Consolidate your noise-solutions toolbox

Finding the solution to a noise problem on the digital side of the circuit versus the analog side does not always require a different set of tools. Many times, you can track down noise to one root cause: your layout. Specifically, your layout may have sensitive traces or pins next to noisy traces or pins. If you find that you have this problem in your circuit, try to physically separate the noisy traces from the high-impedance lines. If this simple solution isn't feasible, an alternative tool exists.

If you think of this problem from a digital perspective, switching noise may be interfering with your system oscillator. A perfect example is a PWM pulse close to a low-power oscillator circuit. Because the oscillator circuit draws little power during operation, it has a high internal impedance. As a result, the switching noise of the PWM pulse capacitively couples from trace to trace into your oscillator circuit. This scenario, in turn, causes clock jitter and, at worst, irregularity. Under these conditions, the pins may need a guard ring (Figure 1a).

In an analog circuit, the dynamics behind noise generation may be different, but the outcome is equally catastrophic. Take the example of a high-impedance pin of an operational amplifier residing close to a trace that has a different voltage potential or, worse yet, switching noise. The easiest circumstance to overlook is the trace that has a different voltage potential from the high-impedance pin.

For instance, if the input pins of an op amp have extremely high impedance, the input bias current can be as low as a few picoamperes at room temperature. The leakage current between the higher potential trace and the high-impedance trace makes the amplifier's input bias current appear higher than it is. This leakage current could be a result of close traces; humidity, board contamination, or dust could aggravate it. The leakage current can sometimes be as much as five to 10 times higher than the input bias current. This error may create severe problems in the application circuit. Under this type of condition, the pins may also need a guard ring (Figure 1b).

A low-impedance guard ring is a great tool to have in your toolbox. "Low impedance" implies that the guard ring has low resistance as well as low inductance. Whether it is around the digital-device pins or the analog pins, it can separate sensitive operations from higher voltage or noisy events. In Figure 1a, the guard ring surrounds the processor or controller pins, oscillator clock, and capacitors. Ground connects to the guard ring to establish a low-impedance potential. The guard ring in Figure 1b connects to a low-impedance voltage that is close in magnitude to that of the amplifier's input pins. You can find these low-impedance voltages in three common places. Ground is the most likely connection, and this scenario makes sense if one of your amplifier inputs connects to ground. The guard ring can also connect to the amplifier's inverting input as long as the feedback impedance and source impedance to the amplifier are relatively low. A third option is to connect the guard ring to the amplifier's noninverting input as long as this pin connects to a low-impedance-voltage source.

Tackling noise problems can be tough if you don't dig in and see the problem from the inside out. With the above-mentioned digital and analog circuitry, impedance mismatches are at the center of the noise problems. Once you understand the source of the problem, the tools in your toolbox could service either domain with ease.