Measuring Noise
Noise on any power bus often emits EMI, which may cause regulatory issues during system testing. Load transients, switching ripple noise, and switching transient noise on a power bus can create unwanted radiation. Reducing noise via the methods discussed may significantly reduce system EMI.
Too often, intermittent system problems relate back to power problems that could have been addressed during the early stages of a design. A solid study of the power planes early in the debug stage might uncover addressable noise issues, thereby eliminating potential residual problems before they surface.
When measuring noise on a power bus, understand what you see. Large amounts of energy can radiate from power circuits, so a less-than-ideal probe without proper ground connections can cause improper measurements. Use one or two high-frequency probes with a small ground stub; a ground wire of any length can result in false noise measurements.
The oscilloscope must clearly display and trigger up to the noise frequency to which your circuit might react. At minimum, I suggest a 500-MHz oscilloscope and a passive probe with a ground stub. For better results, use two probes with the oscilloscope set for a math function that adds the two inputs, with one probe set to invert (Fig. 11).
This differential method reduces the risk of ground currents that may occur when using a single probe. For best results, I use an active differential probe to measure power-supply noise. Again, use the shortest connections from the probe pins to the output and measure directly across the high-frequency output capacitor if possible.
When measuring noise on a split-rail power supply, you should measure the noise from the positive to the negative rail. Since the noise sensitive circuits are often powered from the differential rails, one should measure the noise differentially across these rails.
Trigger the oscilloscope using normal mode, with the threshold adjusted for extreme peaks. Make two measurements, one with the scope triggering on the positive edge and the other with the scope set for negative slope. The difference between the two numbers is the peak-to-peak noise voltage. Measure the frequency of the ripple noise to verify the switcher is running at the correct frequency. Keep in mind that the ripple voltage is measured to the peaks of the triangle wave, which is usually lower than the switching transient voltage peaks (Fig. 5, again).
You can measure the frequency of the transient noise by triggering on the voltage peaks and setting the time base higher. Be sure that no bandwidth limiting is enabled on the oscilloscope. As mentioned earlier, these transients can be limited by adding a snubber(s) across the switching elements. In some cases, noise components may be at frequencies higher than the system can respond to, and thus not cause problems.
Noise often results from system operation, possibly a function of a circuit turning on or off. Or in the case of digital systems, a processor may be executing a high-performance subroutine along with low-power idle modes. A thorough understanding of the system operation will help identify other sources of noise affecting analog circuitry. Synchronizing system software with analog operations can often improve analog performance.
Spectrum analyzers can be used to quickly identify system noise components on a power bus in a complex system. The spectrum analyzer helps identify the noise sources by providing the exact frequencies of various noise components.
Taking the time to properly observe the power supplied to analog and high-speed devices is worth the effort. Some bench time early in the prototype debug stage can save lots of debug time later on. Very often I see system power problems surface as unrelated system anomalies or performance limitations.
Conclusion
Good analog designs starts with a clean analog power supply. The circuit presented provides a very clean and fully isolated +5 V and – 5 V rail. It easily fits on a small double-sided PCB with all components placed on one side of the board (Fig. 12). The layout shown can shrink further by using components available in smaller packages. If isolation is unnecessary, size and cost reductions are possible by removing the feedback isolation circuit.
Higher voltage outputs are possible, and higher currents are attainable, by using the higher power LM5000. Though the transformer needed for a flyback design does cost more than a single inductor, other approaches will often require more than one inductor and more than one regulator, resulting in a higher overall solution cost, and likely higher noise. By employing an integrated flyback IC along with post filtering, one can design a high-performance split-rail power supply with superior system results.