No matter how packed an embedded system design is, no matter how dense the board layouts are, and no matter how difficult it is to access signals, one thing is for sure: You’re going to have to scope things.
Today’s oscilloscopes continue to improve in terms of their bandwidth, memory depth, built-in analysis capabilities, and abililty to test in multiple domains. But it doesn’t matter how capable your scope is if you can’t get a probe on the signal you need to see, or if the probe can’t use all that bandwidth.
Fortunately, test-equipment vendors put just about as much effort into probing technology as the scopes themselves. As a result, probe technology is improving along with the scopes.
There’s No Ideal Probe
An “ideal” probe is nice to dream about. But in reality, no probe is ideal. Even a simple length of wire is still potentially a very complex circuit. For dc signals, a probe appears as a simple conductor pair with some series resistance and a terminating resistance. However, for ac signals, the picture changes dramatically as signal frequencies increase.
Any length of wire has distributed inductance (L), and any wire pair has distributed capacitance (C). The distributed inductance reacts to ac signals by increasingly impeding ac current flow as signal frequency increases. The distributed capacitance reacts to ac signals with decreasing impedance to ac current flow as signal frequency increases.
The interaction of these reactive elements (L and C), along with resistive elements (R), produces a total probe impedance that varies with signal frequency. Through good probe design, the R, L, and C elements of a probe can be controlled to provide desired degrees of signal fidelity, attenuation, and source loading over specified frequency ranges.
Given that the ideal doesn’t exist, what will designers accept from their scope probes? According to customer research done by Tektronix, there are three themes that dominate:
- Designers don’t want to see their probe impact the circuit’s behavior.
- Given the form factors, package styles, and tight pin pitches of today’s chips, designers need smaller profile attachments, leads, and solder-in probe tips.
- Designers need longer ground leads without distorting signals.
If anyone needed convincing about the effects that scope probes can have on a signal, consider this example of a logic gate (Fig. 1). The measurement at the gate’s output (VOut) would be a square wave, of course. Adding a second probe to the input would increase the capacitance at C1.
If that second probe is a low-capacitance probe, the added loading brings the gate’s switching speed from about 77 MHz down to 58 MHz. A standard probe degrades the situation further, dropping the switching rate to 33 MHz. Given that a Schmitt trigger inverter gate like this is a basic building block of digital design, one can see how much probe loading can adversely affect circuit behavior.
For general-purpose probing, the scope makers have had to examine their offerings with an eye toward satisfying the above criteria. The answer, it seems, is a low-capacitance passive probe that improves on bandwidth while reducing the loading effects on the circuit under test.
Ringing on signals is a product of inductance and capacitance. If a probe’s input capacitance is minimized as much as possible, then the ground leads, which add inductance, can be lengthened somewhat.
One example of how this is approached is found in Tektronix’s TPP0500 (500 MHz) and TPP1000 (1 GHz) passive voltage probes (Fig. 2). Whereas most standard passive probes exhibit about 8 to 9 pF of capacitive loading, these probes present only around 4 pF of loading, which, according to Tektronix technical marketing manager Randy White, does not affect the flatness of signal traces. “It only impacts the leading edge where all the high-frequency signal content is,” says White.
With the lower input capacitance, engineers can use longer 3-in. ground leads with these probes while incurring no penalty due to the added lead inductance. Thus, they find that not only can they browse digital signals, but also higher-voltage supply rails, or signals that would usually be browsed with a high-speed differential or active probe.
Almost every embedded system has some variety of wireless capability built into it, be it Bluetooth, Wi-Fi, ZigBee, or a cellular radio. It’s a trend that’s accelerating rapidly, so design teams must be prepared for a great deal of mixed-signal debugging. Moreover, the availability of low-cost, off-the-shelf RF modules for things like ZigBee means that it’s less likely that you’ll have a distinct RF or analog group to lean on. More and more, this mixed-signal debug is falling into the laps of digital teams.
Traditionally, scopes have been the tool of choice for analog and digital debug, while spectrum analyzers come into play for noise analysis. It’s not uncommon, for example, to have the output of a processor or DSP feeding the signal input of a radio. You’ll want to look at the output of the amplifier stage of that radio with spectrum analysis (Fig. 3). Lately designers have seen the need to look at both the frequency and time domains and achieve correlation between them.
Probing a mixed-domain system isn’t necessarily straightforward, especially when it comes to RF probing. One new twist from Tektronix for its MDO series of mixed-domain scopes is the TPA-N-VPI adapter, which enables active, 50-Ω TekVPI probes to be used on the scopes’ RF input (Fig. 4). As a result, engineers can perform true RF browsing on a board. With the active probe, it’s possible to see RF signals of very low amplitude in the presence of much larger signals.
Another common issue is troubleshooting of electromagnetic interference (EMI) problems on boards. Almost every designer of mixed-signal boards wrestles with EMI at some point, largely due to the fact that it can come from so many different sources, such as nearby components or through radiation from an enclosure slot. A typical approach to EMI issues is two-pronged involving both a scope and a spectrum analyzer.
Now, with the multi-domain capabilities of scopes such as Tektronix’s MDO4000 series, designers gain the ability to trace EMI transients with correlation back to their sources, which may be high-speed serial packets or switching power supplies. Through work with a third-party company, Tektronix offers its 119-4146-00 near-field probe set spanning a range from 100 kHz to 1 GHz. Here again, the TPA-N-VPI adapter comes into play, allowing the wire-loop near-field probe to be connected directly to the MDO4000 scopes’ RF inputs.
Probing Power Signals
Determining the amount of power consumed by a given component is a profoundly fundamental measurement. One simply applies a differential voltage probe for a voltage measurement and, typically, a clamp-on ac-dc current probe for a current measurement. Multiplying the results provides power in watts. But because these probes are based on different technologies and have different characteristic propagation delays, the timing (or “skew”) between them must be removed before measuring power.
Differential probes are recommended for power measurements because many power signals are not referenced to ground and contain high voltages, so standard oscilloscope probes can’t be used. High-voltage differential probes provide a selection of accessories to safely and conveniently measure floating voltages on small and large devices.
Current probes measure current flowing through a conductor by encircling the conductor. To measure the current, ac-dc current probes use a combination of transformers and Hall Effect devices. These current probes are also convenient because the jaws of the probe head can be opened and clamped around the conductor.
What matters most to engineers undertaking power measurements isn’t so much the oscilloscope as the probe itself. Does the probe have enough rated bandwidth and input range? Is it categorized properly in terms of safety for the measurement at hand? Does it have the insulation and mechanical design to facilitate the measurement? These attributes might not be apparent in a datasheet.
Tektronix’s latest high-voltage differential probes represent an interesting approach to the issues of power measurements. The THDP0200 probes use interchangeable tips to provide different attributes depending on the measurement requirements at any given time (Fig. 5). The probes offer a broad bandwidth of 200 MHz and voltage capacity of 1.5 kV.
Also, the probes sport improved safety features, such as longer alligator clips and relatively thick wires to accommodate higher voltages and currents. But that’s not always a plus in terms of measurements, as the increased inductances can hinder some measurements.
Thus, Tek includes the interchangeable tips. There are very long tips for measurements where you need more clearance, or to make measurements inside a test chamber, although you do take a bandwidth hit of some 20 to 30 MHz. The probes also carry dampening resistors to hold down in-band resonance and ringing on high-voltage signals with high slew rates.
Good, robust high-voltage differential probes are becoming important diagnostic tools as gallium-nitride (GaN) power devices begin making their way into the market. GaN power semiconductors switch faster and have a wider bandgap, making them good candidates for high-temperature applications such as power conversion in hybrid vehicles (see “Working With GaN/SiC Power Semis Poses Test Challenges”).
Probe Bandwidths Rising
High-speed active probing technology has improved greatly over the past decade and is continuing to march upward in bandwidth. Agilent Technologies’ InfiniiMax probing system has seen two revisions since its 2002 debut, each bringing with it a bandwidth boost. The current generation, InfiniiMax III, sports bandwidths of up to 30 GHz and is intended to complement Agilent’s high-end Infiniium scopes (Fig. 6).
Within the InfiniiMax III system are three probe amplifiers with bandwidths from 16 to 30 GHz. The previous generation’s amplifier offered a 13-GHz bandwidth. But to achieve the higher bandwidths in the new probes, Agilent had to abandon the old version’s thick-film and printed-circuit board (PCB) technologies.
Within the probe amplifiers, Agilent uses a low-loss, low-k dielectric over a ground plane on a standard alumina-substrate thick-film process. As a result, the probe amps have microstrip lines of smaller geometries to limit dispersion and quasi-coax lines to limit coupling and radiation. This technology replaces the old approach, which used thin films in machined cavities. A 200-GHz indium-phosphide IC process with backside ground vias gives the amplifier circuit extra “oomph.”
In designing the probe heads, Agilent eschewed the usual approach in which the PCB is cut down multiple times to achieve a well-tuned result. The more costly thick-film substrate and close-tolerance components dictated fewer cuts to achieve the desired performance.
To accomplish this, extensive electrical simulation and mechanical solid modeling helped arrive at an optimal design before implementation. As with the Tek probe heads, a damping resistor at the tip was used to eliminate high-frequency (HF) peaking caused by resonance of the probe at high frequencies. The resistor yields flat response across the probe’s bandwidth.
A wide range of heads is available to allow connections to circuits using a zero-insertion-force (ZIF) tip, a browser tip, 2.92-mm or 3.5-mm surface-mount assembly (SMA) cables, or solder-in tips. The browser uses a “criss-cross” blade grounding system for very low-inductance grounding as well as a poly-iron wrap of the coax tips to greatly reduce standing waves.
The resistor tips, which deliver very low parasitics, are replaceable. The tip also sports integral white-LED lighting. The browser achieves a 30-GHz bandwidth with capacitive loading only slightly higher than that of the ZIF probe head.
The ZIF probe head’s tip is soldered to the point of measurement. The tip is attached to the probe head by simply pinching a lever on the head. Probe loading is a scant 35 fF at a bandwidth of 28 GHz.
DDR3 Probing And Logic Analysis
Probing of Double Data-Rate Type 3 (DDR3) SDRAMs comes with its own set of issues. The speed of DDR buses continues to rise even as the I/O voltage falls. Moreover, these memory devices are found in more and more consumer electronics applications, so it’s imperative that their power consumption be held to a minimum.
Data is captured from a DDR3 bus by a logic analyzer in sync to the application. Either a clock or some other signal from the application facilitates data capture at a specific time relative to the system. If the logic analyzer is to accurately recognize the signal, it must be within the data eye opening, or data-valid window, of the analyzer.
To capture DDR signals, the logic analyzer and probing system needs to accept a smaller eye opening. The eye opening for a DDR3 signal is shrinking due to the higher signal speeds and lower voltages.
On top of of signals becoming harder to qualify based on the data-valid window, there is the issue of how to probe the bus. There are three fundamental ways to access signals in a DDR environment.
The first is direct probing, which is facilitated by a board layout that includes probing pads or connectors. Area concerns often rule out this approach. A second approach is to attach DDR3 chips directly to the board using memory-chip interposers. A third is to use dual-inline memory module (DIMM) interposers, as DIMM-packaged DDR3 chips are attached to the interposer, from which signals are available for probing.
Tektronix recently announced a DIMM interposer for DDR3-2133 and DDR3-2400 devices for use with its TLA7000 series logic analyzers. The interposers enable test engineers to reuse their existing TLA7BBx modules to acquire address, control, command, and data signals for all DDR3 speeds as well as for low-voltage DDR3 devices.
The link between the interposer and logic analyzer is through cables soldered directly to the interposer. The DIMM form factor accommodates a 64-bit bus. Depending on whether the bus is X4 or X16, this would mean using from eight to 16 probes to capture from 90 to 100 signals simultaneously.
These interposers are the first to use Tek’s silicon-germanium (SiGe) hybrid ASIC technology to provide a form of equalization similar to what’s used in serial data test. As a result, cable losses at high data rates are accounted for and the size of the eye is maintained.