Fullscreen HTML5Med Res (??MB)HTML4

To effectively use an oscilloscope, you must know how the scope's probe affects the signal waveform you are measuring. Here's a rundown on how scope probes work. By JONATHAN GORDON UNDERSTANDING OSCILLOSCOPE PROBES The most common type of oscilloscope probe used by technicians is the voltage-sensing passive probe. Because passive probes are so common and used so routinely, their peculiar effect on electronic circuits is universally experienced - though very often misunderstood. Nothing can replace the kind of troubleshooting knowledge that comes from viewing different waveforms from various types of circuits, such as digital, analog and radio frequency (RF). However, understanding how an oscilloscope probe loads the circuit under ·test will give you a new troubleshooting skill that no amount of experience can equal. Fig.1 is an equivalent circuit of a typical xlO scope probe and the vertical input of the associated oscilloscope. The probe's head has a 9-megohm resistor (Rl) that is shunted by a 4-Z0pF trimmer capacitor, Cl. The shielded coaxial cable that connects the probe to the scope has a distributed capacitance, C3, of approximately 80pF. An electronic signal travels through the probe tip, the probe head and compensating network, the shielded coaxial cable, and then to the scope's input connector. The input impedance of a typical scope is 1 megohm (RZ) shunted by a Z0pF capacitance (CZ). The scope's input characteristics are often printed near the input connector. The idea of compensating a passive probe is to balance the pro- COMPENSATION C1 CAO PROBE 4-2OpF PROBE TIP \ PROBE HEAD __ l_ ---1 + COAXIAL CABLE ___\___ I I CAO INPUT I r--~~I-----q- - - - - - - r'\ -:- I C2 20pf1 R2 1M .,. FIG.1: INPUT EQUIVALENT CIRCUIT of a x10 probe and oscilloscope. The probe's head includes a 9MO resistor which is shunted by a 4-20pF trimmer capacitor {Cl) to provide compensation. 4 SILICON CHIP be impedance and the scope impedance so their time constants are equal, as shown in the following formula: RlCl = RZ(CZ + C3) If their time constants are equal, electrical waveforms will be communicated from the probe tip to the scope input without the probe adding distortion to the signal. The amplitude of the displayed pulse will merely depend on the rnsistance ratio: Vscope = (RZ/(Rl + RZ))Vinput For example, using a xlO passive probe, a lV p-p input at the probe's tip will yield a 0.1 V p-p reading on the scope's display. The resulting decade attenuation of the xl0 probe is highly desirable because it affords a greater tip resistance (10 times the 1-megohm scope input resistance) to minimise circuit loading. Once the probe has been compensated by adjusting trimmer capacitor Cl, the probe and scope input are further reduced to the equivalent circuit shown in Fig.2. Copyright 1989, Gernsback Publications. Reprinted with permission from January 1 989 Radio-Electronics. PROBE TIP \ Another compensation scheme is shown in Fig.5 which uses a 4-20pF timmer capacitor to shunt the resistor in the probe's head. Pulse waveforms Fig.2: INPUT EQUIVALENT CIRCUIT of a xl0 compensated probe as seen at the probe's tip. Any circuit under measurement will now see a single impedance at the probe's tip of 10 megohms, Rp, shunted by an 11.2pF capacitor, Cp, which is a specification that is often printed on the probe's head or compensation box. Scope probe manufacturers have been clever in the methods they use to shunt the resistor in the probe's body. For example, Fig.3 shows the inside of a Tektronix Model 6006 xlO probe. The coaxial cable's centre conductor is extended into the probe's body where it connects to a cylinder that slides over the resistor. At one end of the resistor is a shorting slug that makes contact with the cylinder. A capacitor is thus formed by the cylinder and the resistor. The probe is compensated by screwing the probe's body into the locking sleeve , which is then tightened. During that adjustment, the inner cylinder slides a varying distance over the resistor and this varies the shunt capacitance. Note that the coaxial cable 's outer conductor is connected to a second cylinder for shielding. Fig.4 shows the inside of a Tektronix Model P6105A xl0 probe. In this unit the cylindrical tubing forms a fixed shunt capacitance across the resistor. No probe adjustment can be made in the probe's body to compensate the probe assembly. Instead the assembly uses a compensation box at the connector end that houses a trimmer capacitor, C2, which is connected from the centre conductor to ground. The shunt capacitance C3 and trimmer C2 are in parallel. A common variation of this compensation arrangement uses a trimmer in the BNC plug housing for the probe. This is used by probe manufacturers such as Coline and Hitachi. In general, probes are compared by how well they transfer an electrical pulse to the oscilloscope's input without causing distortion of one kind or another. But what is the real nature of an electrical pulse? Fig.6(a) shows what a rec- tangular pulse looks like on an oscilloscope as amplitude vs. time. However, as shown in Fig.6(b), the same rectangular pulse displayed on a spectrum analyser is transformed into a chart of amplitude vs. frequency - ie, it shows what the pulse would look like if broken up into its individual sinusoidal harmonic components. As you can see, the rectangular pulse is made up of both even and odd order harmonic components. CYLINDRICAL SLEEVEE LOCKING SLEEVE ALLIGATOR GROUNDING CLIP Fig.3: THIS PROBE IS COMPENSATED by adjusting the locking sleeve. C3 COAXIAL CABLE SHUNT CAPACITANCE ·, INNE,, CONDUCTOR CYLINDRICAL SLEEVE FORMING FIXED SHUNT CAPACITOR ACROSS RESISTOR ,::;,:._/ ALLIGATOR GROUNDING CLIP FIG.4: THIS PROBE IS COMPENSATED by adjusting a trimmer capacitor housed in the compensation box at the end of the probe's cable. JUNE 1989 5 Rl COAXIAL SHIELDED CABLE 9M , I , I I I \ t \ =...._ PPOBE TIP \ TRIMMERIADJUSTMENT SCREW FIG.5: THIS PROBE IS COMPENSATED by adjusting a trimmer capacitor housed in the probe's head. w Cl :::, PULSE w PULSE Q :::, ...:i t: ~ "'"' "'"' ...... TIME (a} :,:; :,:; "' en FREQUENCY (bJ FIG.6: THE PULSE WAVEFORM in (a) shows how an oscilloscope displays an . electrical pulse as amplitude vs. time. As shown in (b ), the same pulse waveform is displayed on spectrum analyser as amplitude vs. frequency. Unlike the rectangular pulse, a perfect square wave is made up of only odd-order harmonics (that is 1, 3, 5, 7, 9, etc). For example, if a lkHz square wave is input to an oscilloscope, the lkHz fundamental (1st harmonic) sinusoid up to the 9kHz harmonic sinusoid must be reproduced without attenuation or phase shifting. As shown in Fig.7(a), the 1st' and 3rd harmonic components produce a rather poor square wave. In Fig.7(b), the square wave looks a little better when the 1st, 3rd and 5th harmonics are present. And, as shown in Fig.7(c), the square wave looks better still when the 1st to 7th odd-order harmonic components are present. The waveform will appear sufficiently square and undistorted when frequencies are present out to about the 9th harmonic sinusoid. Obviously, the shape of a square wave displayed on an oscilloscope depends on the amplitude and phase relationships of the harmonic components. To accurately reproduce an electrical pulse it would be 6 SILICON CHIP necessary to design a circuit that responds equally well to an infinite number of harmonic frequencies , so that all harmonics are included. In practice, that cannot be done so a compromise between pulse shape and circuit design must be made. Compensation Every electronics technician has recorded pulse waveform data such as risetime, width, amplitude and repetition rate, only to realise too late - that the probe wasn't compensated. Because an improperly compensated probe can distort an otherwise perfect waveform, the probe's compensation capacitor must be correctly adjusted and the test data remeasured. Let's now examine how to properly compensate a probe and, additionally, how faulty compensation can affect the measurement of pulse waveforms. Because probes should often be compensated, most scopes provide a square-wave calibration signal accessible from the front panel. There are other types of probe calibrators, such as a linefrequency calibrator, a lkHz square-wave calibrator and other more exotic types. However, they are used less often than the more convenient front-panel scope calibrators. The front-panel probe calibration signal is a lkHz repetitive square wave. A lkHz square wave is used because the sinusoidal harmonic components are very close together, so the slightest offset in the probe's frequency response will affect the amplitude and phase relationship of many harmonic components at once, resulting in a visually distorted waveform. Fig.8(a) shows a lkHz calibration signal from a properly compensated probe. Notice that the square wave is undistorted. Fig.8(b) shows the same calibration signal from an over-compensated probe. Here, the leading edge high-frequency harmonics are passed while some of the lower-frequency harmonics are slightly attenuated. Some phase shifting has also occurred. The greater the drooping effect of the waveform, the greater the degree of low-frequency harmonic attenuation and phase shifting that has occurred. Fig.8(c) shows the same calibration signal from an under compensated probe. Notice the lack of high frequency components in the leading edge of the square wave (indicated by rounding). Now let's examine the relationship between pulse shape, rise time and the capacitance of the circuit. When the driving pulse has a slow rise time, or the pulse width is comparatively wide, the stray capacitance can have a fairly high value without producing visible distortion. That's because slower rise times and greater pulse widths correspond to fewer high frequency harmonics. However, the same value of stray capacitance can become intolerable when the driving pulse has extremely fast rise times or a very narrow width. The waveshape then depends critically on the preservation of high-frequency harmonics. As more and more stray shunt capacitance is added to the circuit under test, the shunt (bypass) capacitive reactance decreases in value. Harmonic frequencies that comprise the pulse's · edge will now be shorted to ground by the lower shunt-capacitive reactance. Assume, for example, that a computer circuit is working just fine and that you want to observe the 20MHz master clock. So you connect your xl scope probe - and the whole system crashes. The clock's waveform displayed on the scope looks a little distorted. You then remove the probe from the circuit and the system immediately comes up. What happened? FUNDAMENTAL PLUS 3RD HARMONIC ' sistive attenuation ratio, the lower the probe-tip shunt capacit-ance. For example, xl, xlO and xl00 probe attenuation factors might have a 54pF, 11.2pF and 2pF tip capactance values, respectively. Continuous wave (a) FUNDAMENTAL ....I (b) (a) FUNDAMENTAL PLUS 3RD AND 5TH HARMONICS ✓-\ \ \ \ \ \ (b) (c) FIG.8: A PROPERLY COMPENSATED FUNOAMENT AL PLUS 3RD, 5TH AND 7TH HARMONICS I I I \ \ (c) FIG.7: ALL PULSE WAVEFORMS are built up from sinusoidal harmonic components. As shown in (a), the fundamental (1st harmonic) plus the 3rd harmonic create a rather distorted square wave. In (b), the 5th harmonic is added, thus creating a less distorted square wave. In (c), the 7th harmonic is added, thus minimising distortion even further. probe will display a lkHz calibration signal as a perfect square wave as shown in (a). However, in (b), when the probe is over compensated, the same lkHz signal shows lowfrequency attenuation and phase shifting as indicated by the drooping effect. As shown in (c), when the probe is under compensated, the high-frequency components are lost, as indicated by the rounded leading edge. One possible explanation is that the added shunt capacitance of the probe degraded the clock's rise time, which threw off the system timing. As shown in Fig.9, slower rise time translates into a wider pulse width. In this situation, you should try using a xlO probe instead of a xl probe because the xlO probe has a lower shunt capacitance. In general, the greater the re- When measuring a continuous signal from the output of a sinewave oscillator, the probe-tip's capacitive reactance (X_p) at the operating frequency should be considered. In Fig.10, the total probe tip impedance, designated Zp, includes the probe's resistive, capacitive and inductive elements. The capacitive and resistive elements make up most of the probe's impedance. However, some probes also include additional inductive elements that are designed into the probe itself to offset the capacitive loading. For worst case analysis, use the probe's capacitive reactance formula: Xp = 1/2-n-fC where C is the probe-tip capacitance. This value is often marked somewhere on the body of the probe. For example, the Tektronix model P6105A passive probe has a 10-megohm input resistance with a tip capacitance of 11.2pF. The Xp will equal 290 ohms at 50MHz. Depending on the impedance of the source, the probe's loading could have a major effect on the signal amplitude and possibly interfere with the operation of the circuit under test. The typical curves for probe impedance vs. frequency vary for each probe type so consult your probe's specifications. For sinewave amplitude measurements, a probe should have the highest possible impedance at the frequency of interest. Voltage derating The maximum voltage (DC and AC) that can be safely handled by a probe varies with frequency. Fig.11 shows the voltage derating curve for a Tektronix Model P6105A passive probe. The curve may be summarised by saying that the maximum voltage handling capability is JUNE 1989 7 AMPLITUDE 10M RISE TIME It= I 90 % X' 50 % I 10 ¾ \ I / 1M \ I .;;- \ ;l I\ e, I I~ -1 '' 100k ' '' = >< ~ < \ 10k ~ WIDTH WIDTH ~ ·, \zp \ 'Xp' ' , ' 1k '' I\ \, '' ' ......\ "'" TIME FIG.9: PULSE WIDTH IS MEASURED at the 50% marks on the waveform. As the circuit's shunt capacitance is increased, the rise time slows, causing the pulse to become wider. 100 .01 0.1 1 10 ~,... 100 1000 FREQUENCY I '1!Hz) FIG.10: WHEN MEASURING SINE waves, it's important to know how the probe's input impedance (Zp) changes with regard to frequency. 1000 '\ \. 100% [\ G' < 97% 'r- "' ~ ~ 100 'r-. ~ 70.7% 3dB e. w "'< :; 0 > 10 1 10 30 100 FREQUENCY I MHz) FIG.11: THE MAXIMUM VOLTAGE that a probe can handle decreases as the frequency increases. inversely proportional to the frequency. Most scope probes are supplied with their own voltagederating specification. Bandwidth Scope probes are often rated for bandwidth. It's best to use a probe that has a bandwidth equal to or higher than that of your scope. If the probe's bandwidth is less than that of the scope, then the input frequency will be limited by the probe. Fig.12 shows the response curve of a probe having a 100MHz bandwidth. By definition, bandwidth is the upper frequency where the scope's displayed voltage is down 8 SILICON CHIP 100 FREQUENCY 1MHz) FIG.12: PROBE BANDWIDTH is the point where the voltage amplitude is down 3dB from a starting reference level. 3dB from the reference frequency voltage. The formula to calculate decibels is: dB = 20 log Vout!Vin For example, if the input is a 1V p-p 100MHz sinewave, then at - 3dB the scope 's waveform will show an amplitude of only 0.707V p-p (which is an amplitude accuracy of 70.7%). For an amplitude accuracy of 90% or better, you must limit the input frequency to about 30MHz. Another useful equation shows how bandwidth is related to risetime (tr): Bandwidth = 0.35/tr From this it follows that the faster the risetime, the greater the bandwidth. For low-frequency applications (audio frequencies), you can choose a xl passive probe because it costs the least and will do the job. But be aware that the xl probe has a limited bandwidth - less than 40MHz. On the other hand, use the xlO probe for general digital, analog and RF measurements. A xlO probe has low capacitance and a bandwidth upwards to 400MHz, depending on the model and the cable length. Note also that if you are using switched probes which are the most commonly available, their stated bandwidth only applies to the x10 mode. In the direct (xl) mode, their bandwidth is much less. For example, the Coline SPlO0 probe has a rated bandwidth from DC to 100MHz in the xlO mode but its bandwidth drops to 10MHz in the xl mode. Its capacitive loading also increases from 16pF to 55pF (plus the input capacitance of the oscilloscope). Although not discussed in this article, for frequencies higher than 400MHz choose one of the active probes (FET) with a xl sensitivity. This type of probe will provide high sensitivity, low shunt capacitance and a bandwidth greater than 900MHz. Rp CP 10M 11.2pf GROUND LEAD L FIG.13: GROUND-LEAD INDUCTANCE will reduce the high-frequency response through its series-inductive reactance. RINGING ,J l\tWrG--. Grounding How often have you touched the probe tip to an IC pin only to see a waveform you know from experience isn't right? Then, after moving the probe's grounding clip from the chassis to the IC's ground pin, the scope's trace immediately shaped up and became recognisable as the waveform you've seen before. This leads us to the obvious question: how does the probe's grounding lead affect the circuit measurement? The obvious answer is that improper grounding will generally distort the waveform by allowing excess noise to be picked up. That's true but it's only part of the reason. Fig.13 shows an equivalent circuit of a passive probe connected to a voltage source. Notice the series ground-lead reactance, L, whioh represents the ground return path. Rp and Cp represent the equivalent impedance as seen at the compensated probe's tip. When measuring any signal, the series inductive reactance will be proportional to both frequency and inductance by the formula: X1 = 2-irfL The higher-frequency harmonics will therefore see a larger inductive reactance than the lower frequency harmonics. The pulse waveform displayed on the oscilloscope will show distortion and aberrations because the p-p voltages of the higher-frequency FIG.14: PULSE RINGING OCCURS when the ground-loop inductance and probe tip capacitance form a series resonant circuit that is shock-excited by a very fast rise-time pulse. harmonics have been attenuated and phase shifted across the ground-lead inductive reactance. Now let's get back to the original problem. If you move the probe's grounding clip from the chassis to the..IC itself, then the ground-loop inductance will be reduced. That allows the high-frequency harmonics to reach the scope's input, so the trace shapes up. As a rule of thumb, when making any kind of measurement - such as amplitude, rise time and pulse width - you should use the shortest grounding path possible. As shown in Fig.14, loop inductance may also manifest itself as ringing on the leading and trailing edges of the signal pulse. The ground-lead inductance and probetip capacitance form a seriesresonant circuit with only a lOMQ resistor for damping. When shockexcited by a pulse, the resonant circuit will ring with a predictable damped oscillation. For example, an 11.2pF passive probe having a 15cm ground lead will ring at about 140MHz when hit by a fast rise-time pulse. As the ringing frequency increases, it tends to fall outside the scope's passband and is highly attenuated. It's therefore desirable to try to increase the ringing frequency. To do that, use the shortest possible ground lead and the probe with the lowest shunt capacitance. It's ironic but for the reasons just mentioned, you're more likely to see ringing on an expensive high passband 300MHz scope than a low passband 20MHz scope. Mechanical properties Often, while touching or rearranging a probe or lead wire, unpredictable, confusing and nonrepeatable effects are produced on the observed waveform. That kind of problem may have more to do with the mechanical nature of the probe than anything else. If the probe's inner signalcarrying conductor is poorly shielded, then the probe's cabling will be susceptible to external electric fields. (The shielding could become frayed due to constant flexing of the cabling over many months or years of use). A poorly shielded wire can act as an antenna and pick up all types of interference such as electrical noise from fluorescent lamps, radio stations and signals generated by nearby equipment. To virtually eliminate any external field pickup, always use a probe with coaxial cable shielding of the centre conductor. As a final note, the probe tip should be clean or a poor circuit connection will result. Also, be sure to check printed circuit boards for a conformal coating which is sometimes used to guard against humidity and static. This can easily prevent an electrical connection between the probe tip and the circuit. It may be necessary to scrape off some of that coating to make a good connection to the circuit. ~ Footnote: next month, we plan to describe a practical x10 CAO probe that you can build yourself. This unit is essentially a practical version of the circuit shown in Fig.5 on page 6. JUNE 1989 9