This is an extremely grey area as there are so many factors that can attribute to the failing of a component being fed from a power supply. To generalize could be dangerous, so from the outset we are claiming these are guidelines only!
The most obvious cause is power supply output sags or surges, and with electronics the timings applied to mains power are not at play here. Even a 1µs glitch is now more than 1000 clock cycles on the modern PC CPU. Imagine how much could go wrong in this brief period!?
The problem is worsened by the fact that it does not have to be a full sag or surge but merely a deviation of less than 10% to have the modern CPU fall over dramatically. This comes from the fact that memory devices (EPROMS, RAM, CMOS, etc.) are designed for reliable reading at and around their working voltage. Changes in the supply voltage change the amplifiers that 'look at' the memory cells and could have them reading the wrong values.
Ok, so we went into a little deep technical stuff there for a short while. Let's get back to reality.
Things are, fortunately, a little less complicated as the component on the output of the power supply is a pretty fat capacitor. Glitches of µs are very difficult to achieve, but milliseconds are very probable (meaning we have even more CPU clock cycles to play with!).
Power supplies are the most obvious route where mains power disturbances make their way through to the circuitry. It is not uncommon on cheap versions of power supplies to, at least, find the output affected by mains noise or disturbances. Often, the higher the frequency of the noise or disturbance, the greater the output variations. We did say the least was output variations. Such mains disturbances have been known to cause power supplies to completely shut down.
But sadly, this is where we need to stop the story as practice has proven it is seldom a power supply that causes such damage (not saying it cannot, just seldom does!). Many 'experts' turn their attention to the mains power in an attempt to hide their badly designed systems.
The bad design comes in the form of ground loops and weak ground systems. They are not always obvious at first but upon closer investigation they usually make themselves clear. The moment the system is connected to 'the outside world' there is a possibility of a ground loop.
More obvious connections are serial ports, printer ports, external SCSI, etc. Less obvious are LAN, modem, and other similar connections. These former types have a signal ground which many times obtains the ground from the motherboard and not the chassis. The latter type can, and often do, have capacitive coupling to ground which at 50Hz is a high impedance but by 50µs (10kHz) is starting to look like a dead short!
Ground currents have two possible sources. The first is leakage from the mains down to ground such that the Earth is 'moved', this voltage then finds its way down the signal cables. The second is the Ground is 'moved' by an external means (lightning strike?) yet the mains input can be stable - the currents again flowing down the signal cables. Common mode noise is dealt with in a later section.
With extraneous currents making their way through the circuitry rather than protection (such as the chassis) there is now an increased possibility of damage, and one that the PQI needs to be aware of so as to remove suspicion from the mains power.