Occasionally, I’ll get an idea and it will nag at me until I try it out. For better or worse, this build is a perfect example.
The basic idea is to make a band saw that uses a series of smaller wheels to replace the upper wheel.
These smaller wheels would follow the natural arc of a much larger wheel to (supposedly) reduce the bending stress on the blade.
Of course, there will be more stress on the blade from this, but I think the benefit of the extra capacity in a tighter package might be worth the trade off.
Along with this extra stress on the blade, the build complexity goes up quite a lot, since there are a lot more components involved. There are more advantages and disadvantages, but those are the big ones.

Since I wanted to make this quickly, mainly to test how well the blade would track on multiple wheels, I kept it simple. I started with a large piece of melamine (29″ x 42″) to mount the parts on, then decided on the wheel size that would best fit in the space and still work with the only blade length I have (120″).
I marked out two 12″ wheels with my beam compass from 3/4″ plywood. These are smaller than they should be for the thickness of the blade I’m using (0.032″), but should be fine for a quick test.
I also rounded up some other parts – roller blade wheels, a motor, some bearings and a piece of 5/8″ shaft:

The radius for the upper “wheel” is bigger than the max capacity of my beam compass, so I just cut a stick to the right length and use a finish nail as the pivot point to draw the circle. I then laid out the axle locations for the roller blade wheels and drilled and tapped the holes directly into the particle board:

Nothing really fancy for the wheel mounts, just strips of plywood screwed through spacers to the backer. The axles for the wheels are just 1/4″ x 2″ machine screws:

Again, this spacing for the wheels is not ideal, either. It would be better if they were smaller and tight together around the curve, rather than spaced. But, I didn’t want to go to the trouble of making a bunch of wheels, and all I could find was seven of these.

With those attached, I put the blade on and clamped the two 12″ wheels in place to see how it looked. I did some rough layout in SketchUp before this, but nice to actually see that it fits.
Next up, I needed to mount bearings in the idler wheel. I used my compact compass to draw the outline for the bearing I would use, then drilled them out close to that size. Since I don’t have this exact size bit, I sanded the hole to the right size with my spindle sander:

I could then push the bearings in and secure them with washers that I screwed on that lip over the edge. I used a 5/8″ bolt for the shaft and two washers to space the wheel away from the backer:

The other 12″ wheel drives the blade, so I needed to mount that rigidly to the 5/8″ shaft. I used a 3″ pulley and some counter top nuts to clamp it onto the wheel:

The goal is to get this done fast and reliable, without a lot of messing around to make it look pretty.

I had two of these cheap pillow blocks, but could only find one, so I made another with a regular bearing:

To adjust blade tension, the drive wheel pivots on this long arm that the first pillow block is bolted to. At this point, I’ve done all I can with the backer laid flat, so I screwed on a big scrap of plywood to support it and stood it upright to finish working on the motor assembly:

The motor has a 2″ pulley on it already, and I just needed a bigger one to put on the drive wheel shaft. I made one quickly on the lathe:

That’s a tip for getting the 5/8″ shaft hole centred – drill part of the way in before mounting it on the lathe. When the turning is done, you can the finish drilling the hole.

It takes very little time to turn a small pulley like this:

I used the same method I used on the drive wheel to attach this pulley.

I didn’t have a rubber belt the right length, but these link belts are great to have on hand for this kind of thing. Even if I wanted to use a rubber belt in the final build, the link belt lets me test and can be used to figure out the exact length needed:

I won’t lie, it took a while and a fair bit of messing around to get the blade tracking. But once it was, it was completely stable. That answered my main question about tracking, but there was also the question of how well the blade would hold up to the extra bending stress. To get an idea, I left it running for what turned out to be all night.
I checked it after it was running for an hour and was pleased to see it was still on and still unbroken. Then I kind of forgot about it until around midnight, but when I looked in on it, it was still going strong and was still going at 7 the next morning when I went back out.

That was a surprise. I fully expected the blade to be off (or broken) the next morning. I turned it off, made the video that’s at the bottom of this page, then turned it on again while I cleaned the shop (as seen in the video). I left it running, but when I came back out again after about 30 minutes, the blade was broken.

The conclusion I can draw from this test is that the idea would work – the blade tracking is not really an issue and blade breakage is likely not one either. Like I said at the beginning of this article, the conditions were far from ideal and using the two 12″ wheels probably over-stressed the blade more than the series of spaced smaller wheels. Using a tighter group up top and a single, larger diameter wheel on the bottom would be much better.
However, I don’t think the potential benefits are worth the extra complexity. There isn’t a lot gained from getting more capacity from a shorter blade, since it’s always better to use the longest blade you can.
It was an interesting experiment, though, and certainly a novel one, for me at least. I’m not sure if this has ever been done before (probably, you name it, it’s been done for most woodworking tools), but I don’t regret the day I spent putting it together.

Here’s a short demonstration video going over the main points: