Let's say you have a 100 mic input channel console (just for grins). Maximum theoretical phantom current is 1.4A, but the inrush current to the phantom capacitors will be more than that. If that inrush current is a problem for any devices in the power supply, then there has be be inrush current limiting. There are a number of different ways of doing that, but let's focus on the switching sequence:
power source --> master power switch --> power rail --> phantom switch --> phantom rail
There are going to be large capacitors at the main power rail and the phantom rail, but the phantom supply is *usually* somewhat independent of the main rail (either a separate tap on the transformer or a voltage multiplier off the main rail). So at some point there has to be inrush current to the phantom rail; it doesn't matter whether or not the phantom switch is on when the master power is turned on, there will be inrush current if not limited or managed (if it indeed needs to be).
Of course, there will always be inrush current on the main power rail, it doesn't matter whether or not the phantom switch is on. If the phantom switch is on, then there *may* be more total inrush current. So as a power supply designer you have to anticipate the total maximum possible inrush current. If you build a PSU that will die at less than that maximum, then it will die sooner or later, probably sooner. So the supposition that phantom inrush can be damaging but the master inrush is not requires a PSU designer that was smart one day and an idiot the next. Or two different designers, one of whom is an idiot.
Consider that many large format consoles never have the phantom supply turned off and are used every day with no ill effects. Even the lowly Mackie at my church is switched on every day (twice a day on weekends) with the phantom switch always on and six condenser mics always attached. That's been installed for 12 years now--about 5,000 power cycles.
How stupid do you think SD would have to be to design a box that would die if the phantom wasn't switched after the unit was on, and yet they made the phantom switch a DIP? If it was an actual problem, their devices would be so unreliable as to be worthless for field use.
Remember that a lot of large consoles are used in the field too . . . probably more than in studios.
Chris said this:
Voltage regulators also fail due to inrush currents exceeding the capability of the regulator. But again this is mostly on large format consoles.
If a vreg fails because of inrush current (note that a lot of vregs have built-in current limiting) then the designer is incompetent or the manufacturer was too cheap to install a few pennies worth of inrush current limiting. If that's the case, then the main power rail is going to fail irrespective of when you switch on phantom. That gear should be considered what it is, crap.
More on the topic here:
http://prorecordingworkshop.lefora.com/2011/04/03/why-cant-you-turn-lfacs-off/
http://www.searsound.com/pdf/leaveiton.pdf
I love theory, but in reality I have repaired many consoles that the front end preamps were damaged by phantom power spikes. Midas, Soundcraft, Neve. I could go on. Its not just dc voltage we are talking here Its also ac voltage. The bottom line is phantom power spike damage from hot patching does exist and over time can and does damage preamp circuits. Its been very well documented and I have personally fixed many consoles that were damaged by it.
Here is a link to some useful information.
http://www.benchmarkmedia.com/wiki/index.php/Phantom_Power_DamageAlso THAT corporation also has some interesting advice and views on the subject.
Mackie designs also mentions the problem and have come up with a solution. The gear that does not have this protection is NOT crap.
Slow degradation (and eventual failure) of mic preamps resulting from hot patching is one of the mixing console manufacturing community's dirty little secrets. If you route a phantom-powered mic through a patchbay, it's roughly equivalent to shorting out a cable every time you patch it. The mic preamp's input transistors progressively break down (called zenering), while the mixer channel gets noisy and can eventually totally croak.
Mackie engineers weren't interested in any "solution" that even slightly affected the preamp's sound. A lot of research and considerable, old-fashioned trial and error listening resulted in the addition of ultra-high-speed, large-geometry input diodes to the front end of our input circuit. It completely protects XDR mic preamplifiers from the consequences of hot-patching and direct short circuits in cables carrying phantom power. Yet it has no affect on sound quality.