Mic Pads: When And How To Use Them

...from a response I recently gave on another forum where the question was asked when to use the -10dB switch on a Neumann U87:

The concept of how the pad switch on most transistor condenser mics works, and when to use it, is fairly simple to understand and implement:
 
The -10dB pad switch lowers the amount of voltage coming from the capsule into the input of the FET (the transistor's gate, in this case) by 10dB.
 This allows you to pump the electrical equivalent of 10dB more into the mic's capsule before the FET will clip/distort.
 
If you do not change the amount of acoustic energy that goes into the capsule when you engage the -10dB switch, the output of the whole mic will simply go down by 10dB. 

That all seems straightforward and user friendly. But consider the negatives before you click the pad switch:
 
The first problem:
 As the mic's basic noise floor always stays the same, regardless of whether you use the pad or not, engaging the switch decreases the s/n of the mic by 10dB. 
In other words, the mic's amp has 10dB less signal to work with, but it will still produce the same background hiss/noise. Remember: the -10dB switch only reduces the amount of energy coming from the capsule, not the mic's amp gain and noise! 

The second problem with using the -10dB switch is the way how Neumann and other manufacturers lower the capsule's gain: through negative feedback (NFB.) And using NFB in mics and many other audio components produces audible artifacts from the inevitable time delay/phase shift.
 So, you are not just losing a clean 10dB of gain when you flip the switch, you are adding something unwanted and clearly noticeable.
 
Some mics use another pad method to reduce gain from the capsule: feed less polarization voltage to it. This avoids the side effect of NFB (see above) but unfortunately adds noise, as is always the case when you starve the capsule from voltage necessary to operate it at its lowest noise level. (See also Volker's mention of a third way to lower gain, below)
 
The third problem with using the -10dB switch, specifically on U87 mics: 
The K87 capsule has a tendency to mechanically distort in the high frequencies when pushed hard, just about the point when the FET starts distorting. 
With other words: you may engage the switch, but may not entirely get rid of distortion, because, other than backing off, there is no way to prevent mechanical capsule distortion when too much SPL is hitting the diaphragm.
 
All in all, I consider the pad switch on most condenser mics, regardless of brand and type of impedance converter as merely an emergency switch, to be used only in rare cases of extremely high SPL-exposure, and with the understanding that the side effects may not be worth using the pad after all, but that it may be a cleaner solution to back off the mic from the sound source, rather than padding it down.
 
A final word of explanation on the seemingly confusing fact that the Neumann U87A has higher output but lower head room (i.e. given the same sound level, the U87A will distort about 6dB before the old U87 will distort):
 
The limiting factor in single-FET mics (like all U87 models) is the pitifully low headroom of the FET. So it makes sense that if you pump yet another 10dB more into the FET, as the U87A's capsule does (hence the higher output of the mic) the FET will poop out that much sooner than on the old model. 

The nice gain increase of the U87A in comparison to its U87 predecessor is therefore only useful in situations of moderate sound pressure level exposure.