What happens depends on what sort of amplifier you have. All the modern "sane" audio amplifiers that I know of have a very low output impedance. This means that the amplifiers will fight each other, and dissipate a lot of heat doing so. If they're very powerful at all, they'll both get damaged.
Many older (and modern crazy) tube amplifiers have higher output impedance, and may survive just fine. You can't just say "tube amplifier" and walk away happy, though -- some tube amplifiers (like the "super linear" topology) also have voltage feedback in them, which makes for low- impedance outputs that can't be blithely paralleled.
If the output stage is a complementary PNP/NPN follower, which is common, the base/emitter of your output final transistors can be reverse-biased by up to the full (dozens of volts) power supply range. That can kill the expensive parts of your two audio outputs.
All audio sources have an impedance. In most cases at signal level that act s as your mixer, and no harm is done. You just don't always get the mix rat io you wanted.
At power levels, as Whit3rd said you quickly wind up with output stages fig hting and frying. Overcurrent, overdissipation and excess Veb are all probl ems.
Valves are always high impedance unless there's a transformer in the way, o r a most unusual multiparallel tranformerless output. They can pretend to b e low impedance upto a point with nfb, but the valve itself is still a high impedance source and can thus only maintain the low impedance effect withi n narrow limits of current, outside of which its high impedance becomes evi dent.
If each audio source is a line output with an output impedance in the range of 100-1k and the input impedance of the following device is much larger than the parallel combination, then the two sources should sum by superposition forming a crude mixer.
Assuming each of the two sources has the same impedance and same output level, if the two audio sources are uncorellated then the sum output at the junction will have approximately the same audio power as one output in isolation
The original poster, however, was talking about connecting two outputs, and did not suggest that the outputs were driving complementary signals. When complementary output signals are available, a (resistive) load between them is effectively a resistor to ground (the Thevenin or Norton equivalent of a resistor to ground), which ought not to be a problem.
With non-complementary signals, it might be a problem.
Some amps have one channell inverted. I think the Carver 1.5T is like that. The reason given is that only one channel is sucking down the + or - suppl y on bass, which is usually in phase. that kind of amp is alrdy easily brid gable as long as you feed it mono equal to each channel and don't go under double its minimum impedance load, as you said.
Seems ironic to those who aren't hep to this stuff that if you bridge such an amp, if you drive one channel harder, the OTHER channel will get hotter and more likely to blow first. Seems exactly opposite of how it seems it sh ould be but nope. The harder driven channel will run cooler.
h an amp, if you drive one channel harder, the OTHER channel will get hotte r and more likely to blow first. Seems exactly opposite of how it seems it should be but nope. The harder driven channel will run cooler.
** Here is an easy way to think about it.
In the normal case, with matched outputs, there is a point in the middle of the load that is always at zero volts - imagine it is grounded. So each ch annel sees load of half the total impedance.
Imagine one channel is driven at half the level of the other - then the ze ro point moves towards that channel and it sees a load of 1/3 the total imp edance.
Imagine one channel is driven at 1/10th, then the load is 1/11th of the tot al.
The worst case is with no drive at all to one channel - it *still* has to c ontinuously sink current equal to what the other channel is sourcing but th e voltage across each power device is fixed at maximum level so has much hi gher heat dissipation than the driven channel.
In this last case, the current limiting circuits will often trigger and cau se gross distortion of the sound.