Conflicting crosscut sled accuracy problem

[Dana Fujikawa]

Help! I've spent the last year trying to get consistant 90 degree cross cuts on my table saw with my large crosscut sled and it's driving me insane. The problem I'm running into is that I can get my smaller test peices to cut at 90 degrees - smaller being about 20"x20", but when I goto cut large panels around 3 by 2 feet, the cut is off by a hair.
My sled is a single runner design with the fence mounted on the back and the sled base is around 4x3 feet of 1/2" mdf. After test cutting a large panel, the end where the edge that was against the fence tends to be a hair longer than at the other edge which was the last to pass by the blade. Heres the test I'm doing: I cut one end, flip the peice over end for end keeping the same side against the fence, and then cut the other end. I then repeat this process on a second test peice. After I'm done cutting I take the second peice, flip it side for side, lay it on top of the first peice and then align the peices. Finally, I run my fingers along the ends checking for any differences between the lengths of the two peices which effectively doubles any errors - and more importantly - exposes any tiny errors that can't be seen visually. During each test pass I take off about 1/8" of an inch or so at each end.

Originally I suspected a problem with either my plywood fence or the UHMW runner. So in desperation I bought an Incra Miter 5000, rippped off the fence/protractor system as well as their adjustable steel runner and integrated them into my sled. I also added metal re-enforcements to each side of the slot in my 3 foot outfeed table to ensure that the sled remains straight after fully travelling off the back off the saw. The sled is therefore rock solid in the slot throughout the whole length of the travel. Even with all of these changes I still have the exact same problem. I then suspected it might be blade flex, so I also tried switching from a thin kerf to a regular kerf blade. Again same problem. The blade is parallel with the miter slot and I see no signs of the test peice shifting as I'm cutting. Off square end cuts appear to be straight which seems to rule out any sort of blade flex or runnout problems.

This is extremely frustrating, because as far as I know I've covered all of the variables and in theory all reasonably sized panels should cut exactly the same way with this configuration. At this point I"m not sure what else I can try. I could tune the fence to square up large panels, but then smaller panels would be out. Now that I think about it, it's entirely possible that correcting for the large panels could be the right way to go, and that maybe it's the smaller panels that are wrong.....

Any suggestions on what else I can try?

[Matt Edwards]

Based on everything you have already done, I'd tune the sled by adjusting it to the largest size panel you can or will cut with it. In theory that is the most accurate way to do it, if the large panel is square, then the small has to be.

HTH
Matt

[Lee Schierer]

Your cut description indicates to me that your sled is twisting as you make the cut. Perhaps the runner is bending (flexing) or is moving relative to the mdf.

Try waxing your table saw table and make the same cut. Reducing the drag on the underside of the table will lessen the tendency for the bar to flex.

[Glen Butler]

If you are certain your sled is staying parallel to its original tragectory along the entire length of the tragectory then the next thing to look at is fence perpendicularness. There are a few reason a large panel could be off while a smaller panel is not. Larger size will inherently show imperfections in both sled runner straightness and fence perpendicularness. It is important that the fence it as straight as possible. Even .001 off will result in .002 after you flip the pieces as you indicated you do. That is enough to feel and see. And IMHO it is not enough to worry about. How much are we talking?

[Pete Bradley]

Does it matter? From what it takes to detect this, it's not clear that the error is meaningful.

[Fred Belknap]

If you are using adjustable runners the difference might be when one adjuster is in the miter slot and when there is two adjusters in the miter slot. I think the Incra use a two point system.

[Mike Vermeil]

My experience with crosscut sleds is that you will never stop chasing them. Just when you think you've got it cutting accurately - WAM - it's out again. I finally gave up & started using automotive feeler gages to shim the fence accurate every few weeks or so.

[dan sherman]

Try the 5 cut method for squaring your fence, It's best method around imo.

http://www.thewoodshop.20m.com/five_cut_method_swf.htm

For a big sled, the three things you must absolutely get right are:

1. The fence must be square & I mean really square as in 90.00 degrees. If the fence is out of square by a tenth of a degree, a 3' cut will be off by 1/16".
2. The fence must be dead strait. That means jointing the board, and then double checking it,with a strait edge or another jointed board. A cup or bow of a few thousandths will cause inconsistencies in your work.
3. The fence cannot flex at all, as this can cause the same problem as a bowed fence. Thus the fence needs to be bolted (to allow tweaking over time) to the sled every few inches. I think mt big sled is on ~8" centers.

Based on what you have posted I bet your problem is with #2.

[Rick Davidson]

I am far from a expert, but I would guess it's off on the small pieces as well, but because it is smaller and therefore a smaller difference you are not noticing it. I would square it up to the largest panel and then see what it does to a small panel.

[Ken Shoemaker]

Is it possible that the outside end of your rear fence is slight concave??? If it were, your 20X20 panel might show up as good, but then the longer panels might show a slight deviation. Try shiming a large panel, at the blade, to see if it comes into square.

Good luck!! This can be very frustrating.... damhikt... Ken

[Dana Fujikawa]

Thanks for all the great replies. I put a precision straight edge against the Incra Fence and discovered that it does in fact have a slight concave shape to it. To prove if this was the cause of my problem I ripped a 6" wide peice of plywood spanning the length of the fence and ensured both sides were parallel. I then placed this up against the fence to "simulate" a longer peice and then did my test cuts with the smaller test peices against this plywood (I also clamped on a temporary stop block). Sure enough these smaller 20"x20" peices now cut exactly the same way as the large peices which is great news - in other words the cuts are now consistent. My plan now is to add an auxillary peice of 3/4" MDF to the fence and then shim it. With the cuts now consistant it will just be a matter of tuning the fence to 90 degrees. I'll write in again once I prove the out.

On a side note: my sled originally had a wooden fence - which was based on that from FWW's Ultimate crosscut sled - which was in fact dead straight. However I was using an UHMW runner at the time which I suspect had some play in it. I beleive this runner play is why my original system also had this problem.

Therefore there are two main learnings here:
1) Make sure the fence is dead straight and test it with different lengths of peices. The Incra fence is aluminum so it should remain the same shape over time, however the MDF and paper shims I plan to use could potential change shape in the future. So it will be critical to constantly re-check for straightness.
2) Ensure that the miter bar runner has absolutely no play throughout it's full range of travel. Incra's miter bar has 5 sets of adjustable discs to eliminate side play, although it remains to be seen how well the wear. I like sleds with the fence on the back as it ensures that more of the runner is engaged at the start of the cut and also helps to counter balance the weight. Adding some sort of channel or re-enforcement on the outfeed table's slot is critical to ensuring the sled continues following the same path at the back of the saw and to also eliminate runner play at the same time.