Oscilloscope Probe Selection and Use
by Dean Huster
Choosing oscilloscope probes is not necessarily an easy task. There are many parameters to consider: What kind of input connectors are on the oscilloscope? What is the scope bandwidth and/or risetime? What is the input resistance of the scope? What is the input capacitance of the scope? Do I need any probe attenuation? What frequencies am I going to measure? What is the impedance of the circuit under test?
System Bandwidth
If you have a scope with a 50 MHz bandwidth, you don't necessarily want to choose a probe with a 50 MHz bandwidth. Anytime you add another instrument (a probe IS an instrument) into the signal path of the instrument used for the measurement, you affect the overall system bandwidth. In fact, you will ALWAYS lower your system bandwidth. Choosing the proper probe will minimize this reduction in bandwidth.
The risetime of a system is directly related to bandwidth. The standard equation for this is:
rt = 350/BW,
where risetime (rt) is in nanoseconds and bandwidth (BW) is in megahertz
If you have a175 MHz oscilloscope mainframe with a risetime of 2ns and a 175 MHz plug-in with a risetime of 2ns, the system risetime (input, through the preamplifier plug-in, through the mainframe circuitry to CRT) will be the square root of the sum of the squares of the individual risetimes. So the square root of (22 + 22) = the square root of 8 or 2.8ns. A 2.8ns risetime corresponds to a 125 MHz bandwidth. This means that a 175 MHz plug-in installed in a 175 MHz mainframe will not have a system bandwidth of 175 MHz, but will be 350/2.8 or 125 MHz! If you add a probe, it gets even worse.
Let's assume you have a 100 MHz mainframe, a 250 MHz plug-in and a 200 MHz probe. The risetimes of those three will be 3.5ns, 1.4ns and 1.75ns respectively. This calculates to a system risetime of the square root of 3.5² + 1.4² + 1.75² or 4.15ns which calculates to a bandwidth of about 84MHz. Although the lowest bandwidth element of the system is the 100 MHz mainframe, the overall bandwidth is even worse at 84 MHz.
If you're using a portable scope with a 50 MHz bandwidth (you don't have to worry about a plug-in with a portable), you'll have a system bandwidth of 35 MHz with a 50 MHz probe, 45 MHz with a 100 MHz probe and 48.5 MHz with a 200 MHz probe. If you can outfit your scope with a probe with a bandwidth of at least twice the scope bandwidth, you'll be doing pretty good with system bandwidth.
Don't forget that the better brands of scopes (Tektronix, Hewlett-Packard/Agilent) are pretty lax in their bandwidth specifications. A 100 MHz Tektronix 465 can often have an actual bandwidth of around 130 MHz. If it actually is 130 MHz, a 200 MHz probe will give you a system bandwidth of 109 MHz, better than the scope's catalog bandwidth specification!
For a "fast" system bandwidth calculation, the equation is "the inverse of the square root of the sums of the inverse bandwidths". This method eliminates the frequent bandwidth-risetime calculations. Here's how it works:
If you have a 500 MHz mainframe, a 250 MHz plug-in and a 350 MHz probe, here's the calculator steps for system bandwidth:
500 [1/X] [X²] + 250 [1/X] [X²] + 350 [1/X] [X²] = [1/X] [SQR X]
which should get you about 188 MHz.
The bottom line with bandwidth is this: get the highest-bandwidth probe that you can if you're interested in getting all the bandwidth possible from your oscilloscope system.
Probe Compensation
Yes, probe compensation is important. A feature of all attenuator probes (not non-attenuating 1x probes) is that they have a compensation adjustment to match the probe's attenuating capacitance to the input capacitance of the scope. A properly made adjustment will keep the system's frequency response flat throughout the system bandwidth. An improperly compensated probe will cause an erroneous fall in amplitude or increase in amplitude as the frequency is increased. This could cause a scope with a normal ±2% amplitude error specification to actually be in error by ±5, 10, 15 or 20%!
Poor compensation is often noticed with poor waveshape on square waves having a frequency in the 100 Hz to 2 kHz area. Unfortunately, most non-rectangular waveforms will not show any problems at all because only the amplitude is affected, not the waveshape.
The best way to compensate a probe is to simply use the probe compensation signal available on the front panel of every scope. I say "every" here because you don't have much of a worthwhile scope if it doesn't have that signal available. Most lab-grade scopes have it available in several amplitudes and at a fairly accurate frequency. It's usually a square wave of around 1 kHz and the probe is adjusted (see probe manual) for the "flattest-topped" square wave possible.
Selecting a new probe for your scope must include the determination that the range of capacitance to which it can compensate will include the input capacitance of your scope. Beware of older and lower-bandwidth scopes. They sometimes have higher input capacitances. I've seen as high as 33pF and 47pF and many probes, especially high-bandwidth probes in the 200- and 300 MHz area, will only compensate between 10pF and 30pF. 20pF has become pretty much the standard input capacitance for most oscilloscopes.
Vertical amplifier input resistance is also very important with regard to compensation. The probe is designed to compensate such that the capacitive reactance (Xc) of the probe to the Xc of the vertical amplifier is the same ratio as that of the attenuation resistance (9M ohms for a 10X probe) to the input resistance of the vertical amplifier. 1M ohm is the standard high-impedance input resistance of any industrial-grade scope made since about 1955 or so. There are some older service-grade scopes made for the TV industry that may have an input resistance that is different, such as 3.3M ohms or 4.7M ohms. Not only will it be difficult to compensate a modern probe to an older scope such as this, but the attenuation factor of the probe will be wrong as the probe depends upon the input resistance/reactance of the scope to be 1M ohm for accurate attenuation. If the scope's input resistance is higher than 1M ohm, the attenuation factor of a 10X probe will be much less than 10X.
Passive Attenuating vs. Non-Attenuating Probes
Non-attenuating probes are available as passive probes (relatively inexpensive) and as active FET probes (get ready to spend about a thousand bucks). The FET probe will give you very high bandwidth, low circuit loading, easy volts/DIV calculation, less money to spend on other equipment, a probe that is limited in maximum voltage that can be measured and a probe that is very easy to destroy when connected to the wrong circuit. From now on, we'll only be discussing passive probes.
Passive non-attenuating probes are much more robust over an active probe. The advantage of a non-attenuating passive probe over an attenuating probe is that it allows you to have the maximum vertical sensitivity available from the scope. If your scope can go all the way down to 2mV/DIV, a non-attenuating (1X) probe will provide you with that sensitivity at the probe tip. If you have a basic scope, another advantage is the fact that the attenuator setting is the sensitivity of the scope at the probe tip. There're no calculations involved, nothing to forget to do. A third advantage is that you don't ever have to remember to check the probe compensation. There's no adjustment for that. That's the end of the advantages list.
The disadvantages of the 1X passive probe is that it has the lowest bandwidth of all probes. It also puts the input impedance of the scope right at the end of the probe where it can load down high-impedance circuits and alter the actual voltage giving you a significant measurement error. The third disadvantage of a 1X passive probe is that it presents a very high capacitance and therefore a very low XC to the circuit under test, an even more significant loading factor when you're working with AC signals.
The highest-bandwidth 1X probe that I've ever worked with is the Tektronix P6101 in a 1-meter length. It has an 11 MHz bandwidth. Regardless, it still has horrible loading factors to the circuit under test and is useful only on lower-impedance, lower-frequency measurement situations.
An attenuator probe works by putting a 9M ohm resistor in series with the scope's 1M ohm input resistance. A capacitor is placed in parallel with this 9M ohm resistance to compensate for the fact that at high frequencies, the capacitance and Xc of the resistor will begin to alter its effective value and therefore the attenuation ratio of the probe.
Let's look at the scope input itself and see how it loads down a circuit. With a direct current input, all the signal will see is the 1M ohm resistance of the scope's input. The 20pF capacitance (a pretty standard input capacitance for most oscilloscopes made since about 1970) that shunts (is in parallel with) this resistance appears as an open circuit or infinite resistance/reactance.
With an AC signal of say 1 MHz, things begin to change. That same 20pF begins to have a reactance in parallel with that 1M ohm. This reactance will have a value of just a little under 8K ohms! At only 1 MHz, the scope will present a load of less than 8K ohms to the circuit under test. At 100 MHz, this reactance will be less than 80 ohms and at 500 MHz, it will be less than 16 ohms! That's a darned heavy load to any circuit, let alone a high-impedance circuit.
A typical 10X passive attenuator probe is basically a 9M ohm resistor in parallel with maybe 2.2pF of variable capacitance (the variable nature is the probe compensation adjustment). While the 9M ohms in series with the scope's 1M ohm provides the 10X attenuation factor (at a 5mV/DIV attenuator setting, the scopes actual sensitivity will be 50mV/DIV at the probe tip), the 2.2pF in series with the scopes 20pF input capacitance takes over to keep the attenuation ratio correct at higher frequencies. In fact, the Xc of the probe/scope combination "swamps" the (1M ohm/9M ohm) resistance combination. At 1 MHz, the Xc of the scope and probe is about 8K ohms and 72.3K ohms for a total of about 80K ohms instead of the scope's 8K ohms. That's a factor of 10 less circuit loading.
If you want even less loading, use a 100X probe for more like 800K ohms of loading or even a 1000X probe for more like 8M ohms load. By the way, a 1000X probe is NORMALLY used to measure high voltages rather than to lessen circuit loading, but the lower circuit loading can be used to an advantage in otherwise normal measurement situations of lower voltages. However, the typical 1000X probe costs three or four times what a 10X probe costs, is the size of an extra-large turkey baster and often requires a filling of Freon for its operation.
Also, don't forget that if you're using a 100X probe for lower circuit loading, the scope's 5mV/DIV attenuator setting now equals 0.5V/DIV at the probe tip; a 1000X probe means a 5V/DIV sensitivity at the probe tip. So there's a definite down-side to using attenuator probes, especially the 100X and 1000X models.
The 10X attenuator probes are still the handiest to use and that's why most scopes come with a pair as the supplied accessories.
Switchable Attenuation Probes
There are dual attenuation probes available from both Tektronix and as imports. The attenuation ratios are 1X and 10X, giving you the better parts of both worlds in one probe. The cost of the Tek version will take your breath away, but then, they're high-bandwidth probes. A switchable probe has one major disadvantage: the 1X position has horrible bandwidth compared to a dedicated 1X probe. For instance, the older Tek P6062B has something like 200 MHz bandwidth in the 10X position, but only 1 MHz at the 1X setting. Except for the convenience, you're really better off having two separate probes rather than a switchable model. You can buy two separate Tek probes for the cost of a Tek switchable.
Remember that you'll be using a 10X probe for most applications to reduce circuit loading, so the ability to immediately switch to a 1X attenuation ratio isn't that big of a deal as opposed to simply swapping out probes.
Probe Voltage Rating
There is a maximum voltage that a probe can handle. As the attenuation factor increases, typically the voltage rating increases. A 10X probe has a higher voltage rating than a 1X probe; a 100X probe can handle higher voltages than a 10X probe. The old P6006 10X probe could handle around 600v while the P6007 100X probe could handle more like 1,100v. The 1000X probes are designed for the tens of thousand volts, hence their large dimensions. Sometimes the attenuation selection is not for the actual attenuation factor or the loading factor, but for the voltage rating. It's just something to keep in mind.
CONTINUED IN THE NEXT POST BECAUSE OF LENGTH RESTRICTIONS