Powermatic belt sander tracking

A customer tasked me with repairing, if possible, tracking trouble on a 6 x 80 Powermatic belt sander. Apparently the mfr is aware of the problem because they actually replaced the machine.
I have some guesses about where the trouble may lie, but I will wait to see if anyone else has confronted or solved this trouble, before I poison the well with my speculations

I have had the same machine for over 15 years and have found that the tracking is very over sensitive, in a conversation with powermatic it was pointed out that because of the short distance between centers that this sensitivity cannot be corrected completely, also the movement of the motor to move the belt up and soen compounds the problem, I have found that as the belt is used more it becomes more stable

Any belt sander I have seen that had belt tracking issues had no or not enough crown in the non driven drum. On a 6" wide belt sander you want the center of the non driven drum to be about .080 to .100 bigger than the ends. This will correct a over sensitive tracking issues.

Moonlight machine is absolutely correct - On belt sanders (or any other flat belt machine, really), the pulley crown is crucial.
If the idler pulley is flat, you'll never get it to track right - You can check this out, by simply wrapping a few layers of tape around the center of the idler. I can almost guarantee that you'll see it improve.
A more permanent solution is to have it machined - It doesn't even have to be a perfectly rounded crown, but just leaving an inch or so width at the center high, and the sides sloping in a straight line will do it.

Thanks for the replies. I agree insufficient crown is part of the trouble. As-built the only "crown" is that the center 2 1/2" of the idle pulley is .003" bigger than the ends. There is no smooth crown only those .003" steps.

But I see more problems with the design

The oscillation is driven by rocking the motor and drive pulley assembly. But the pivot point is several inches below the center of the driving pulley, so in addition to skewing relative to the CL of the belt, the drive pulley rises and falls, and the belt tightens and loosens.

Does anyone else agree that the rise-and-fall, tighten and loosen incidental action of the rocking of the motor, intended to drive the oscillating of the belt, is part of the problem? Much more work to correct than just cutting some crown on the idler.. I want a definitive fix, but the customer does not want to pay for too many design interations.

My guess is that the crown is missing so the idler pully will be better to use for internal radius sanding. And I thought the best oscillating sanders moved the entire platen/motor/idler assembly up and down.

Crown idler helps with tracking, but it does stretch the belts center and can make the belt bulge slightly on the platen. Best to remove tension when not using it. Something I never do.

Good point about the reason for minimal crown being to make the idler more useful as a drum-sander. Compromises.

I had a maintenance contract for a while in a shop with a Timesaver (huge wide-belt sander). The Timesaver belts oscillated, but the design did not ask for stability, but used broken light beams to reverse the tracking between limits. And that gave a fair amount of trouble.

Here is the rest of the story.

I offered a significant re-design to the owner, including adding crown (minor) and building a new motor cradle to put the center of nutation at the center of the driving drum (major). Cost was more than he wanted to pay (since he already had been given a new Jet equivalent as a replacement for the Powermatic.

So per the original deal he offered me, the problematic Powermatic became mine.

And suddenly a craft project of my daughter appeared, that needed a belt-sander (making chess-boards out of log slices)
So now I had to make the thing usable
First, I added crown, about .020 on both driving and driven drums (several very high-tech helical wraps of electrical tape)

Then I completely removed the gearbox that drives the cam that rocks the drum from the back end of the motor. That included the fan cowl, but over a couple of hours use the motor only got sligthly warm.

The cam action was opposed by a spring. I replaced the spring with a little pipe spacer bushing, so I could adjust with shims and lock the drum angle solid.

The belt (after a little adjustment) now tracks well enough. Shifts a little as you bear down on the work, because the long motor shaft that carries the driving drum deflects. Needs an outboard bearing. Only two other drawbacks remain. One is that the cam used to loosen and re-tension the belt for installation has too little travel,and the spring it drives is too short. There is essentially no allowance for slightly differing belt lengths. Also the motor is feeble..not too hard to stall it, sanding a 12" diam oak disc.

But it is now a usable machine..