In April of 2010 the Geo-not-so-stationary satellite Galaxy-15 stopped responding to ground commands. It operates as a "bent-pipe" or Transponder system, which means - roughly speaking - it will amplify and rebroadcast back to earth whatever TV programming it happens to receive from Earth (within prescribed frequencies).
Geostationary satellites need to perform regular station-keeping maneuvers to stay put, and since commercial satellites are typically not able to do this, regular instructions are sent from the ground and must be received in order to keep the satellite wherever it's supposed to be.
So when Galaxy-15 stopped responding to commands, it started to drift uncontrolled within the geostationary "belt". That is a potential collision problem, but the other problem was that all the other systems worked. It maintained it's solar panels pointing towards the sun and kept "Earth lock" meaning that it had its receiving and transmitting antennas pointed down towards Earth.
Once it drifted into another "box" which happened to have TV signals beamed up from the ground in the same microwave band, it would dutifully amplify them and broadcast them back to Earth, and so even without a physical collision it could cause serious interruptions to other satellite's duties. This mindless capture and rebroadcast without control is what makes the term ZombieSat fitting.
The following is from Dealing with Galaxy 15: Zombiesats and on-orbit servicing:
Whatever malfunction did occur did not affect either the satellite’s ability to re-broadcast signals or its ability to keep its transponders pointed at the Earth and solar panels aligned with the Sun (known as “Earth lock”). This allowed the spacecraft to continue to receive and transmit signals. What it did affect was the ability of Intelsat’s ground controllers to maneuver Galaxy 15 to maintain its orbital position. Intelsat issued between 150,000 and 200,000 commands to the satellite in an attempt to get a response to either turn off its communications payload or maneuver. When these efforts failed, the company attempted to send an even stronger signal to try and force an overload of the satellite’s power system and cause it to shut down. This too failed. As a result, the satellite continued to drift slowly eastward through the GEO belt. What had seemed like a small problem was about to get much bigger.
My question: Is this a bug, or a feature? Isn't there some kind of automatic gain control or AGC? I am guessing that the idea in this attempt to shut down the transponders is to rely on the (preseumably) fixed gain of the amplifier - that stronger-in means stronger-out - and so one can trip something analagous to a circuit breaker by shouting really loud into the microphone so to speak. Was the inability of the amplifier to limit its gain designed in as a feature (a brute force shut-down when all else fails), or was it necessary due to other constraints, or did someone just forget to add an AGC?
One might argue that an AGC might diminish weak signals too far while attempting to limit a strong signal, but dynamic frequency-dependent gain has been around in fiber optic Dense Wavelength Division Multiplexing (DWDM) with 40 or 80 wavelengths, (although those are a different types of modulation e.g. digital QPSK)) so I'm thinking that that would not be a reason to avoid AGC.
That they attempted this suggests that either there was no AGC or that they knew it may be inadequate under extreme conditions. So its not even clear to me if the hoped-for shut down would have been exptected behavior, or wishful thinking. Could somebody with a really big steerable dish start shutting down lots of communication satellites this way? Could a day come when billions of people would experience cable TV withdrawl syndrom all at the same time?
above: "Basic +20dB, automatic gain control cell - as used in 1970s telephone networks." A simple example of a circuit which will adjust the amplifiers gain if the signal is too strong, from here.