Fender 15G: why FET in front of TDA2050?

I looked at some other TDA2050 discussions but I have not seen this mentioned.

Take a look at the Fender Frontman 15G schematic

http://support.fender.com/schematics/guitar_amplifiers/Frontman_15G_Schematic_68F.pdf

Although Fender lists this as the 15G/15R schematic, the reverb version is not there for downloading.

The PDF does not print out very well from my printer. One annoying thing is that the schematic was drawn with net junctions shown in the form of crossing lines without big black dots at the connection. When I went to school, crossing line junctions was forbidden on a schematic, you had to draw this as two T junctions, the crossing line had to jog over from one T junction to the other. Oh well.

Preamp output U3A (TL072) sends the signal to the input of the U4, the TDA2050. However, there is an interesting FET arrangement of Q1 (J111) at the input of the power amp stage. I first thought this might be something that was leftover from the reverb version, but now I think the FET is configured as some sort of amplitude-dependent variable load.

I have not seen this arrangement before. Fender did not put it there for no reason, what does it do?
 
When I see a FET along the signal path of a solid state guitar amp, I assume its there to add some tube-ish distortion as you suggest. But when I first saw this schematic, I thought Fender had drawn C39 incorrectly. This is a polarized cap, its anode connected to the FET gate and cathode to ground.
But that gate is connected to the negative supply through a 470K resistor.

Now that I have thought about it a bit, I am guessing the reverse-biased cap's leakage current is big enough that it will not allow the FET gate to reach a very negative Vgs level. So if the Vgs stays more positive than -2V then the J111 is not pinched off, and since its more negative than 0V then its not fully switched on, its resistive. From the J111 curves, if the voltage across the FET drain-to-source is small, say less than 1 volt, then its drain current curves do look triode-ish, and that Vds will be proportional to the preamp output level. At higher Vds, the curves flatten out like a pentode, meaning less "distortion" due to the FET at high signal level.
 
I was reminded of this old discussion recently while looking at a different Fender solid-state amp, a model they called the H.O.T. amp.

I am going to try linking a couple of schematics into this post. Hopefully I can link the thumbnails. The first one is the output stage of the Fender 15, showing the clamping FET in front of the TDA2050.

[IMGHTTPDEAD]http://img823.imageshack.us/img823/6982/fender15outputstage.jpg[/IMGHTTPDEAD]

OK. Now here is the output stage of the Fender H.O.T.

[IMGHTTPDEAD]http://img199.imageshack.us/img199/9189/fenderhotoutputstage.jpg[/IMGHTTPDEAD]

In the H.O.T., the FET is used as a compressor to limit the peak output signal excursion, if the clean switch settings are selected. If the Crunch or Lead distortion settings are selected, the compressor is not used.

I modified the H.O.T. circuit by substituting R46 with a 5K pot plus a series 3.87K resistor. This allows me to adjust the compression. I also defeated the Crunch/Lead disabling of the compressor.
 
The amp's marketing literature states "the 25 watt power amplifier section of the HOT utilizes Fender's "Clean Comp" compressor to assure clean clear sound when the "Bright" and "Full" presets are selected. "

(Full is the clean preset, Bright is also clean but with a little boost in the mids and highs, and some more preamp gain.)

I agree with you, this is not really an AGC-style compressor with a lot of range. But the J230 is a good audio amp FET, here the TL072 is bang/bang but the RC values to the FET gate means it is not driven like a switch, it turns on quickly but turns off softly. So unlike some limiter circuits, this one is not too harsh sounding. The original value of 9.1K for R46 means the limiting does not begin until the output peaks exceed about 22V, so on the Full setting I never heard any limiting, if it was happening I couldn't hear it. When I reduce the R46 resistor value, the limiting action begins earlier, now I can hear the attack/release pumping if I lower the value too much.