There is a requirement for high-performance test instruments at all stages of a customer?s high-speed design and validation process. Not only do these instruments require ultra-high bandwidth, signal fidelity and acquisition performance, they also need a wide range of trigger capabilities to aid in analysis and compliance testing.
A system known as Pinpoint triggering uses a dual A- and B-triggering system that allows over 1400 trigger combinations. For customers who need to ?zero in? on fault conditions in debug applications and specific waveform areas of interest for signal integrity and data acquisition applications, this type of triggering, in conjunction with an ultra-high-speed oscilloscope (such as the Tektronix TDS6804B with 8 GHz real-time bandwidth), enables them to quickly isolate, capture and analyze events of interest.
The term ?Pinpoint? summarizes the precision with which the trigger system can isolate individual events or combinations of events. The system?s timing specifications are joined by features that increase the user?s choices in defining what to capture and when.
Pinpoint triggering provides ultra-low trigger jitter (<1.5 ps) and can capture glitches less than 200 ps wide for detecting events of interest with very high accuracy. The high-speed trigger path ensures that narrow events and fast edges will cause a trigger and an acquisition, so that the user does not miss transient events.
In addition to the A-trigger, some advanced oscilloscopes give the user flexibility to use B-triggering to capture an event of interest after a specified delay time or number of events. However, in these systems the B-event can only be qualified by an edge trigger, and the oscilloscope then has to wait indefinitely for the B-event to occur. With the dual triggering offered by Pinpoint triggering, both A- and B-events can be fully qualified with a suite of trigger types including edge, glitch, width, timeout, runt, transition, window, and set-up/hold. Pinpoint triggering not only provides high performance and a broad selection of trigger types: it also offers the most flexible range of choices, qualifiers and parameters available in any oscilloscope.
A designer can now use the B-trigger to look for a suspected transient which might, for example, occur hundreds of nanoseconds after an A-trigger has defined the beginning of an operational cycle. And, because the B-trigger offers the full range of triggering choices, the serial design engineer can actually specify, for instance, the pulse width of the transient he or she wants to find. In addition to the more than 1,400 possible trigger combinations that can now be qualified, the system also allows sequences to include a separate horizontal delay after the A-trigger event to position the acquisition window in time.
The reset trigger function makes B-triggering more efficient. If the B-event fails to occur, the oscilloscope, rather than waiting endlessly, resets the trigger after a specified time or number of cycles. In so doing, it re-arms the A-trigger to look for a new A-event, sparing the user the need to monitor and manually reset the instrument.
The timing diagram in figure 1 shows a common application in the disk-drive industry, which could also apply to other digital debug applications where the trigger system has to ignore portions of the waveform. In this example, the need is to identify data defects only when the read gate signal of the drive is high. In this case, channel 2 is connected to the read gate signal and channel 4 is observing the data being read. Thus, what is needed is a trigger that ignores the data signal when channel 2 is low and triggers on channel 4 if there are too many pulses in the data. Previous trigger systems do not allow the system to stop looking for a B-event: they will simply trigger on the next B-event that comes along or wait indefinitely until the next even comes along.
Pinpoint triggering adds this trigger reset feature to the A?ÃœB sequence system, directing the instrument to stop waiting for a B-event when a certain reset criterion is met.
The addition of reset triggering and dual A- and B-event triggering has opened up many new triggering possibilities: for example, in validating serial bus designs. High-speed serial buses are finding their way into many of today?s designs across all industries. Standards such as PCI Express, XAUI, InfiniBand and Serial ATA (SATA) transmit data and clock using a single differential pair for a transmission line. The clock is embedded in the data using NRZ (non-return-to-zero) signaling and 8B/10B encoding. Pinpoint triggering is useful in performing validation and compliance measurements on these serial buses.
Another important application is in trigger-qualified jitter analysis. A primary concern in good serial transmission is that it is transmitted with a BER (bit error ratio) of less than 10-12 bits. Serial validation measurements on timing, amplitude and jitter can be made using serial data compliance and analysis software such as RT-Eye. The RT-Eye software and the Tektronix TDSJIT3 advanced jitter analysis software use the spectrum approach to jitter measurements to provide an estimate of total jitter at 10-12 BER. While accurate and repeatable, this approach requires that at least 100 repeats of a jitter test pattern appear in the acquisition. Unfortunately, when in a system, skip ordered sets (SOS) are added to the data being transmitted. If a jitter analysis is attempted on a signal that contains an SOS, the software cannot complete the analysis and gives an error message indicating that a repeating pattern is not present.
While the serial trigger does not support 2.5 Gbit/s, using Pinpoint triggering the user can be creative and use 1.25 Gbit/s where the data is a known pattern of 11, 00, or xx combinations. Each single bit defined at 1.25 Gbit/s is two bits at 2.5 Gbit/s. The serial trigger can be set to 1.25 Gbit/s with a combination of ?0 11×0 011x? to trigger on the K28.0 character in the SOS. Once the SOS is isolated by the trigger, the horizontal position can be positioned to 0%. Figure 2 shows the trigger setup and capture of the signal, while figure 3 shows the result of the jitter compliance test after the position has been adjusted to 0%.
Record Length, Triggering
It is often thought that record length can be used as a substitute for a good trigger system. The answer becomes very clear when one considers how much real time is actually captured ? even with a very fast long record length oscilloscope. For example, the Tektronix TDS6000B captures and displays a long record of 32 megasamples every 0.75 sec. The amount of data captured in each acquisition is 1.6 ms at full sample rate. Thus the amount of real time being captured is approximately 0.2%. Even if the post-processing software is infinitely fast, the user will only see the event if it happens to be captured. If the event is random, this is highly unlikely. Using long record length with post-
processing software to capture the event is like taking several short videos and then searching through the videos frame by frame for the point of interest. This process is not very efficient and is very computing-intensive. Hence, although both tools are very important, record length is not a substitute for triggering.
Ever since Howard Vollum introduced the first Tektronix triggering sweep oscilloscope, triggering has been transforming the oscilloscope from an instrument that provides vague impressionistic results into a critical tool for capturing random events or qualifying data for analysis. Signaling speeds in modern computer, data-communications and telecommunications devices need more advanced trigger systems because of their high speeds and complex nature. The Pinpoint trigger system keeps up with the fastest and most complex signals using silicon-germanium trigger chips and new trigger features such as dual A- and B-event triggering, window triggering, logic qualification and reset triggering.
Chris Loberg is design and manufacturing market segment manager with Tektronix Inc.