Avoiding common pitfalls will get you better data with minimum effort. Steve Sandler takes a look at four common scope mistakes and shows how to avoid them. This part looks at problems with probes.
The engineering community uses oscilloscopes more than any other piece of equipment, yet many of the published results are questionable at best. Some errors are very common, so we can eliminate a great deal of bad data by considering a few simple but key points. This second part of this four-part series covers probe issues: choosing an incorrect probe or using it improperly.
The three most common probe issues are not calibrating the probe (either for capacitance or for time skew), ringing due to the ground wire, and using 50-Ohm coaxial cable connected to a high-impedance input. The scope probe is represented as a high impedance in parallel with a high Q capacitance. Any series inductance (either connected to the tip or the ground) will result in a high Q tank circuit and ringing. The ringing frequency will generally be lower than the stated bandwidth of the probe.
The probe selection impacts the measurement rise time and the fidelity of the edge. A ring is created due to the probe connection.
An unterminated coaxial cable offers shielding, but it will still resonate with the scope input capacitance and with high Q. This example, using a 36" coaxial cable, can achieve very high Q at high frequencies, which results in ringing that is due to the probe, rather than the measurement.
The forward transfer gain of a 50-Ohm coax cable
terminated into 1 MegOhm.
Figure 7 shows the transient response measurement of a point of load (POL) in response to a current pulse. The current pulse is shown in the lower panel, with a step of 100-500 mA and rise and fall times of approximately 8 nS. The upper panel shows the voltage response with the same coax connection AC coupled through a Picotest J2130A bias injector into a 50-Ohm scope input. One trace (M1) shows the responses measured with the coax cable terminated into 50 Ohms. The other voltage trace (M4) shows the same response with the scope input set to an input impedance of 1 MegOhm. Note the severe ringing, which is due only to the unterminated coax cable.
Voltage response of a POL regulator to a step load using a 50-Ohm coax cable connection terminated into 50 Ohms and into 1 MegOhm.
Another mistake to avoid is influencing the measurement with the probe. The top panel in Figure 8 shows the measurement from our first example, and the results are approximately the same as reported above. In the lower panel, a 500-MHz, 9-pF probe was connected to the output as well as the 50-Ohm connection. The 9-pF probe tip capacitance results in a 55 percent error in the 50-Ohm measurement. One of the more common probe influence issues is the measurement of the switching frequency of a pulse width modulation (PWM) converter via the oscillator ramp pin. The capacitance of the probe can easily alter the switching frequency.
The addition of a scope probe on the signal we are measuring alters the result of the measurement. The 500-MHz probe (lower trace) is also poorly calibrated. Often this occurs when more than one piece of test equipment is connected to a single high-speed signal.
For all high-fidelity measurements up to 100 MHz, remove the ground clip from the measurements. Above 100 MHz, consider an active probe.
A typical oscilloscope probe presents 10-15 pF of capacitance. Many circuits cannot tolerate this, and, at the least, this capacitance can cause significant ringing. In most cases, an active probe is required.
Sometimes the best probe is no probe at all, and a coaxial cable is perfect. This solution also maximizes the sensitivity of the measurement. However, remember that a 50-Ohm coaxial cable must be terminated into 50 Ohms.