Small Leslie motor control: variable voltage or PWM?

[rufustf]

I've been going nuts trying to get this leslie motor happening: It's a small 12/24VDC/300ma motor that's got a resistance of about 8Ω across it. Here's a schematic of it's current motor control:
http://www.monkeyview.net/id/3604/schematics/index.vhtml
It's true, I've got a lot to learn about my sch. Software.
Anyhow, what's happening is that I'll turn it on w/ the fast speed (rotary sw. bypassed), and it'll be at 2.1 volts on the motor, a nice fast speed. Then, about 30 sec. later, it winds down to about a volt, and slows to a "not very useable" speed somewhere between "slow" and "fast".
With the motor disconnected, the voltage stays at 2.1 volts
With a motor that ISN'T turning a baffle, it stays at 2.1 volts.
So, the baffle is causing the motor to try to draw more current and that's bringing the voltage down, or at least that's what it seems like.

Should I try a PWM system instead? Will I be able to get a nice slow speed (40-50 rpm) with a PWM? Should I get a different motor that has more power & slower speed? A servo motor? A geared motor?

I hesitate to get a different one because I've had this one do both fast and slow speeds but not in the same configuration. Anyone here a whiz with motors?

While this may not be a true stompbox question it's still an instrument effect that's controlled by a metal hammond box on the floor with a big stomp switch that resembles a silver nipple. Ought to be enough criteria, right?

Thanks.