1. Technical Field
The present invention relates generally to communication link circuits, and more particularly, to digital signal transmitters having selectable drive capability and power consumption.
2. Description of the Related Art
Interfaces between present-day system devices and also between circuits have increased in operating frequency and complexity. In particular, high-speed serial interfaces include transmitters and receivers that typically consume relatively large amounts of the power budget of an integrated circuit. However, depending on channel conditions and parameters, the maximum output signal level of a transmission circuit may not be required for proper signal reception at the remote end of the interface. For example, the channel physical length may be shorter in some applications than in others, reducing signal degradation and thereby reducing transmit power requirements for the same receiver complexity.
Due to limited design resources and the need to satisfy the requirements of multiple interface applications, customers and channel conditions, transmitters and receivers within above-described interfaces are typically designed for the worst-case bit error rates and environmental conditions, leading to relatively complex receivers and high power transmitters. As a result, it is not always possible to provide a transmitter having lower power consumption when a high channel quality is available.
The above-incorporated patent application discloses an interface in which the transmitters and receivers have adjustable and/or adaptive parameters for finely tuning an interface to manage power consumption. One of the controllable parameters is the transmitter power level of the interface driver circuit(s). However, typical transmit power adjustment performed by changing the driver voltage and/or current levels is not always a preferable mechanism to adjust transmit power. Communications links within and between computer subsystems have reached bandwidths of between 5 gHz and 10 gHz and interface frequencies can be expected to increase in the future. Driver circuits operating at such high frequencies do not typically scale operation well over voltage or bias current adjustments, as internal impedances change with such adjustments causing mismatch and loss of power. Delay also typically increases with reduced transmitter power, compromising the integrity of the data window.
Transmitter drivers as described above are not generally simple digital buffers or inverters that switch power rail levels using an effectively near-zero impedance onto the interface line, but are typically linear driver circuits providing multiple stages of amplification or switches having progressively increasing and controlled signal current levels. The cascaded driver circuits progressively raise the power level of the signals internal to an integrated circuit or subsystem to the level required for transmission across the interface channel.
When providing an adjustable power consumption transmitter as used in the above-incorporated patent applications, several different power levels may be required to provide optimum selectability of power level vs. driver performance. In particular, with respect to the Application entitled “METHOD AND SYSTEM FOR INTERACTIVE MODELING OF HIGH-LEVEL NETWORK PERFORMANCE WITH LOW-LEVEL LINK DESIGN”, a large number of unevenly-spaced driver power levels may be required.
It is therefore desirable to provide an interface transmitter circuit having selectable power consumption that provides multiple unequally-spaced power levels. It is further desirable to provide an interface transmitter circuit with a minimum of control logic.
The objective of providing an interface transceiver having multiple unequally-spaced selectable power levels with a minimum of control logic is provide in a method and apparatus.
The apparatus is a transmitter circuit for transmitting a digital signal. The transmitter is divided into slices, each having multiple cascaded driver stages extending from the transmitter input to the transmitter output. The slices are connected in parallel at each driver stage, and include an enable input so that any of the slices can be disabled, providing a programmable power level for the driver circuit. By using unequal power levels for each slice, the range of power selectability is improved over a set of equal power level slices, which can produce only an N:1 range of values where N is the number of slices. Control logic can also be reduced thereby, as the number of slices can be reduced while still meeting a desired range of power level/performance selectability.
The selection and power control process may be programmable by a logic connection, register bit or via a signal from an interface quality measurement circuit. A remote transmitter may also be power-managed at the other end of the interface by transmitting a control signal to the remote transmitter.
The foregoing and other objectives, features, and advantages of the invention will be apparent from the following, more particular, description of the preferred embodiment of the invention, as illustrated in the accompanying drawings.
The novel features believed characteristic of the invention are set forth in the appended claims. The invention itself, however, as well as a preferred mode of use, further objectives, and advantages thereof, will best be understood by reference to the following detailed description of an illustrative embodiment when read in conjunction with the accompanying drawings, wherein like reference numerals indicate like components, and:
With reference now to the figures, and in particular with reference to
Transmitters 16A, 16B of the present invention incorporate power control mode selection inputs provided from programmable register 19 that reduce the power consumption of the transmitter when channel conditions permit. The transmitter (driver) circuit is divided into multiple slices 20 of unequal weights that are calculated to provide optimum coverage of various interface power consumption vs. performance.
Slices that are unneeded within transmitters 16A, 16B are disabled by isolating one or more power supply rails from internal inverters or buffers in each disabled slice. Alternatively the inverters or buffers can be designed with series pass transistors having gates coupled to an enable signal for blocking current through the inverters/buffers to prevent or substantially reduce leakage current flow through the device. Multi-threshold CMOS (MTCMOS) design can be used to further reduce leakage current flow in disabled slices, by using higher threshold voltage devices for the power supply rail isolation devices.
A decoder 15 receives the power level selection signals from programmable register 19 and produces control signals for each slice 20 that enable and disable the individual slices 20 independently. Thus, the above-described transmitter circuits provide a selectable power consumption that can be used to provide lower power usage and dissipation within transceivers 12A and 12B, when channel conditions are good, while maintaining low bit error rates (BERs) using a higher power consumption state when channel conditions are poor.
The selection of power consumption states via programmable register 19 may alternatively be hard-wired, externally programmed using an external data path to programmable register 19 or may be programmed via a reception from receiver 14A. Receiver 14A, transmitter 16A or both may be controlled by one or more selection signals, for example, multiple bits may be provided for each of transmitter 16A and receiver 14A so that power consumption may be very finely traded off for receiver processing power or transmitter signal strength, etc. Details of receivers having selectable power consumption are detailed in the above-incorporated patent applications and those receivers may be used in concert with the transmission circuits of the present invention.
Transceiver 12A is an example of a transceiver having external power mode selection via register programming or external connection. Wired or register selection of power modes is very useful in integrated circuits and systems, including computer systems, communication systems or peripherals. In such systems, external terminal(s) 17 can be hard-wired depending on the application (e.g., known short shielded cable length attached to a peripheral dictates a high channel quality or connection of two transceivers on a high-quality circuit board also dictates high channel quality).
Transceiver 12B is an example of a transceiver having automatic channel-quality-based power level selection in response to a measurement performed by interface quality measurement block 18, which may be an eye-diagram circuit, an error detection circuit or other mechanism for detecting that the channel quality is less than a desired threshold. Select signal SEL B is provided by an output of interface quality measurement block 18 and automatically selects a level of power in conformity with the measured channel quality.
Another type of transceiver power consumption control is provided by an interface link wherein a register such as programmable register 19 may be set via reception of a command code sent over interface 10 and received by a receiver such as receiver 14A. Remote interface link control is useful for informing a transceiver about channel conditions when the transceiver being programmed has no ability to determine the channel quality or does not have information regarding channel conditions (such as cable length). Also, return of channel quality from the remote side of an interface provides absolute information about the transmitted signals, where measurement at the local side provides only an estimate based on reception of signals from the remote side. Therefore, in practice, it would be desirable to provide interface quality measurements at both ends of the interface and return channel quality measurement information to the originating side for controlling transmitter power in accordance with the present invention.
Referring now to
Also shown in details of driver slice 20A are different mechanisms for handling the enable control signal within slices 20A-20D. For inverters/buffers (such as inverters I1 and I3) in close proximity, it is more practical to share the power supply control devices P1 and N1, and thus reduce the overall number of devices required. When the power supply common internal rails (“virtual power supply rails”) would extend over an impractical distance with respect to the desired sizes of power supply control devices P1 and N1, the enable signal is provided to another set of power supply control devices N2 and P2. Inverter I4 supplies a complementary enable signal, but alternatively, the output of inverter I2 may be connected to subsequent stages for controlling the positive power supply rail. In general, any combination of shared virtual power supply rails or bussed enable signals may be used, depending on the physical layout of the driver and the physical length between stages.
By selecting the current scaling for each slice as a particular discrete fraction of the total current for the overall transmitter/driver circuit, a variety of disparate interface requirement goals vs. power consumption goals can be met. If the slices are set to equal current levels, the selection range is N:1 in steps of unit current. If a “binary” division is chosen where each stage has a current level scaling factor of 2n/2N−1 where N is the total number of slices and n is the particular stage, then a reduction in the number of devices and the size of the control architecture is achieved. However, there are a multitude of other non-equal current scaling that can be used to produce results optimized for particular applications and in accordance with other selectable interface features such as receiver processing power. The values can be arrived at experimentally or derived according to methods such as those described in the above-referenced patent application entitled: “METHOD AND SYSTEM FOR INTERACTIVE MODELING OF HIGH-LEVEL NETWORK PERFORMANCE WITH LOW-LEVEL LINK DESIGN.”
The methodology described in the above-referenced patent application determines a transmitter signal strength for particular interface configurations based on required bit error rates (BERs) and jitter constraints. The determination is made in conjunction with other parameters such as receiver complexity, etc. so that the interface performance is assured under the selected conditions. By providing an arbitrary set of transmitter power levels based on the scaling factor of each driver slice 20, the ability to select driver power in conjunction with the other interface parameters and channel conditions provides a robust solution to interfaces that must serve a wide variety of applications.
While the invention has been particularly shown and described with reference to the preferred embodiment thereof, it will be understood by those skilled in the art that the foregoing and other changes in form, and details may be made therein without departing from the spirit and scope of the invention.
The present application is related to co-pending U.S. patent application Ser. No. 10/289,777, attorney docket No. AUS920020561US1 entitled “INTERFACE TRANSCEIVER POWER MANAGEMENT METHOD AND APPARATUS”, filed on Nov. 7, 2002 and Ser. No. 10/829,829 attorney docket No. AUS920040086US1 entitled “METHOD AND SYSTEM FOR INTERACTIVE MODELING OF HIGH-LEVEL NETWORK PERFORMANCE WITH LOW-LEVEL LINK DESIGN”, filed on Apr. 22, 2004, both of said applications having at least one common inventor with, and assigned to the same assignee as, the present application. The specifications of the above-referenced patent applications are herein incorporated by reference.
Number | Date | Country | |
---|---|---|---|
Parent | 11050019 | Feb 2005 | US |
Child | 12024448 | US |