How to judge the quality of high bandwidth measurement probes

There are two important performance issues to consider for oscilloscope probes, namely how much influence and interference the probe has on the measured signal, and how well the probe can reflect the actual signal. In order to obtain the best performance, it is usually necessary to keep the probe connection as short as possible, but sometimes this cannot be done, so it is necessary to understand the performance of various probes and their accessory configurations and understand the balance between various indicators.

To understand how the probe interferes with the signal being measured, you need to measure the effect of the probe's input impedance. To understand how well the probe reflects the true signal being measured, you need to measure the probe's response, or V _OUT ?V _IN .

Figure 1 shows the test setup for making these two measurements, either in the frequency domain using a three-port network analyzer or in the time domain using a high-speed step pulse source and a dual-channel oscilloscope. Regardless of which instrument is used, the signal source drives one input of the network analyzer or oscilloscope through a 50Ω transmission line, the probe output V _OUT is connected to the second input, the probe input is connected to a 50Ω thru fixture, and some probe accessory is used to measure V _IN , which represents the probe input. Using an oscilloscope with a probe allows for qualitative analysis, but to accurately evaluate the probe performance, a high-bandwidth sampling oscilloscope should also be used.

Connecting the probe to the 50Ω thru fixture and observing how V _IN at the probe changes can determine the effect of the probe input impedance on the signal. For this measurement, the time domain effect may be of more interest.

Figure 2a shows the effect of poor probe input impedance on a 1.2GHz clock signal. _VSRC is the signal that does not pass through the probe, and VIN _is the signal that is connected using the probe. Obviously, the probe has serious interference with the signal at this time, and such a probe configuration is considered unacceptable. In comparison, Figure 2b shows the result of the same measurement with a better probe. At this time, the probe has much less interference with the signal, so it can meet the measurement requirements.

The probe response can be evaluated by calculating the ratio of V _OUT to V _IN . To accurately represent the input signal, the probe should have a flat response within the specified bandwidth. The measurement should be made using the frequency domain method. HSPACE=12 ALT="Figure 2: Effect of probe input impedance on a 1.2GHz clock signal. A) Loading effect of a poor quality probe. B) Loading effect of a higher quality probe.

Figure 3a shows a poor probe response. When observing a signal with a fast transition edge, the +3dB peak at 1.8GHz will cause considerable overshoot and ringing. Some manufacturers have specially designed the probe frequency response to have a peak response to compensate for the signal loss caused by the probe input impedance. This practice may make you mistakenly believe that the signal looks good, but in fact the probe has greatly attenuated the measured signal, thus masking the error caused by the probe peak response. In comparison, Figure 3b shows the response of a better probe. The probe response is flat to 2GHz, and then produces a -3dB attenuation at the expected 3.5GHz bandwidth point.

Conclusion

In order to accurately measure high-bandwidth signals with an oscilloscope, the probe used must not have a significant impact or interference on the measured signal. Severe interference may cause a normal circuit to work abnormally or make a good signal look bad on the oscilloscope. The probe must not only accurately reproduce the measured signal, but also minimize the impact on the measured signal. The measurement setup provided in this article can accurately evaluate the probe configuration, so that you can make high-bandwidth signal measurements with greater confidence.