来源：Power Tip #6: Accurately Measuring Power Supply Ripple - EE Times
By Robert Kollman, Texas Instruments 12.03.2008 1
(Editor's Note: To see a linked list of all entries in this series, click here.)
Measuring power supply ripple properly is an art in itself. Figure 1 shows an example where a neophyte engineer grabbed a scope and got it all wrong. His first mistake was using a scope probe with long ground leads. His second was to place the loop formed by the probe and the ground lead in the vicinity of the power transformer and switching elements. His last mistake was allowing additional inductance between the scope probe and the output capacitor.
The problem is high frequency pick-up demonstrated in the ripple waveform. There are numerous high-speed, large-signal voltage and current waveforms within the power supply that can easily couple into the probe. They can include magnetic field coupling from the power transformer, electric field coupling from switching nodes, and common mode current generated by the transformer interwinding capacitance.
Figure 1: Improper Ripple Measurement Yield Poor Results.
(Click this image to view a larger, more detailed version)
The measured ripple can be substantially improved with a corrected measurement technique. First, ripple is usually specified with a bandwidth limit to preclude picking up high-frequency noise that really is not there. The oscilloscope used to make the measurement should be set to the proper bandwidth limit. Second, the antenna formed by the long ground lead can be eliminated by removing the probe “hat” and forming a pick-up as shown in Figure 2 . A short piece of wire is wrapped around the ground connection of the probe and is used to make the ground connection to the power supply. This has the additional benefit of reducing the length of the tip that is exposed to the high electromagnetic radiation near the power supply, thereby further reducing pick-up.
Finally, in isolated power supplies, substantial common mode currents are generated that can flow in the ground connection of the probe. This creates a voltage drop between the power supply ground connection and the scope ground connection, which shows up as ripple. To combat this, good attention to common mode filtering in the power supply design is required.
Additionally, wrapping the scope lead around a ferrite core can help minimize this current flow. This forms a common mode inductor that does not impact the differentially voltage measurement, but reduces the measurement error created by the common mode current. Figure 2 shows ripple voltages of the exact same circuit, but utilizing this improved measurement technique. High-frequency spikes have been virtually eliminated.
Figure 2: Four Simple Changes Drastically Improve Measurement.
(Click this image to view a larger, more detailed version)
In reality, power-supply ripple performance will be even better when it is integrated into a system. There will almost always be some inductance between the power supply and the rest of the system. The inductance may take the form of wiring or simply etch runs on a PWB. Also, there will always be additional bypass capacitance near the chips that are the load for the power supply.
These two create a low-pass filter that will further reduce power supply ripple and/or high-frequency noise. Taking an extreme example of a short run of one inch with 15 nH of inductance and 10 µF of bypass capacitor, the cut-off frequency of this filter is 400 kHz. In this case, this means there will be large reductions of high-frequency noise. Many times the cutoff frequency of this filter will be below the power supply ripple frequency and substantial ripple reduction is possible. A resourceful engineer should be able to find a way to use this in his test procedures.
To sum up: 1. Do not use grounding clips with long leads. Use short spring grounding clamps.
2. Do not form a test loop above the inductor and avoid it.
3. The oscilloscope needs to enable bandwidth limitation. My experience is that the bandwidth limit is 20M.