 |
| Fig. 1: VRM-switching noise is a self aggressor that can be identified because it is synchronous with the PWM clock. |
An example of self-aggression would be VRM-switching noise. Figure 1 shows ripple on a 900 millivolt rail (yellow trace) at a time when no load is present. One of the things that tells us this is switching noise is that it is synchronous to the PWM clock (red trace). Ripple that is synchronous with the switching clock is a typical figure of merit for identifying switching noise.
 |
| Fig. 2: On-die noise (blue) can bleed through to board-level power rails (green). |
Another example of self-aggression noise occurs on the on-die Vcc rail, shown in Figure 2. As general-purpose IOs switch (one IO being probed is shown by the yellow trace), the IO driver that connects it to the Vcc causes noise spikes on the Vcc rail (the blue trace). This bleeds through to the board power rail (the green trace), as well, causing the dip you see around the time the Vcc spikes are highest. This is a case where on-die noise affects the board, despite the low impedance of the board itself.
It can be challenging to probe the Vcc in order to measure noise. One way you can observe Vcc self-aggression noise is to leave one IO switched to a permanent high state, which is essentially connected through to the Vcc rail. That allows you to observe what's happening on that rail as you induce other effects in the system, such as switching multiple IOs.
The same type of self-aggression occurs within the on-die Vdd, the core logic power rail within the device, although that is much more difficult to observe and usually requires a specially instrumented die.
For more about self-aggression noise, see our on-demand webinar: Fundamentals of Power Integrity
No comments:
Post a Comment