A steady stream of advances has elevated test and measurement instruments to the point where they can reveal minute details of signals with lightning-quick rise and fall times. So, then, what about test probes? The last thing test engineers want or need is a probe that’s going to influence their measurements or fail to deliver the full bandwidth that’s available to them on the scope.
Fortunately, today’s high-end probes are constructed to sidestep these issues. In this article, you’ll learn about what makes modern active test probes tick and how they can help you pin down (literally as well as figuratively) the signals you need and want to see. There also are considerations related to squeezing the maximum lifespan and performance out of active probes (see “Pick The Right Probe And Get The Most Out Of It”).
THE PERFECT PROBE
An active probe has active components such as transistors, amplifiers or preamplifiers, and sometimes FETs. The basic distinction between an active and passive probe is that the active probe requires a power source applied to it.
In a perfect world, the perfect test probe would serve as a completely electrically transparent interface between the device under test (DUT) and the oscilloscope’s input. It would impart no effect whatsoever on the signal being acquired, accurately reproducing the signal under test with more than adequate fidelity.
Unfortunately, here in the real world, many physical factors come into play to make the perfect test probe practically unattainable. The probe that’s incapable of loading the circuit under test has not been and cannot be built. “All probes will load the circuit,” says Jae-Yong Chang, Agilent Technologies’ product manager for test probes. “The question is how much loading the probe imposes and how much you can tolerate.”
The challenge for test-instrument makers, then, is to come as close to perfect as possible. A number of considerations must be balanced in the creation of high-bandwidth active probes. Many of these considerations are borne of the market drivers behind the probes’ development.
Today’s serial high-speed data standards, such as PCI Express, are among the top challenges for test engineers. “Signals for all of these serial data standards are differential signals,” says Andy Heltborg, product marketing engineer at Tektronix. “Many of them are moving to multiple lanes, faster speeds, and lower voltages. So this is where the market is going, and it creates issues we need to address with probes.”
HOW PROBES AFFECT TESTING
Traditional active test probes have always had the facility to add accessories to the tip of the probe to enhance the actual connection to the DUT. These accessories allow users to more easily make measurements, but they also affect measurement performance in terms of bandwidth, loading, linearity, and flatness of bandwidth (Fig. 1). But once signals get into the section of the probe where its amplifier resides, there’s little or no further effect on these signal parameters.
A closer look at the nature of the connection of an active probe reveals more about its electrical characteristics (Fig. 2). The interconnect itself dominates these characteristics.
“We view it as an LC-resonant tank circuit,” says Chang. “Depending on the length of the cables, input leads, and connectors, you have inductance and capacitance on the front end that make up a resonant circuit. Because of the resonance, you get certain characteristics over the frequency range.”
Input impedance falls to nearly zero at some frequency, which is the LC-resonant frequency. “At that point, the response of the circuit rises at the resonant frequency. This is an inevitable characteristic of the probe,” says Chang. A common approach to correcting this peaky frequency response, which Agilent employs in its high-end InfiniiMax active probes, is to use a damping resistor of about 100 to 150 Ω at the probe tip.
“This resistor prevents the input impedance from going below the damping resistance,” says Chang. The added resistance reduces the loading and peaking of the probe caused by input connection conductors. So while the damping resistance helps to flatten out the probe’s frequency response, it does not influence any increase in the probe’s bandwidth.
Equipment makers can use and have used various design techniques to squeeze a bit more bandwidth out of their scope probes. Historically, the simplest way to improve probe bandwidth has been to make the overall probe as physically small as possible.
As part of this effort, the manufacturers strived to minimize the distance between the probe tip and amplifier, decreasing the overall size of the probe. They sought to increase the amplifier’s bandwidth while simultaneously making the amplifier itself smaller. Finally, they reduced the probe tip’s physical size (and ground wire) while also limiting the type of probe tip to a fixed “browsing” style.
According to Agilent’s Chang, applying all of these design techniques yields a maximum probe bandwidth of 6 GHz “The tradeoff, though, is in usability,” Chang says. “Probes designed in this manner are rendered unusable in many applications because you can barely see where you are probing. If the fixed tip breaks, your probe is ruined. Then there’s the matter of a fixed distance between the probe tip and the ground connection. Circuitry has to be laid out with this constraint in mind, and that’s unfeasible.”
PROBE-TIP SIZE DOES MATTER
Getting the most out of active test probes—or any kind of test probes, for that matter—requires an understanding of the ways in which they attach to the DUT. Differential signaling standards sometimes call for common-mode measurements, which can complicate test setups.
“If you’re making single-ended or common- mode measurements, you have to connect your probes differently,” says Tektronix’s Andy Heltborg. “With features getting smaller, it gets harder to attach probes to signals. That’s often the most time-consuming part of making the measurement. It may take an hour to get things hooked up. Switching between different measurements takes planning.”
There are various ways to approach probing, some of which will affect the design of the DUT’s circuit board. For example, scope makers can provide permanent, designed-in embedded sockets to which a scope probe is attached. “If you want the best signal fidelity you can get and have space on the board for it, we can work with customers to help them design probe tips into the board,” says Heltborg. “This can be a very good test method in early stages of the design for prototyping.”
Another critical means of attaching probes to hard-to-reach circuit points is with solder-in tips. Sometimes the test engineer will just solder down the lead of the damping resistors used to smooth out frequency response. In other cases, a slightly more elegant solution might be found in using solder-in zero-insertionforce (ZIF) tips.
An example of such a connection is Agilent’s N5426A 12-GHz InfiniiMax ZIF tip (Fig. 3). “Some users wanted handsfree probing, but still wanted to be able to browse around,” says Agilent’s Chang. “The ZIF tip provides a hybrid of handsfree, solder-in, and browsing.”
ZIF tips can be soldered into various circuit points that will require frequent retesting; users simply snap the probe head into them when needed. Otherwise, the probe is free for browsing around other areas of the circuit.
EXPLOITING PROBE VERSATILITY
The time-consuming nature of connecting and disconnecting probes to finepitch IC leads has resulted in some creative thinking among test-equipment makers. “This was a key pain point in measuring differential signals,” says Heltborg, “and it’s why we came up with the TriMode probing feature in our high-end probes.”
The TriMode probes, with bandwidths of up to 20 GHz, have different input selections (Fig. 4). Users hook up the tip once, and by switching the input to the probe, they are able to make all singleended, common-mode, and differential measurements without moving the probe connections.
Using a single TriMode probe to make these various measurements can potentially replace two individual probes. “The advantage is the single connection,” says Dave Fink, Tektronix’s oscilloscopes and probes product marketing manager. “The alternative is a dual probing method, in which a common-mode measurement requires two probes. In a densely packed circuit area, this can be very difficult to perform.”
WORKING WITH RF
Today’s test-probe technology can also make it easier for designers having to work with RF signals. Probe tips embedded within circuit boards can facilitate the practice of damping the connection with added resistance, yielding a connection that has minimal parasitics.
These types of SMA embedded probe tips offer benefits above and beyond simply connecting directly to the oscilloscope. “If you are looking at multiple signals, such as an HDMI stream, you may want to see two or three different differential signals at once,” says Heltborg. “These tips can let you look at multiple signals at once with a single scope.”
Furthermore, HDMI signals have a high dc-bias level that’s problematic for going directly into a scope. With the SMA tips, users can set a termination voltage to match this dc bias level. As a result, they won’t load down the probe’s amplifier, and they can still acquire signals.
Embedded probe tips also provide users with improved voltage standing-wave ratio (VSWR) and skew matching. “We can control the probe much better than individual channels of the scope,” says Heltborg. The result is enhanced performance.
GOING THE CUSTOM ROUTE
Another avenue to consider for probe tips is the custom route, which can yield perfectly tailored results for a given application. Custom probes can be the answer for probing extremely fine-pitch IC packages. One vendor of such probes is AlphaTest Corp., whose probes can be constructed as small as 0.25 mm, center-to-center spacing. Probe makers like AlphaTest can deliver custom test fixtures as well as probes with custom contact forces and lengths.
“At least half our customers come to us with concerns about mechanical connections,” says Dan Rogers, AlphaTest’s vice president of marketing. The company’s probes are constructed simply, yet durably, of steel. It builds custom test fixtures into which the pins are inserted. The fixture holes serve as guides for the pins, which are accurate enough to hit a pad that is 5 mils in width.
Because AlphaTest built its own lathes for machining the probe parts, adjustments can be made to coil thicknesses, curves, and pitches. The angle of the coil within the tube is also adjustable, yielding different amounts of contact force and travel.
The vagaries of making a good connection with the circuit under test constitutes just one aspect of getting the most out of active probes. Most of today’s high-end active probes are differential probes that examine serial data streams. These data streams are referenced to each other and not to ground. Such probes come with built-in differential amplifiers that subtract the two signals, resulting in one differential measurement displayed on one scope channel.
For the design of its WaveLink 13- to 20-GHz differential probes, LeCroy Corp. went the non-traditional route. The probes use a differential traveling-wave amplifier architecture that’s commonly seen in ultra-high-frequency broadband amplifiers. This architecture brings a number of benefits, including maximum gain/stage and minimal attenuation.
“One key benefit is that there is no feedback, which always increases noise even as it stabilizes the amp,” says Ken Johnson, LeCroy’s senior product manager for scopes and probes. “As probe bandwidth rises, the expectations for noise performance grow more stringent. We have to make dramatic gains in noise spectrum density to make any headway.”
Too much noise in an eye diagram can cause excess thickness at the top and bottom. “That thickness is related to test equipment, not the user’s signal,” says Johnson. “In such instances, they can no longer discriminate between noise in the signal and noise from the probe and scope.”
The traveling-wave architecture involves a synthetic transmission line through the signal path that yields extremely low probe noise of 25 nV/√Hz. “We strove to make the amplifier more linear with as flat a frequency response as possible out to 20 GHz,” says Johnson. The amplifier comprises a line of active elements (FETs). Each transistor has an associated parasitic capacitance to which inductance is added to create a continuous impedance from stage to stage (Fig. 5).
A critical specification for active probes is ac loading. For the WaveLink probe, LeCroy specifies ac loading as just 175 Ω minimum. Johnson says ac loading is more a function of tip design. “It’s a matter of having a resistor that’s very small and attached as close to the point of contact as possible,” he says. “We have patents on how to do custom compensating resistors. It’s a simple and elegant way to control parasitic capacitance, rather than using excess capacitance to swamp the parasitics.”
Another benefit of this architecture is its ability to minimize the gain needed from each individual stage. “Each stage’s gain is additive to the previous stage, so each one isn’t being pushed too hard. The synthetic transmission line is a trick we play to get the gain stages to add,” says Johnson.