Written by Min-Jie Chong
The Need for New SAS-4 Storage Standard
The increase of data traffic due to the advent of internet of things has driven the need for faster backbone and storage transmission to meet this need. The Serial Attached SCSI - 4 (SAS-4) is a new enterprise storage standard that is being created to meet this need. It supports data rate of 22.5 Gb/s which doubles the data throughput of previous generation SAS-3 standard.
What is SAS-4 standard?
The SAS-4 working committee decided to leverage the OIF-CEI 3.1 specification to speed up the development of SAS-4 specification. The consequence is the test methodology in SAS-4 will deviate from the previous SAS generations. In the previous generation, reference transmitter and receiver are defined, which describe how a “perfect” design would handle the outgoing and incoming signal. However, the OIF-CEI specification does not provide any reference designs, which changes how the SAS-4 designs are tested.
First thing first, accessing SAS-4 signals
SAS-4 specification has a new recommendation of the insertion loss profile of the test fixture being used for testing. The intents are to more accurately test the 22.5 Gb/s signal without the effect of test fixtures so the industry can get more consistent results and avoid marginal design from passing using better test fixtures, but not meet the actual performance in real world. Keysight has found the Wilder Technologies SAS-3 test fixtures to be suitable for SAS-4. They perform better than what the specification recommends. This is a good outcome because it is easier to supplement loss using the embedding methodology, using Keysight’s N5465A InfiniiSim software toolset.
Oscilloscope bandwidth requirement
The SAS-4 specification does recommend a minimum of 33 GHz oscilloscope bandwidth for transmitter test. Keysight’s Z and V Series oscilloscope models (i.e. DSAV334A, DSAZ334A and DSAZ504A) have bandwidth that meet this requirement.
Making SAS-4 transmitter measurements
Keysight’s new N5412E SAS-4 transmitter test application provides step-by-step instructions to guide an engineer through the process of configuring the test setup, selecting the tests and connecting the signals to the oscilloscope. After everything is setup correctly, the application will then make the necessary measurements and analysis, and then presenting a pass or fail status of the signal under test. A test report will also be automatically generated at the end of this process, documenting the test results and measurement screenshots. This can really remove the complexity of learning the specifications, which can save engineers a lot of time and effort.
Figure 1: Keysight N5412E SAS-4 automated test application for the oscilloscope, which covers all the required transmitter test requirements.
The application includes all the transmitter requirements listed below.
- Spread spectrum clocking (SSC)
- Transmitter signal quality (TSG)
- Transmitter equalization (TXEQ) coefficient request and circuit response
- Out-of-band (OOB) signaling
SAS-4 is highly susceptible to crosstalk
SAS-4 interface packs a lot of high speed lanes densely in a connector, which makes it highly susceptible to crosstalk effect. It is important for oscilloscope jitter separation algorithm to be able to handle presence of crosstalk. The earlier, more common jitter separation with the “spectral” method is not capable of separating crosstalk from random jitter. Keysight oscilloscope uses a newer, more advanced “tail fit” method that can correctly separate the effect of crosstalk from the random jitter.
After determining the presence of crosstalk, the Keysight N8833A crosstalk analysis tool can provide deep analysis and debug capabilities. The tool is able to identify which potential aggressor is aggressing at the victim, quantify the amount of crosstalk the aggressor is coupling into the victim and then removing the crosstalk from the victim signal. We can check if the design can pass the specification and how much margins can be recovered without the crosstalk. This can assist in making important design decisions such as whether improving the crosstalk can make our design passes the specification, and which part of the design needs to be fixed.
Figure 2: Keysight N8833A crosstalk analysis application showing the eye diagram before (top) and after (bottom) removing crosstalk from the signal. The eye height and width can be measured to see the improvements of the signal without crosstalk.
Vendor specific SAS-4 receiver equalization implementation
While SAS-3 mandates 5 tap of decision feedback equalizer (DFE) implementation to recover a closed eye at the receiver, the SAS-4 does not mandate any specific number of taps. It is left to the vendor specific implementation how many taps will be sufficient to open up the eye. Keysight’s N5461A equalization tool with DFE allows engineers to recover the eye with up to 40 taps. Engineers can specify the value for each tap and check the effect on the eye opening or use the tap optimization feature that will compute the values based on the constraints given by the engineers. This feature is very useful to reproduce the eye opening that the receiver sees after the DFE process.
Figure 3: Keysight’s N5461A equalization tool is used to open up the closed eyes at various SAS data rates, and what the SAS receiver would see after DFE is performed.
Keysight has been a key contributor to SAS-4 and previous standards, and understand the test requirements. New test and interoperability challenges exist at 22.5 Gb/s and Keysight has the solutions to overcome these challenges. The automated N5412E SAS-4 test application covers the complete transmitter test requirements. Other tools such as the N8833A crosstalk analysis and N5461A equalization application can provide deep analysis and debug capability. In addition, Keysight has other comprehensive test solutions from design simulation to physical layer testing that includes transmitter, receiver and channel for SAS-4 standard.