A highly misunderstood area of power supply design is the rectifier and smoothing capacitor section. Many a highly competent engineer has been known to under-design this rather crucial section. The reasons are not clearly understood until the waveforms are analysed.
Before we continue it may be prudent to understand the two basic specifications of diodes.
Peak current: This is the maximum forward current for which the positive-negative junction that makes up the diode is designed.
Peak Inverse Voltage (PIV): This is the maximum reverse voltage that may exist across the diode. This usually happens every half cycle and in a bridge rectifier is roughly equivalent to the peak input voltage (RMS x 1.414). In certain types of rectifier circuits this can be twice this value.
As said earlier, if it were not for the smoothing capacitor, the output from a rectifier would be a series of pulses from zero to full peak voltage playing havoc with the electronics it is supplying. Although the circuit being supplied is a lot happier with the smoothing cap in place, the rectifiers are now dealt a severe blow.
The reason is fairly straight forward. The rectifiers will conduct for only a small portion of the total cycle being from where the incoming voltage is slightly greater than that on the capacitor through to where the voltage falls below that of the capacitor. This happens just short of to just after the peak. During this time not only does the circuit want the current it demands delivered to it, but also the capacitor needs to have sufficient energy pumped into it to keep the circuit going until the next cycle peak.
During the conduction phase it has been known for the peak currents to reach in excess of ten times the average. This would mean a circuit using 1A would have the diodes conduct a peak of 10A, obviously for a very short period. Although the average dissipation of the diode is as though 1A were flowing through it, the active material still has a maximum capability that must not be exceeded. Doing so will cause the junction to "fuse" (melt together) and go short. In extreme cases the current may be so high it actually blows the junction open, but this is rare.
Back in the good ol' days, the first component found within a piece of "hi-tech equipment" was the mains transformer which then fed the rectifiers and smoothing capacitor. This component has, by the pure fact it is two mutually coupled inductors, impedance. Having impedance means it is also frequency dependent which translates into being a higher impedance at higher frequencies. This factor helps in limiting currents that may otherwise damage components. There are few designs, especially if the transformer is large and only a small proportion is supplying e.g. a control circuit, that have an inrush resistor or other current limiting components around the rectifiers (you know, just to be sure to be sure!).
Within the Switch Mode Power Supplies (SMPS) in present day equipment this inherent 'current limiting' device is no longer present. Although there is an EMC filter before the rectifier the inductance values are so small the current limiting properties are by no means that of an equivalent mains transformer.
In order to protect the rectifier from sharp switch on currents a Negative Temperature Coefficient resistor (NTCR or just NTC) is fitted. During the first few cycles the resistor is a high value but soon warms up and drops dramatically in resistance allowing most of the current to flow through to the rectifiers. Although the resistance range of the device depends on the current required it is usually only a few ohms when warm.
(you've probably started asking yourself what has this all got to do with power quality!? we're getting to that part now!)
Under perfect conditions well designed and specified rectifier circuits will work for years (discounting any latent defects within the components). However, modern day power is all but perfect. We indicated earlier that the conduction phase of a rectifier is small and that during this time the diodes were under severe strain. Should a transient or surge happen right at the peak i.e. during this high current conducting portion, the current the diodes need to handle may well exceed their capability - and they go short. This problem is also exacerbated with lower network and/or feed impedances.
Although some designers do investigate the peak current carrying capability of diodes and choose one appropriate for the situation making sure the peak current is catered for, many forget to also build in the reserve required to handle surges and transients. Designers of PC power supplies are the most guilty. One survey done at a company revealed all failed PC power supplies to only have only one damaged component.... the main incoming rectifier.
Now we've concentrated a lot on current as this is the primary failing of input rectifier diodes. However, there is one further issue and it would be remiss not to mention it here (especially as we spoke of PIV earlier!).
During normal operation the diode is usually expected to withstand no more than the peak voltage of the incoming mains voltage i.e. at 230VAC the peak is 325V. However, during any transients or surges the PIV can be exceeded and the junction then "breaks down". Unfortunately the energy is not controlled and usually results in component destruction. Fitting devices with diodes with PIV ratings close to the typical peak voltages that may exist will almost certainly result in failure.
If the diodes withstand the operating voltages then we need to start looking whether the components that follow the smoothing cap can.