As test requirements have changed, so have the capabilities of oscilloscopes.
BY BRAD WEBER, Tektronix Inc.
Throughout the communications industry, designers and technology re searchers are racing to develop equipment that will deliver ever-increasing amounts of capacity. Just about every component and system level is affected, from metro and access networks to individual optical crossconnects and optical add/drop multiplexers. New 10-Gbit/sec technology will go to the core networks; lower layers, currently running at OC-12 rates (622 Mbits/sec), will receive upgrades to 2.5-Gbit/sec capacity.
Test-equipment technology is on the same fast track. The sampling oscilloscope, the industry-standard tool for waveform acquisition and analysis, has made major strides in recent years. Today's top models are configured as communications signal analyzers, with specific features and performance capabilities to address high-speed optical-network elements. The newest instruments are not only faster, but also better.
One of the fundamental enablers for network capacity advancements is DWDM technology. DWDM over SONET or SDH increases the number of usable channels from one to eight, 32, and eventually 128 and beyond. Interestingly, the total power "budget" for the data transmission remains the same; it is shared among ever-more channels.
Accompanying the DWDM density trend, there is an industry-wide push to maximize the distance a signal can travel via fiber before it must be regenerated by an inline optical amplifier.
These concurrent increases in density and distance can affect designers and researchers trying to characterize system signals and overall performance. Edges are faster, signals are briefer, and a little noise goes a long way-literally.
In other words, progress comes at a price. Signals traveling at 10 Gbits/sec are a challenge for existing test equipment. The best of the sampling oscilloscopes in service today (a generation of tools that has been in use for at least a decade) can be equipped to capture 10-Gbit/sec signals under normal circumstances but might have difficulty with today's DWDM-based signals.
Ask a group of DWDM equipment or component designers what's most important to them, and you'll get a litany of measurement concerns:
- "My signals are very small; sometimes less than -10 dBm."
- "I'm dealing with signals that are embedded in a fairly noisy environment."
- "I'm having trouble getting my 10-Gbit/sec eye-patterns to stabilize on the oscilloscope screen."
- "When I need to extract an embedded clock signal, fooling around with external adapters and fixtures costs me time I can't afford to spend."
Realistically, most users would like a solution that answers all these concerns. Can the sampling oscilloscope rise to the occasion?
From the viewpoint of the equipment designer, researcher, or manufacturer, the emerging generation of sampling oscilloscopes is welcome news. It provides solutions for the near-term and ensures a path to the future. The latest sampling oscilloscopes, like their predecessors, can look forward to a long, useful life.
The era of the uncompromised 10-Gbit/sec optical sampling oscilloscope has arrived. While earlier measurement systems offered the raw bandwidth to capture signals in this range, the new tools are much better matched to today's challenges-thanks to a whole package of improvements. Triggering functionality has received particular emphasis in the new designs.
Trigger stability is critical. In truth, it's impossible to design a trigger circuit that is utterly free of jitter across the whole span of an acquisition; physical and thermal realities just won't allow it. But every attempt must be made to keep jitter to the lowest possible tolerance; the reason becomes clear when signals are measured at various points across the full 125-microsec width of a SONET frame. Trigger jitter is trivial at the beginning of the frame but begins to accumulate later and later in the frame. Unless special measures are taken, trigger jitter at the end of the frame makes the acquisition so uncertain as to be meaningless. An instrument that adds jitter like that is actually "stealing" precious test margins.
Certain new-generation sampling oscilloscopes provide a solution for this nagging problem: They provide two optimized trigger modes. The "short-term" mode is for signal viewing close to the trigger point and holds jitter to the 1-psec range (plus or minus a small tolerance). The "long-term" mode locks the trigger to a highly stable oscillator. Using this mode, it is possible to view edges and eye-patterns all the way to the end of the SONET frame, with a two order-of-magnitude improvement in jitter performance. Dual trigger modes assist designers in viewing signals that have been elusive until now.
Earlier, we mentioned the big industry-wide effort to push optical signals farther before re-amplifying them. The motive here is simple: Inline amps cost money, and it behooves every network to use as few of them as possible. Also, distance equals attenuation. In addition, DWDM technology is cramming more data into a fixed power budget per link. The simple truth is that signals are getting smaller and weaker even as they convey more information (see Figure 1).
The equipment designers' mission is to create receivers that can deal with these tiny signals. They are doing just that; elements that can examine signals down to the -10 dBm (and lower) are coming to market. From the test and measurement standpoint, the problem is not one of detecting or amplifying such a small signal; the challenge is to distinguish it from its surrounding envelope of noise-while being careful not to add even more noise.
The obvious solution is to keep measurement system noise to an absolute minimum, of course. New-generation tools accomplish that by using the latest low-noise optical-component technologies behind their input connectors. Typical RMS noise levels have dropped by a factor of two (compared to previous-generation equipment).
In addition, some sampling oscilloscopes have ways to improve upon even this exceptional level of performance. They use special measurement modes that take many readings at a particular point in time, then average the figures. An example of an averaged measurement is shown in Figure 2.
Because the noise on the signal is random in nature, its effect on the validity of any single measurement point tends to cancel over time. By averaging many acquisitions of the same point, it is possible to reduce noise to almost insignificant levels. Of course, the repeatability of this kind of measurement depends, once again, on the stability of the instrument's trigger and time base.
Optically transmitted signals have embedded clocks. No problem-receiver circuits in the field extract optical clock signals all day long. But the engineer trying to extract that same clock is in a different situation. In the past, the task has called for complex external hardware and cabling plus, of course, the time spent setting it all up. It's an issue that cries out for an integrated solution.
Other test-by-test configuration requirements are similarly cumbersome. A typical setup procedure might include changing external filters, cabling together external clock recovery units, and swapping out optical sampling heads to accommodate various standards and data rates. In some instances, the system must be recalibrated after any element is added or removed. Obviously, this process stands in the way of today's urgent time-to-market deadlines.
A new generation of integrated solutions has come to the rescue. Sampling oscilloscopes now bring virtually all of these (formerly external) features into the mainframe or the sampling modules (see Figure 3). The benefits can hardly be overstated; accuracy and repeatability improve, recalibration is unnecessary, and it is a simple matter to view filtered or unfiltered signals at various data rates without changing any part of the hardware. These characteristics equate to a major boost in throughput, whether the user is an individual engineer with eight hours' work to get done before lunch or workers on the production line with a responsibility for shipping product in volume.
Any self-respecting optical-communications equipment designer knows how to set up and measure the characteristics of a fiber-optic signal. But who wouldn't benefit by having some of the key measurements automated? Modern "smart" sampling oscilloscopes simplify device characterization and conformance testing by conducting complex measurement and analysis steps under internal software control:
- Extinction ratio (ER) is a measure of the modulation depth of a source transmitter. ER expresses the proportional relationship between the power levels of the binary "1" and "0" signals, averaged while transmitting a pseudo-random binary-sequence signal filtered at 75% of the maximum bit rate. ER measurements are notoriously hard to make-an ideal candidate for automation.
- Eye-pattern mask testing encompasses rise time, fall time, overshoot, undershoot, ringing, noise, and jitter in one measurement. Like the extinction-ratio test, mask tests are performed with a sampling oscilloscope plus a reference receiver set to filter at 75% of the bit rate.
As explained earlier, many different rates, from OC-12 to OC-192, will co-exist in next-generation networks. Because automated testing is software controlled, it adapts readily to emerging standards and diverse data rates.
As we have seen, the latest sampling oscilloscopes are optimized to match the requirements of emerging generations of high-speed telecommunications equipment. They are designed to help engineers characterize the performance envelope of new devices with higher data rates and smaller signals than ever before. Moreover, their modular, integrated nature and software-driven automation provide a solid foundation for future growth to meet communications measurement needs.
But sampling oscilloscopes aren't the only tools that have moved ahead with the users they serve. Communications test capability can be found in oscilloscopes at every performance and price level these days. Mask-testing features are available in inexpensive portable instruments as well as high-performance, real-time oscilloscopes.
General-purpose lab oscilloscopes now offer bandwidths up to 4 GHz, with sampling rates to 10 Gsamples/sec. Virtually all of these instruments offer automated communications signal measurements and high-frequency probing accessories, and many can be fitted with reference receivers and other communications-oriented accessories.
With regard to measurement tools, the prospects for the communications-equipment designer and researcher have never been better. Although DWDM technology presents a host of new measurement problems, there is a new generation of sampling oscilloscopes that addresses these problems point-by-point. New capabilities range from optimized trigger modes to automated measurements aimed at DWDM conformance testing requirements, and sampling oscilloscope platforms have the "hooks" to add new features as the need arises.
Clearly, the oscilloscope will remain an enabling tool in the development of new communication technologies for years to come.
Brad Weber is a product marketing manager for sampling oscilloscopes with Tektronix's Instrumentation Business Unit. Tektronix Inc. is based in Beaverton, OR.