Proclamation
Be it known that the Fifty Ohm Party (FOP) has mandated all voltage signals to have 50-Ω source impedances. Direct connections to test instruments are to be made via coax cables. Scope probes have become historical curiosities and are relegated to museums. The only exception is 20:1 low-impedance probes, which require an FOP license.
When waveforms look odd, the scope probe or its misuse may be the cause.
Until the FOP rules, we’re stuck using voltage probes to bridge the gap between signals and scope inputs. It would be difficult enough if the gap were only electrical, but probes also must span the several feet that may separate signal and instrument. And, components on the equipment under test (EUT) may be inaccessible even for the smallest probe body and tip. Of course, the goal is to connect the signal to the instrument without loading the source or introducing distortion.
All measurements are made between two points, the signal and a reference. For the majority of voltage probes, the reference is ground: the probe and scope form a single-ended measuring system. More general-purpose differential probes can measure between two nonground signals as well as between a signal and ground.
Matched pairs of probes can be used with two separate scope channels to provide a differential capability at relatively low frequencies. The gain and frequency response of the pair of probes/channels must be adjusted to maximize common-mode rejection. The channel amplifier outputs are subtracted in the scope and the difference displayed. However, a large common-mode signal will reduce the dynamic range available for the differential signal of interest.
Differential probing up to 100 MHz requires much better matching than can be accomplished using separate scope channels. An external differential amplifier such as LeCroy’s Model DA1855A together with the company’s DXC100A matched pair of probes may be a good solution. An amplifier provides high common-mode rejection, although achieving much greater than 20 dB at 100 MHz is difficult.
For the best high-frequency performance, the probe length must be minimized and the probe/amplifier combination integrated to give predictable and well-controlled behavior. Examples include the miniature Tektronix P6246 with 400-MHz bandwidth and the company’s subminiature 3.5-GHz P7330.
Differential or single-ended, either type of scope measurement relies on probes to maintain signal fidelity. By considering the advantages and limitations of the main types of probes, the best choice can be made for a particular application.
High-Impedance Probes
For signals containing frequencies below a few hundred megahertz, conventional 10:1 high-impedance probes work well. To a first approximation, if the time constant of the probe’s RC network matches that of the scope’s RC input, including about 50 pF of cable capacitance, the signal will be attenuated but otherwise undistorted. Source loading is assumed to be small because of the typical 10-MΩ probe input impedance.
Unfortunately, as shown in Figure 1, the input impedance of a 10-MΩ 10:1 probe with a 12-pF input capacitance already has decreased by 30% at only 1.6 kHz. The source impedance loading caused by probes becomes a serious issue at high frequencies. For example, a 10-pF capacitor has a reactance of 16 Ω at 1 GHz. Passive 10:1 probes are available with a 500-MHz bandwidth, but this is measured when driven from a 25-Ω source impedance—a doubly terminated 50-Ω signal generator.
One way to reduce input capacitance is to use a 100:1 probe instead of the more popular 10:1 style. For example, the Tektronix P5100 100:1 probe has an input impedance of 10 MΩ in parallel with only 2.75 pF. Passive 10:1 probes typically have an input capacitance from 8 pF to 14 pF depending on the model. Of course, the price of saving several picofarads of input capacitance is a signal that’s 10 times smaller than if a 10:1 probe were used.
As the input signal frequency increases, the nonideal behavior of the probe cable becomes a problem. It looks like a piece of ordinary coaxial cable but, in fact, contains a resistive Nichrome center conductor. Because the cable can’t be properly terminated by the scope’s input impedance, the damping provided by the additional resistance is beneficial.
At high frequencies, lack of a well-matched termination means that one must be approximated. This is the role of the additional R and C trimmers in a high-frequency passive probe tailbox, the small molded box on which the probe BNC connector is mounted. Typically, labels or a slide-off cover hide these adjustments from the user, allowing access to only the low-frequency compensation capacitor.
If you have a suitable signal source and reasonable patience, you may be able to improve a probe’s high-frequency response. Alternatively, if you are using a high-frequency probe with a scope to which it has not been matched, unless you do adjust the probe, your results may be misleading.
Even with a correctly compensated probe, attaching the probe to the EUT is problematical. Some circuit nodes simply can’t be reached except by adding a couple of inches of wire to the end of a probe. Two inches of wire add significant inductance that will resonate with the probe input capacitance.
You can reduce the sharpness of the resonance by using a small resistor rather than plain wire to connect to the circuit. Damping the resonance allows you to make meaningful signal measurements but at reduced bandwidth. Without damping, the ringing introduced by the extension wire may mask the actual signal behavior.
Figure 1 shows such a damped resonance resulting from adding 10 nH in series with 50 Ω between the probe tip and the signal source. Although the simplified probe model cannot accurately describe the behavior of a real probe at high frequencies, a resonance will occur. Its location depends on the series L and R values and the probe input C, but the probe cable and tailbox components will alter its effect on the probe output. The overall result is reduced bandwidth.
Low-Impedance Probes
If you’re working with logic signals rather than very sensitive high source-impedance analog signals, a 10-MΩ input resistance is unimportant. By using a properly terminated 50-Ω coaxial cable as one part of a resistive divider, exceptionally well-behaved, high-frequency probes can be made. Typically, the input resistor is 950 Ω to give a 1-kΩ input impedance and 20:1 attenuation. This type of probe exhibits only very small input capacitance, so loading effects are much less than those of a 10-MΩ 10:1 probe at high frequencies.
Examples of such probes include the Agilent Technologies 54006A 6-GHz passive divider probe with a 0.25-pF input capacitance. LeCroy has a similar low-impedance probe, the Model PP066, specified to have <0.2-pF input C and a 7.5-GHz bandwidth. The Tektronix P6150 probe uses a very small SMA-compatible screw-in tip to achieve 0.15-pF input C and a 9-GHz bandwidth.
Active Probes
The earliest devices, called FET-input probes, were very sensitive to mishandling and easily damaged. Newer generations of active probes have improved ruggedness, frequency response, and size. Active probes achieve low input capacitance by reducing the dimensions of the probe tip and by using low capacitance parts internally.
Dr. Michael Lauterbach, LeCroy’s director of product management, said, “The IEC safety specification, EN 61010, requires a probe with a small distance between the ground and signal inputs to be rated for only 42-V maximum input. However, the front end of LeCroy’s HFP series of active probes actually can withstand several hundred volts. Also, with regard to ESD protection, transient absorbers with less than 0.1 pF of capacitance have replaced older spark-gap technology with about 0.7 pF.”
The present 4-GHz to 6-GHz bandwidth of this type of probe appears to be near the highest practical value. One reason is that the very small dimensions of some multigigahertz probes make them awkward to use for hand-held troubleshooting over extended periods of time. More importantly, the manufacturer’s bandwidth specifications are for the bare probe; they do not include degradation caused by any clips or accessories necessary to attach to the point being probed.
In the multigigahertz frequency range, short wires used to access EUT nodes are fatal to bandwidth. They represent an undefined distributed L-C network that may cause ringing on fast signal edges but certainly will reduce bandwidth. The loss can be as high as 50%.
A new approach to solving the high-frequency probing problem has been introduced by Agilent in the InfiniiMax range of active probes. Figure 2 shows the deliberate inclusion of approximately 10 cm of small-diameter coaxial cable between the probe tip and the probe body containing the amplifier. The 25-kΩ tip resistance in parallel with 0.2 pF forms a divider with the terminated 50-Ω input cable. The amplifier drives the cable to the scope but also compensates for the characteristics of the input network.
This probe architecture separates the very low capacitance tip from the body, which can be made larger to better fit the user’s hand. More importantly, because the behavior of the 10-cm cable and input R-C network is well understood, the overall probe can be designed to provide a guaranteed bandwidth in excess of 7 GHz at the probe tip.
The probe’s lead designer, Mike McTigue, commented that as a scope user, he had been frustrated in the past by the difficulty of probing at high frequencies: “[I felt] there had to be a better way to get connected to your circuit. The circuit topology of the InfiniiMax probe is not new, being based on earlier work by Agilent. But, we’ve combined resistive tip technology, better modeling software, and microwave components to make it a practical reality.”
Three bandwidths are available for the probe body: 7, 5, and 3.5 GHz. You then need to choose either the differential or single-ended connectivity kits depending on your application. Each connectivity kit includes a small browser with an ergonomic sleeve, a socketed probe head, and a solder-in probe head. The performance of all probe heads is fully characterized.
In a complementary development, Tektronix has concentrated on improving performance of the connection between the company’s active probe output cables and scope inputs. TekConnect technology has replaced the familiar but bandwidth-challenged BNC connector with a 22-GHz blind-mate BMA style. The new system eliminates the ambiguous changes in signal appearance that can result from BNC pin-to-socket realignment as the probe cable is moved.
Conclusion
Choosing the best probe for the application at hand is not a difficult process. The first step is to reconcile the actual signal characteristics, the capabilities of your scope, and your expectations. Is the scope bandwidth sufficient? Does it have enough channels to simultaneously present time-related events occurring at different points in your circuit? Are the signal levels so high that safety is your first concern?
When these questions are answered, selection becomes a matter of matching probe loading to the signal source impedance. The decision also is affected by how precisely you must measure voltage levels and event timing, what probes you have available, and the size of your new probe budget.
Low-cost, relatively low-bandwidth scopes often are supplied with switchable 1:1/10:1 high-impedance probes. For the types of signals that you can view using a low-cost scope, this type of probe is appropriate. However, if you will be examining 500-MHz ringing on a switching signal, you will need a better scope anyway.
To pursue the 500-MHz ringing further, a 500-MHz bandwidth 10:1 passive probe and a low-impedance 20:1 probe are the first types to consider from a cost point of view. You may be able to capacitively couple your signal to a low-impedance probe if you only want to examine the ringing.
The series coupling C ensures that your circuit isn’t loaded at low frequencies, but you can choose a capacitance value that passes frequencies above 200 MHz with little attenuation. The overall signal certainly will be distorted but not the 500-MHz ringing, assuming the circuit can drive a 1-kΩ load.
This may be a better solution than the 10:1 probe because at 500 MHz the probe’s typical 10-pF input C has a reactance of only 32 Ω. Also, to view the overall switching signal and the ringing with low loading and distortion, you need a probe and a scope with at least a 1-GHz bandwidth. If you use a 500-MHz measuring system on a 500-MHz signal, the amplitude will be in error by 30% even if the probe is perfect. But, if you only are interested in the frequency of the ringing, a 500-MHz bandwidth may be good enough.
If you choose an active probe to reduce resistive loading, remember that it has limited dynamic range, although some offer a relatively large DC offset capability. So, if the ringing you wish to observe is on top of a 10-V pulse, you can offset the pulse by 10 V to bring the ringing to the center of the probe’s dynamic range. On the other hand, the 300-V and higher signals often encountered in a switching power supply cannot be viewed directly with the usual types of high-speed active probes.
For the fast logic and RF signals common in today’s electronic products, active probes are the best choice. The supply voltage for most logic families is 5 V or less, and RF signals generally are small ahead of the actual power amplifier output, within the dynamic range of most probes. Some active probes also provide a high-frequency differential measurement capability.
Unless you’re dealing with power supplies and hundreds of volts, considering the voltage levels likely to be present in a circuit is unimportant when using a robust 10:1 high-impedance probe. This is not the case with low-impedance and active probes. Low-impedance probes are limited to about 20 Vrms, the equivalent to 0.4-Ω dissipation in the probe resistances.
Depending on the particular model, active probes are rated for maximum input voltages from 15 V to 40 V beyond which damage can occur. The input is protected against overvoltage, and as LeCroy’s Dr. Lauterbach commented, the protection can be effective up to a few hundred volts.
So, overvoltage damage may not occur, but you’ve been warned that it could. With the highest-performance active probes costing more than $5,000 each, checking the voltage you’re likely to apply is a step worth remembering.
Acknowledgement
Thanks to Leslie Green of Gould Nicolet Technologies for providing technical input.
FOR MORE INFORMATION
on high-frequency probes from
Agilent Technologies
enter this rsleads URL
www.rsleads.com/301ee-180
on probe basics from LeCroy
enter this rsleads URL
www.rsleads.com/301ee-181
on all types of Tektronix probes
enter this rsleads URL
www.rsleads.com/301ee-182
Return to EE Home Page
Published by EE-Evaluation Engineering
All contents © 2003 Nelson Publishing Inc.
No reprint, distribution, or reuse in any medium is permitted
without the express written consent of the publisher.
January 2003