the characteristic impedance Z_othen another reflected wave will again be generated, but now travelling in the transmit direction 108. There will be many multiple reflections, and the electric and magnetic field vectors at any point along the transmission line 106 is the vector sum (superposition) of the transmitted field vector and all reflected field vectors at that point. This superposition of the transmitted wave and the reflected waves may cause signal degradation, which typically limits, for longer transmission lines, the speed at which digital data is reliably transmitted from the driver 102 to the receiver 104.

[0011] The first reflected wave can be reduced by terminating the receiving end of the transmission line 106 with a receiver or stub having an impedance matched to Z_o. This may, however, require the use of an off-chip resistor, and furthermore, power may be wasted due to ohmic losses in the resistor. Another negative impact of impedance matching at the receiver end is loss of amplitude, potentially halving the amplitude, which can then make the transmitting signal susceptible to noise.

[0012] FIG. 1 is an illustration of a driver in communication with a receiver over a transmission line, where the receiver is not matched to the transmission line.

[0013] FIG. 2 is an embodiment of the present invention.

[0014] FIGS. 3aand 3bare approximations to the embodiment of FIG. 2 for two particular conditions.

[0015] FIG. 4 is an embodiment of the present invention having programmed linear output impedance.

[0016] FIG. 5 is an embodiment of the invention comprising a pre-drive stage having programmable constant delay.

[0017] FIG. 6 is an illustration of a plurality of p-channel and n-channel stacks with delay/impedande select lines.

[0018] Embodiments of the invention are described for input/output (I/O) drivers having an output impedance that is approximately independent of the output voltage, and which can be adjusted based upon variations in temperature, supply rail voltage, and variations in transistor dimensions (channel length, width, etc.). This allows for matching the driver's output impedance with the characteristic impedance of a transmission line driven by the I/O driver. If the output impedance is matched to the impedance of the transmission line, an electromagnetic wave that is reflected toward the driver will substantially cease from reflecting again. Therefore, signal degradation is reduced, thus, allowing for faster, more reliable data transmission. The exemplary embodiments are provided to illustrate the embodiments of the invention and should not be construed as limiting the scope of the embodiments of the invention.

[0019] Reference in the specification to “an embodiment,” “one embodiment,” “some embodiments,” or “other embodiments” means that a particular feature, structure, or characteristic described in connection with the embodiments is included in at least some embodiments, but not necessarily all embodiments, of the invention. The various appearances “an embodiment,” “one embodiment,” or “some embodiments” are not necessarily all referring to the same embodiments. If the specification states a component, feature, structure, or characteristic “may”, “might”, or “could” be included, that particular component, feature, structure, or characteristic is not required to be included. If the specification or claim refers to “a” or “an” element, that does not mean there is only one of the element. If the specification or claims refer to “an additional” element, that does not preclude there being more than one of the additional element.

[0020] Relevant terminology will first be introduced. Two element Boolean algebra is relevant to switching circuits. For any point in a circuit, the term LOW will denote a set of voltages that map into one of the two Boolean elements. The term HIGH will denote a set of voltages that map into the other of the two Boolean elements. The particular mapping into Boolean elements depends upon the technology used, and may be different for different parts of a single circuit. To avoid dealing with set terminology, we shall say that a voltage is LOW/HIGH if it belongs to the set LOW/HIGH. We also follow the convention that for any given node within a circuit, LOW voltages are generally less than HIGH voltages. Referring to the figures, exemplary embodiments of the invention will now be described.

[0021] In one embodiment of the invention circuit 200 in FIG. 2 is part of an I/O driver or buffer suitable for driving a transmission line. The circuit 200 provides an approximately constant output impedance when the transmission line is being driven. That is, the impedance of the circuit 200, when “looking into” a port defined by the terminal 202 and power supply ground is approximately constant when the transmission line is being driven by the circuit 200.

[0022] Transistor set 206 and 212 and transistor set 208 and 216 are switched in complementary fashion with respect to each other to drive the transmission line. The circuit 200 is shown in FIG. 1 without the capability to tristate the output node 202 because V_nand V_pare shorted to the input node 204. In some embodiments, however, they are approximately synchronous, and V_pand V_nare such that V_ptransitions from HIGH to LOW before V_ntransitions from HIGH to LOW, and V_ptransitions from LOW to HIGH after V_ntransitions from LOW to HIGH, so that p-channel metal oxide semiconductor field effect transistor (pMOSFET) set 206 and 212, and n-channel metal oxide semiconductor field effect transistor (nMOSFET) set 208 and 216 are not simultaneously ON. Circuit 200 can be tristated if V_pis set HIGH and V_nis set LOW. Note that transistors 212 and 210 are considered a stack of pMOSFETs, and transistors 214 and 216 are considered a stack of nMOSFETs, where a stack of transistors consists of at least two transistors that are stacked together.

[0023] The combination of 200 is such that, when V_pand V_nare both LOW the drain of pMOSFET 206 is approximately at the output voltage V_oand nMOSFET 208 is OFF; the drain of pMOSFET 210 is approximately at the output voltage V_oand nMOSFET 214 is OFF; and pMOSFET 212 is ON and nMOSFET 216 is OFF. When V_pand V_nare both HIGH, the drain of nMOSFET 208 is approximately at the output voltage V_o, and pMOSFET 206 is OFF; the drain of nMOSFET 214 is approximately at the output voltage V_o, and pMOSFET 210 is OFF; and nMOSFET 216 is off and its drain is approximately zero volts (ground) and pMOSFET 212 is OFF. Transistors 206 and 208 are considered to be the drive-stage of circuit 200.

[0024] The approximate constant output impedance property of circuit 200 can be understood by considering FIGS. 3aand 3b, which provide approximations to FIG. 2 for the cases in which V_pand V_nare both LOW and in which V_pand V_nare both HIGH, respectively. The circuits of FIGS. 3aand 3b, however, do not model circuit 200 during logic transitions of voltages V_pand V_n.

[0025] For purposes of finding an approximate expression for the output impedance of the circuit in FIG. 3awhen transistors 208. 214, and 216 are ON, let I_ds1denote the drain-source current of nMOSFET 208, and I_ds2denote the drain-source current between nMOSFETs 216 and 214. Let Z_outdenote the output impedance of the circuit of FIG. 3a. Then the output impedance is given by

[0026] Vo=Z_out(I_ds1+I_ds2)

[0027] Noting that the drain-source voltage (V_DS) of nMOSFET 208 is equal to V_o, and assuming that the threshold voltage of nMOSFET 208 is much less than the supply node V_DD, then an approximate expression for the drain-source current of nMOSFET 208 is given by $I_{\mathrm{ds1}}=\frac{{\beta }_1}{2}\left[2V_{\mathrm{DD}}{V}_o-V_o^2\right]$

[0028] where β₁is the beta for nMOSFET 208 and we assume that nMOSFET 208 is in its linear or nonsaturation region. Transistors nMOSFET 214 and 216 are configured to be in their saturation region when ON, and provided their threshold voltage V_Tis much less than V_o, an approximate expression for the drain-source current of nMOSFET 214 and 216 is given by $I_{\mathrm{ds2}}=\frac{{\beta }_2}{2}\left[2V_o^2-V_o^2\right]\simeq \frac{{\beta }_2}{2}{V}_o^2$

[0029] where β₂is the beta of nMOSFETs 214 and 216. Substituting the above two expressions into the expression for the output impedance yields. $\mathrm{Vo}=Z_{\mathrm{out}}\left[{\beta }_1{V}_{\mathrm{DD}}{V}_o-{\beta }_1\frac{V_o^2}{2}+{\beta }_2\frac{V_o^2}{2}\right]$

[0030] If the betas of transistors 208, 214 and 216 are matched and denoted by β, then the above expression yields. $Z_{\mathrm{out}}=\frac{1}{\beta V_{\mathrm{DD}}}$

[0031] As seen from the above displayed equation, the output impedance of the circuit of FIG. 3a(when transistors 208, 214 and 216 are ON) is approximately constant (i.e., independent of V_o). A similar analysis shows that the output impedance of circuit 3bis also (approximately) given by the above displayed equation, provided the betas are also properly matched. Also note that the capacitance and inductance seen at the output node are also approximately constant, whether the driver is driving high or low. This is key to ensure that the imaginary part of the impedance also remains approximately constant, neglecting slight variations due to voltage coefficients from different levels of V_o. This, however, has more direct impact on the output driver slew rate than it does on impedance matching.

[0032] In another class of embodiments, in FIG. 4 transistors 206 and 208 in Block 0 may be joined by a first and second plurality of drive transistors, such as transistors 410 and 412 in Block N, respectively, so that subsets of the first and second pluralities of drive transistors can be selected so as to provide a programmable output impedance. In this case, to match betas, transistors 210, 212 (upper impedance elements), 214 and 216 (lower impedance elements) in Block 0 would be joined by a plurality of transistors, such as transistors 414, 416, (upper impedance elements), 418 and 420 (lower impedance elements), so that the proper subset of plurality transistors can be selected, depending upon the selected subset of the drive transistors, so that the output impedance is programmable and approximately independent of output voltage V_o.

[0033] An embodiment belonging to the previously described class of embodiments is illustrated in FIG. 4, where corresponding components in FIGS. 2 and 4 have the same numeric label. In FIG. 4, signals Sp₀, SP_n, Sn₀and Sn₀on impedance select lines 402, 424, 404, and 426, respectively, are select signals. If Sp_nand SN_nare LOW, and Sp₀and Sn₀are HIGH, then transistors associated with Sp_nand SN_nare OFF and circuit 400 behaves as circuit 200 in FIG. 2. If Sp_nand SN_nare each HIGH, then the transistors associated with Sp_nand SN_naffect the output impedance.

[0034] To determine the output impedance when Sp_nSN_nSp₀and Sn₀are each HIGH, consider betas of transistors 208, 214 to be matched and denoted as β₁, the beta of 216 to be much greater than β₁, the betas of transistors 412, 418 to be matched and denoted as β₂, and that of 420 to be much greater than β₂. Then a similar analysis as discussed earlier yields the approximate expression for the output impedance, $Z_{\mathrm{out}}=\frac{1}{\left[{\beta }_1+{\beta }_2\right]V_{\mathrm{DD}}}$

[0035] By utilizing a plurality of transmission gates, impedance select lines, and matched pairs of transistors as in FIG. 4, the output impedance can be programmed and still remain approximately independent of output voltage.

[0036] The circuit illustrated in FIG. 5 illustrates one embodiment of the invention having a pre-drive stage (502, 504, 518, 520) coupled with a programmable bias circuit that may be used with previous discussed embodiments. This embodiment of the invention features a programmable technique for keeping constant the propagation delay of the driver, which causes the propagation delay to vary from its design value. The particular operating conditions can be compensated for by programing the actual, observed delay via delay select lines Sp₀(514) to Sp_n(516) and Sn₀(510) to SN_n(512). Note that SP_b(522) and Sn_b(524) are always ON. The delay value of delay select lines Sp₀(514) to Sp_n(516) and Sn₀(510) to SN_n(512) would correspond to proportionally large transistors which they are coupled with, namely 526, 528, 532, and 534 respectively. To reduce the amount of delay (for faster operating conditions), smaller transistors and fewer delay select lines would be asserted. Contrarily, to increase the amount of delay (for slower operating conditions), larger transistors and more delay select lines would be asserted. The sensitivity at which the operating conditions are sampled determines the number of delay select lines and correspondingly how tight the interval around a nominal delay can be set to. If delay select lines Sp₀(514) to SP_n(516), and Sn₀(510) to SN_n(512) are OFF, then transistors 526, 528, 532, and 534 are OFF, and circuit 500 would only consist of transistors 518, 520, 530 and 536. If Sp₀(514) to SP_n(516) are ON, then transistors 526, 528 affect the propagation delay. If Sn₀(510) to SN_n(512) are ON, then transistors 532 and 534 affect the propagation delay. Typically, to ensure that V_ois substantially equal to V_i, Sp₀(514) and S_no(510), are both turned on together. Accordingly, Sp_n(516) and SN_n(512) would be turned ON together. The stacked configuration of the selected transistors with the actual drive transistors also serves to prevent hot electrons.

[0037] The embodiments of the invention described above are, of course, subject to other variations in structure and implementation. For instance, additional devices may be inserted between various nodes, terminals, and devices in the above embodiments without materially changing their overall function. For example, voltage drops may be introduced by resistors, diodes, or transistors configured as diodes, to change various voltage levels, or buffers may be inserted between various nodes, terminals, and devices. In general, the scope of the invention should be determined not by the embodiments illustrated but by the appended claims and their legal equivalents.