In general the range of impedance that things like mixers will work into includes zero - so yes the output is short circuit protected, either by a real series resistance, or some sort of current sensing and  limiting arrangement. In an 'on chip' design the few tens of extra transistors this takes are trivial compared to the area of the bond pads to connect to the rest of the circuit.  Where higher poer is needed thermal shut down is also normally part of the design

Jack plugs and jacks have the distressing feature of shorting the contacts during insertion and removal,  even when the correct no of poles are used, so designers have to consider this. Also at the other end of the cable they tend to make the signal contacts first and the earthy side last, so 'live' insertion leads to all manner of clicks, pops and picked up ground loop humming noises until the plug is fully home in the jack.  Good designs have various delays or transient clamping circuits to avoid this.

The real advantage of the jack plug is rapid removal and replacement and that it can spin, and this was the feature that led to them being the telephone exchange connector of choice. Also there is very little tricky machining, mostly  lathe-work for the plug and stamped from flat phosphor bronze spring for the jack,  which suited the largely manual manufacture 1920s when they first appeared.
For high quality audio there are better choices (XLR or some DIN types are nicer..)  and the complexity argument no longer really applies so much , but mass manufacture for telephones and then later headphones and microphones, led to their ready availability & has led to the jack plug being the connector of choice in many cases, even though technically not the finest.

Note that the loudspeaker outputs of a power amplifier are not guaranteed to be current limited however, and it is certainly possible to damage some designs of  PA by driving into too low an impedance -in  my teen years I augmented my income replacing toasted  power transistors in a number of amplifiers where the owners had either mis-wired it or over done it on the loud music !

Mike.

Thanks Mike. I have an Old Maplins 150 Watt M.O.S.F.E.T. amplifier, home made. It is amazing and very rugged. It was available as a kit or fully assembled. It has been described as being bomb-proof. It performed faultlessly at full volume on Saturday despite its age. Circa 1980s. It is very simple and easy to repair if needed. Stuff should be like that these days. The cost of the toroidal power transformer for a stereo set up was more expensive than two Maplin amplifier module kits! 50-0-50 Volts.

Kits or assembled boards based upon the Maplin's design are still available new on fleabay.

Here is a link to what the original amp looked like. It looks like a home made job made for a sound a lighting hire company years ago.

Reminiscing thanks to a Maplin 150W MOSFET Amplifier on my bench. Wyeminster Slave amp repair. - YouTube

Z.

Amusingly I have one of that family too, and it is a design that has never been on the bench for new output devices, nor have I ever had to fix anyone else's either.

There were a few versions for various powers with more or less bosky FETs or parallel FETs  but all based on something like this

Although there is no current limit as such,  the two output devices are configured as source followers, but the gm is much lower than the equivalent bipolar design, where the base emitter voltage moves by a few hundred mV to get from a few mA to self destruction, while in the FETS the IV curve is more like this

Where the current flattens off and limits to a level that is set by the drain source voltage, but changes slowly a few amps per volt.

A bipolar transistor would  be

Where the current shoots off to infinity for a small change in base emitter voltage. (so much so that we normally look at curves of base current rather than voltage, ignoring the diode like base current to voltage conversion. )

So with FETs there  is a reliable element of self  limiting.
As anti-series source followers the two FETs provide current gain but no voltage gain, so the two gate voltages have to swing the full amplitude, but only driving the capacitance of the gates rather than lots of amps into the speaker.

If the FETS do the current gain, the voltage amplification is provided by the pairs of  bipolars sharing an emitter resistor - known as  a 'long tail pair', and that can be viewed as a crude 'op amp' responding to the difference in the two base voltages. When the two base are more than ~ 50mV apart the two collectors are essentially one up and one down as far as they can go. So for all sensible non clipped outputs, the two base voltages are identical, and one is the input and the other is a resistivly divided version of the output.

I agree, bomb proof.

images  from http://www.muzines.co.uk/articles/mosfet-amplifier/2651

and http://www.eeherald.com/section/news/onws20150111001a.html

M.

edit
compare if you like the maplin design with the

ETI 300 bipolar design

(www.solaris.no/.../eti300.html)

the long tail pairs are still there, now with active tails rather than simple resistance, but the output devices are Q12/13/ pulling up and Q 14.15 pulling down, while  Q 17/ Q17 are trying to detect the output current and remove the base drive if the output current is too high. There is a problem with the speed this protection operates, and the current limit circuit is prone to oscillations, and the output devices are not adequately protected against all fault cases. And it is considerably more complex to build, repair and debug. And a few like this have crossed the bench with their paws metaphorically facing upwards,and more than just the output devices shot.

Hm. The gate drive with a shorted output is probably potentially 50 V Mike, although this will be limited to considerably by the gate protection diodes and gate resistor, the real reason why mosfets appear very rugged is that they don't suffer from secondary breakdown which bipolar transistors do, a bipolar will fail with a high voltage and current together almost instantly, it is not the instantaneous dissipation, Mosfets only fail when the melt! 2N3055s are rated at 15A but if you try this current with 50V across them you have ornaments. More modern bipolar transistors are much more robust, the insulated gate power electronics types being particularly good, as you will have discovered if you have tried making a motor drive or similar. The protection circuits should be enough to protect bipolar transistors, but often are not because the area free from secondary breakdown is not well enough defined by the designer as it is far from linear voltage / current. Limiting the current only is nothing like good enough, the current limit needs to follow the inverse C-E voltage as well, and is not simply linear, so complex circuit. A typical high power bipolar amplifier will be seen to have a very large number of output transistors, simply so that worst case current of any one is small enough to stay away from secondary breakdown. A well know 3kW one (1.5kW / ch) has about 30 transistors per channel, these "amateur" designs say 150W from 2 transistors but failure with bad loads is very likely, thus magazines liked the Mosfet types.

Slight lesson in electronics!

Except you forget that the 2sk133 and the 2sj48 of the Maplin designs and knock-offs all  have inbuilt gate-source Zeners to clamp the S-G voltages voltage to about +/- 12~15V from memory, but in any case something well below the gate oxide breakdown voltage.

So the limited current in the preceding  'op-amp' and it's totem pole active load plus the various  series resistors, in conjunction those internal zeners, mean that there may have been 55V from source to the collector of TR4/5 without them, but once you allow zener current to flow, then the bias on the preceding stage is shifted and  the FETs are never more turned on than to the tune of 12V or so. (still maybe ten amps or so, but limited)  then it is I2t dissipation on the die , and a race between  that and the PSU fuse.

Agree 2ndry breakdown can be a thing, but only really if you are  pushing the envelope of the IV curves. Of course value engineered designs of both audio and SMPS do just that.

My point is not that bipolar designs cannot work , just that they are quite a lot harder to protect.

And IGBTs are indeed lovely, but they were not around in the 1970s when the VMOS  (Vertical MOS) came out.

I do suspect that had FETs come out sooner the basic bipolars would have been seen as a niche device for special cases, not the other way about, and the 2n7000 would have been seen like the BC107. .

For what its worth I like to see FETs in my RF power designs too.

Mike.

Except you forget that the 2sk133 and the 2sj48 of the Maplin designs and knock-offs all  have inbuilt gate-source Zeners to clamp the S-G voltages voltage to about +/- 12~15V from memory, but in any case something well below the gate oxide breakdown voltage.

So the limited current in the preceding  'op-amp' and it's totem pole active load plus the various  series resistors, in conjunction those internal zeners, mean that there may have been 55V from source to the collector of TR4/5 without them, but once you allow zener current to flow, then the bias on the preceding stage is shifted and  the FETs are never more turned on than to the tune of 12V or so. (still maybe ten amps or so, but limited)  then it is I2t dissipation on the die , and a race between  that and the PSU fuse.

Agree 2ndry breakdown can be a thing, but only really if you are  pushing the envelope of the IV curves. Of course value engineered designs of both audio and SMPS do just that.

My point is not that bipolar designs cannot work , just that they are quite a lot harder to protect.

And IGBTs are indeed lovely, but they were not around in the 1970s when the VMOS  (Vertical MOS) came out.

I do suspect that had FETs come out sooner the basic bipolars would have been seen as a niche device for special cases, not the other way about, and the 2n7000 would have been seen like the BC107. .

For what its worth I like to see FETs in my RF power designs too.

Mike.

I mentioned the gate protection Mike! Probably about 10A and 50V as you say, 500W possible dissipation for a minute or two.. I like RF FETS too, I have an excellent module in a 2m PA, 12V 80W out and no particular SWR susceptibility and linear, with RF NFB! I could go on about these audio amps for a long time Mike, but not today. It is difficult to get these fairly simple designs to very low THD, as loop gain is a problem, and mosfets don't help either, but that's another story. The Self articles on good design in an old Wireless World were very useful too.