Here is an overview from the driver's side. The PTO driveshaft comes in low underneath on the driver's side, and drives a worm gear at the bottom of the gearbox. The worm gear is permanently engaged to a big gear on the end of the drum shaft. It is very important to understand that the cable drum spins freely on the shaft. It is not permanently locked to the shaft, allowing the cable to be pulled out in free-spool mode. A clutch mechanism on the passenger side of the winch engages the drum to the shaft for pulling. With the clutch engaged, the drum cannot backdrive the PTO driveshaft due to the action of the worm gear. A drum brake is fitted, but only to keep the cable from unreeling with the clutch disengaged. The friction of the brake is light enough so you can manually pull the cable out. It is not designed to restrain any sort of load on the cable.

My winch was damaged when I purchased it. The clutch mechanism had been broken off from the A-frame that supports that end of the shaft. Before I explain what I did, let's take a look at the stock configuration. The following images are used with permission by Josiah Bartlett. They have been edited for clarity.

The passenger side of the drum has two large protruding pins. A sliding clutch dog is keyed to the shaft. The clutch dog is free to slide back and forth, but cannot rotate on the shaft. The clutch dog has a pair of slots to engage the pins on the end of the drum. Note the forks on the clutch lever, which engage a slot around the perimeter of the clutch dog. In this photo, the clutch lever is down, and is held that way by a spring-loaded lever lock pin above it. You can see how the clutch dog is away from the drum, so the pins aren't engaged. The drum is free to spin on the shaft, with only the free-spool brake to keep the cable from unwinding:

Here the clutch lever has been raised, shifting the fork and clutch dog toward the drum. The clutch lever is above the lever lock pin now. Note the oblong slots in the clutch dog to engage the pins. The ends of the pins are not visible in the shadows. The drum is now ready for winching. You can see the wear marks on the end of the drum from the free-spool brake. In this picture, the brake is not riding fully against the drum because the A-frame is backed off slightly:

The stock free-spooling brake is an ingenious device. This image shows the side of the A-frame that faces the drum. Note the heel of the clutch lever, which is below the lever pivot point. When the lever is down to the disengaged position, that heel moves towards the drum. The heel pushes on a spring which pushes the brake pad against the drum. When the lever is raised to the engaged position, the heel retracts and spring tension on the brake pad is released. What happened with my winch is the cable jumped the end of the drum and wrapped around this mechanism. It ripped the mounting ears from the A-frame, and most of these pieces were discarded. A jury-rigged version was then built on top of the bumper:

I called my bank and confirmed that nobody had recently set up a generous trust fund in my name. I would not able to afford these hard-to-find parts, if I could even find them. A broken clutch lever mechanism is a very common problem, making the parts rare and expensive. I was also worried about driving off with the winch accidentally engaged, so I went into DeepPonderMode(tm). A few months and much coffee later, here is what I came up with. A lollipop shaped Clutch Arm, formed from .125" x 2" bar stock, has a pair of pins sticking inward that mimic the stock fork arrangement. The Clutch Arm pivots at a fixed bracket at the very bottom. Midway up the Clutch Arm is a pushrod to move the arm and engage the Clutch Dog against the drum. Inside the hollow of the A-frame, and very difficult to photograph, is a simple bellcrank. The bellcrank is rotated by a shaft and lever where the lever lock pin would have been. In this image, the lever (left side of picture) is straight up, moving the tip of the bellcrank down. The bellcrank is held in this position by a spring, not seen, which travels overcenter relative to the pivot point. The bellcrank and pushrod will be explained in the image after this one:

Here the lever is moved so the tip of the bellcrank is up. Once again, the not seen overcenter spring holds the bellcrank in this position. The barely visible pushrod is a piece of .125"x 1" bar stock with a bolt hole at each end. The connection point between the bellcrank and pushrod has traveled up and over center, so the whole mechanism is locked in the engaged position. Remember when you laughed at your math teacher and said you'd never need to know all that stuff? Well, that is how I figured out all the angles and travel necessary. This was not guesswork or trial and error. It was fun to actually put that math to good use and see it work out exactly:

Click here for a bitmap rendered in glorious TooMuchFreeTimeVision(tm) showing how the bellcrank and pushed operate. This will also give you an idea how the spring is arranged to travel overcenter to hold the bellcrank in either position. Not shown is how the bellcrank was carefully shaped so it hits the underside of the A-frame casting at either end of its travel. Setscrews in the edge of the bellcrank at each end allow fine tuning. The following image is from the end on the passenger side. The lever is straight up to the disengaged position. The travel adjusting setscrew is hitting the underside of the A-frame. There was no way to get a picture of it, but at the inboard end of the bellcrank is another setscrew to adjust travel in that direction. Here you can also see the spring that holds the bellcrank in either direction:

Part of the design was that the lever could be used as a storage point for the hook on the winch cable. If the PTO drive and winch clutch were accidentally left engaged, the cable would pull the clutch lever to the disengaged position. The top picture on this page shows the hook stowed for driving. In practice, it would probably just bend the lever but it is nice in theory.

Now, what about the free-spooling brake? I did not replicate that portion in my new design. With some more coffee, I came up with a very simple brake. Remember, the only function of the free-spool brake is to apply a light friction to keep the cable from unwinding. The friction must be light enough to overcome to manually pull out the cable. Way under all the cable is a hole that I drilled and tapped to the center of the drum. The threads only extend partway down. At the bottom of the hole, against the shaft, is a small piece of brass rod. Above the brass piece is a short, stiff spring, with a setscrew threaded in above it. By adjusting the setscrew, I can change how hard the brass rod presses against the shaft. This is now my poor man's free-spool brake. With the shaft not turning, I set just enough friction so I can pull out the cable by hand. When the shaft is turning and the winch clutch engaged, there is no relative movement between the drum and shaft. The free-spool brake has no effect whatsoever then. But best of all, it works great for rewinding the cable after winching. By leaving the winch clutch disengaged but with the shaft being driven, the free-spool brake has just enough friction to pull in the cable. I can guide the cable in by hand, and even pull it back out for adjustments with the shaft still driving inward.