Ultra320 SCSI and adaptive active filtering: the alternative to transmitter pre-compensation. (High Availability).
SCSI technology continues to evolve to successfully meet the increasing demand for input/output (110) bandwidth. The SCSI interface offers stability, ease of connectivity, a large installed base, and a 20-year history of full backward compatibility. Ultra320 SCSI is the latest evolution and provides the highest-ever data bus bandwidth. Ultra320 SCSI is designed to handle the most demanding enterprise server applications.
The ANSI INCITS T10 Technical Committee (T10), with input from the SCSI Trade Association (STA), develops standards that define the requirements for the parallel SCSI interface. Standardized definitions for new SCSI transfer rates have been devised to prevent the interface from becoming a performance bottleneck.
Ultra320 SCSI, with a maximum transfer rate of 320MB per second, is the latest development for the SCSI parallel interface. Ultra320 SCSI is designed to handle the most demanding enterprise server applications. New features to enhance performance, improve data reliability, and increase ease of use have been developed by T10. Adaptive Active Filtering (AAF) is one of these features and will be discussed in detail in this article. For specifics on all Ultra320 SCSI features, refer to the ANSI draft standard document SCSI Parallel Interface-4 (SPI-4). The latest revision of this draft is available for review at www.t10.org.
Compensating for Transmission Line Signal Losses
One of the primary benefits of the SCSI interface over the ATA interface is its ability to support multiple devices on a single bus while providing higher reliability and signal integrity. By supporting more devices on the bus, SCSI provides better throughput by spreading the load across multiple hard disk drives, aggregating these drives' data rates over the SCSI bus. This results in exceptional system performance for enterprise-class servers and RAID applications. A cable interconnect system is used to provide a communication link between these devices. With this comes inherent transmission line issues, making it more difficult to achieve a good signal-to-noise ratio (SNR) to send and receive data accurately. As digital data transfer speed increases, the inter-symbol interference (ISI) effect in electronic communication via a long cable becomes significant to a point where simple voltage level detection is not reliable to extract data from the incoming signal. Ultra320 SCSI provides two methods to resolve t his problem: transmitter pre-compensation and adaptive active filtering.
Transmitter pre-compensation with cutback: This is an open-loop method of compensating for some of the signal loss that is most severe on the first part of a signal's transition. The transmitting device boosts the amplitude of the first part of the transition or cuts back the signal for the remainder of the transition. This provides additional signal amplitude where it is most needed and then decreases the amplitude to decrease the negative effects of cross-talk and reflections. Transmitter pre-compensation is sub-optimal because it cannot monitor changing conditions in the cable plant (i.e. adding devices to the bus). The signal boost is static.
Adaptive active filtering (AAF): Also known as "receiver equalization with filtering," AAF is a closed-loop method of improving received signal quality by amplifying the fundamental frequency of the signal in the receiver while filtering noise and other undesirable components. Devices implementing AAF establish the gain of their amplifiers by setting the amplitude of the high frequency portion of the training pattern to be the same as the low frequency portion at the beginning of the training pattern. Using the training pattern to perform this adjustment of signal amplitude provides for an inherent closed-loop system that can adjust signal quality for different cable plants and changes in system conditions (e.g., when a new device is added to a system causing the electrical characteristics of the cable plant to change). AAF settings may be adjusted as often as necessary because either the initiator or target may initiate a training pattern sequence. A receiver may disable transmitter pre-compensation in a tra nsmitter as a configuration using AAF performs better than a configuration using pre-compensation.
The AM Advantage
AAF improves signal integrity and, as a consequence, maximizes system performance. Signals processed with AAF enabled are sharper and have a better peak-to-peak amplitude definition. AAF enhances electrical signal margins reducing the effect of signal losses due to back-planes and cables. AAF compensates for variations in temperature, voltage, and process.
Issues with transmitter pre-compensation addressed by AAF:
* One level of cutback is not best for all systems. AAF allows devices to optimize the frequency response that best fits existing conditions.
* Pre-compensation causes increased crosstalk and reflections and should actually be disabled in some systems.
* Pre-compensation is open-looped with no opportunity for adjustment. AAF allows continuous optimization throughout a system's operation.
* Pre-compensation is inefficient, a significant portion of the power added at the transmitter is dissipated in the system before it gets to the receiver. AAF is tailored for the specific system environment.
* Pre-compensation requires significantly more power than AAF. Higher power equates to higher thermal dissipation and reduced reliability.
* Transmitter pre-compensation will not work for all legal Ultra320 SCSI configurations. AAF will.
How AAF Works
AAF is used to counter the effect of inter-symbol interference (151) by processing the signal to make data detection more reliable. The AAF scheme allows the system to perform optimally without intervention. The AAF approach automatically amplifies the frequency components that carry the data information and attenuates nonessential frequency components to reduce 151 as well as the unwanted electrical noise.
ISI cancellation is a technique that has been applied in various digital communication systems and is mainly used in disk drive read channels. One common approach for ISI cancellation is to slim the digital pulse using a filter with a second-order zero ([s.sup.2]) in the frequency domain, such that minimum energy is left from this pulse to affect the adjacent pulses.
Cable resistance (R) and capacitance (C) are the main factors of the ISI effect. ISI limits the maximum allowable speed for digital data transmissions. If the RC effects were compensated, the majority of the ISI effect would be significantly reduced. AAF is a frequency-boosting scheme using a first-order zero (s) up to the maximum frequency of the digital information, attenuating higher frequencies that do not carry information to reduce noise. The gain of the (s) term boost is adjusted such that the amplitude of the high frequency equalized signal is proportional to the amplitude of its DC voltage logic level. The advantage of this technique is its simplicity, allowing the entire equalization scheme, including the analog functions, to be easily implemented in a digital IC fabrication process. This equalization scheme allows the data transfer rate to reach 320MB/sec or higher in a SCSI environment.
When SCSI transfer rate increases from 160MB/sec to 320MB/sec, the frequency components that carry the digital information are doubled, and the cable skin effect resistance is also increased. Therefore, the RC effect becomes more prominent. With a long cable, the signal amplitude varies a lot depending on the data pattern. In certain conditions, the isolated pulse amplitude relative to the threshold level could be 10 times smaller than the DC logic level. In this case it is impossible for a simple comparator, used in earlier SCSI implementations, to reliably detect the data.
The AAF equalization circuit consists of a low-pass filter with a first-order frequency boost, a sample and hold, two comparators, a digital counter and some control logic, as shown in Figure 3. The main equalization function is performed by the low-pass filter, which has adjustable frequency gain boost. The rest of the circuitry is used to determine optimum frequency gain boost for equalization as the optimum gain boost depends on the location of the receiving device on the cable. In a SCSI environment, at the beginning of data transmission, the transmitting device drives the signal to a negated logic level for a period of time long enough for the entire signal path to settle to its most negative voltage level. At this time the negative voltage is captured and stored by the sample and hold circuit. The stored negative voltage is used as the negative voltage reference of the latching comparator. After the initial negative assertion period, the SCSI transmitter will send out a logic pattern of 101010, which is the highest transmission frequency. A comparator is used to detect the 101010 pattern and generate a clock for the equalization circuit. The latching comparator detects any incoming signal that is above the stored negative voltage reference. If the signal is higher than the stored reference voltage, it tells the digital counter to count up, otherwise it tells the counter to count down. The digital counter controls the gain of the frequency boost. The frequency boost gain is proportional to the counter value that is in the range from zero to its maximum gain. In other words, the whole equalization scheme is making the negative amplitude 101010 pattern equal to the negative DC assertion voltage level.
AAF is as simple as:
* Applying a low-frequency pattern.
* Sampling and storing the low-frequency signal amplitude at the equalizer output.
* Applying a maximum frequency "101010..." pattern.
* Adjusting equalizer boost to match the "101010..."-pattern-equalized amplitude to the stored low-frequency amplitude value.
[FIGURE 1 OMITTED]
Jorge L. Fernandez is program marketing manager, Mark Evans is senior technical marketing manager, Russ Brown is senior member of the technical staff Ivan Chan is senior staff engineer--all at Maxtor (Milpitas, CA).
|Printer friendly Cite/link Email Feedback|
|Publication:||Computer Technology Review|
|Date:||Jul 1, 2002|
|Previous Article:||New 15K drives provide more than speed: they offer increased safety in multi-TB storage. (High Availability).|
|Next Article:||Back-end switching in storage server design: improves the performance and availability of storage systems. (High Availability).|