Anecdotal, unfortunately.
Speaking of which, I've plugged and unplugged phantom from my TT rack for 15+ years and never seen a failure from a preamp or a mic.
That being said, driving devices such as BJT or FET type transistors don't usually short through the emitter-collector or drain-source path. They usually short base or gate due to VBE/VGS excursions. More often than that, it's "protection" devices in the circuit that make a sacrificial short.
"multiple AEA active ribbons" dying at the same location by the hand of the same person. I took a quick look and didn't see much chatter online about mass failures. Without further info, I assume that this particular person is doing something wrong or has something wired wrong.
I'm not saying that "phantom" isn't the cause, but in years and years of troubleshooting for people and designing things for myself and for large corporations, I've found that the majority of the time the initial specious reasoning for issues tends to only be skin deep and there are usually more in-depth reasons for failure modes than folks are willing to dig for.
Take an example of mine:CTO of the company decides to take a mature (been selling for years without problems) 1GHz RF switch matrix that I designed (16 input ports, split and switched signal paths to 2 outputs) and put it in a box with other measurement equipment and then call it a "total solution". We sell dozens of these to a customer in Thailand.
Thai company starts complaining that there is no signal on some of the ports. Our SE Asian rep visits the sites and sees NO signal getting through the box.
Cue the usual "blame the engineer" circlejerk.
Point out that we have thousands of switch matrix units in the field with no failures.
CTO points to almost 100% failure of switch matrix in
his "total solution" to prove that it's
my design that has an issue. He strongly suggests that I go to Thailand to figure out the problem and meet with the executives of the CATV company and explain the issue.
I don't know what the issue is yet.
I get to Thailand (quite a long flight indeed) and after a couple hours sleep I'm off to the sites to investigate. Long story made short, the units are indeed dead and the action of plugging an RF cable into the box immediately kills that channel.
I take note of their odd grounding methods over there. Their ground strapping isn't done between network racks as it's done everywhere else in the world. It's done through a thick cable back to a single star point in the BUILDING. So each rack has 50 to 100FT of ground cable back to a huge bus-bar grounding network. I measured the DC difference (network racks are 48V DC, just like phantom) at a few volts, very similar to the residual DC being spoke of in the GroupDIY thread.
I can't make accurate measurements in the field since this is a government owned facility and they won't allow me touching any of their gear. We do hook a ground strap from the next rack where the RF is sourced and hook it to the chassis of our rack and make connections.. They do NOT blow inputs in this case, meaning there is a huge ground potential issue, likely caused by the long ground loops. They refuse to add ground strapping to the chassis due to demarcation issues between government network racks and 3rd party equipment.
Also luckily their executives didn't have time for me so I got a day to visit temples and such before my 24 hour trip home.
Back in the lab I have to set up a mock rack with 48V supplies and do my testing. I probe and prod every single inch of the box and find that floating chassis from the 48V return by a volt or so can cause an amplifier failure. RF amps are notoriously sensitive to being overpowered, moreso than common BJT/FET devices in mics and preamps.
The CTO demands we add more TVS/zener protection, (which is industry standard for telecommunications despite what THAT has to say in their document) and they asked another engineer to do a secondary analysis.
During all my work, I realized that it's not overpowering the amps in a sense of jamming positive *power* (Power=RF signal level + transient peak from momentary DC offset), but rather a case where ground potential rises quickly, then falls more quickly than the signal power has time to settle. The effect is much like a roller coaster where the drop lifts you from the seat because the coaster car drops faster than you, and for a quick moment your positive potential is much higher than normal.
Unfortunately, the other engineer did not understand the concept that *ground* is not an absolute value. He was certain that adding huge amounts of front-end protection in the form of parallel TVS devices was the solution. The CTO agreed with him and they added their mods which did work to mitigate the destroyed amplifiers, but added a huge amount of distortion and loss of high-frequency flatness. My 1GHz switch was now more like 800MHz, but I digress.
I then continued to verify my though process by surmising that if the voltage potential "surge" was indeed more like the ground dropping out, then increasing the size of the AC coupling capacitor on the output of the amplifier (the amps are DC biases on their output pins and are more like single darlington transistors than opamps) would lower the impedance of the output at closer to DC levels and the "surge" would not accumulate at the output. Input coupling was left at 100nF and the output was increased from 100nF to 1uF.
No more failures.
I had fixed it with a single part change, yet neither the other engineer nor the CTO would accept my findings because "it didn't make sense". I then wrote a 10 page paper explaining the failures in Thailand, the reasoning, and the methods to finding a fix that worked around the Thai grounding issues.
They still went with the overkill approach of adding protection diodes all over the place, much like the THAT paper did.