Okay so I have tried the resistor method and yes it does end up using a lot of resistors. I had a design going that used 4 1k Ohm resistors and one 750 Ohm resistor and some relays to create a very basic high powered stepped resistor. I think it gave me 10 or so distinct voltage levels that changed the speaker volumes.

The trouble with going that way is that the resistors themselves are very expensive (Digikey had them going for 6 and 10 dollars respectively) and they were difficult to source in quantity.

I think a stepped autotransformer would work better for what I want but I can't seem to find any. I spoke with a few of the volume control manufacturers and they all said that they cannot sell me the part individually nor can they tell me where they purchase theirs from so I am at a loss on how to locate a simple 5-10 step autotransformer.

That all said, it might sound like crap for the reasons already cited. But a PA doesn't have a real broad frequency range anyway, so there's a chance? I'm almost tempted to try it with my home intercom.

On this note, I have tried out some simple tests with the resistor design and it doesn't seem to have too much of an averse effect in terms of sound quality, but then again, I wasn't playing music or anything like that and the quality of the speakers could have been questionable as well.

 

Interesting test results.

I guess the resistors are expensive because of the power ratings? I have 15W resistors that were only about 25¢. Maybe you could knock down your power rating a bit since, as I noted earlier, the highest power dissipation you'll have in the resistor is 1/4th of what the max power of the speaker is. This happens when resistor and speaker impedances match and the two together take half of what the speaker alone would. So half of one half is 1/4th. Of course you still have to double that (or more) to allow for safe operation.

I've been reading MOSFET datasheets and I don't think my MOSFET-as-resistor idea could ever work. At less than full-on gate voltage, the current allowable through the MOSFET goes flat at a low voltage. That means severe non-linearity, as mentioned by rogs. Live and learn.

 

Yeah so ideally my box would never need to handle more than 40W output on each of the lines but in some cases, multiple 30W speaker spheres will be attached to them so I have been having to overshoot quite a bit. I was using some 50/100W resistors to be safe.

I hadn't actually looked at that MOSFET datasheet in any detail yet but I had assumed the linearity would be difficult. Right now I am thinking I will just buy the standard wall mount 100W volume controls from ATLAS or someone and strip the auto transformers off them to put in my box.

I suspect I could get them cheaper by themselves but I can't find them anywhere for sale although every volume control seems to use the same hardware.