I am making a tablesaw sled and am using William Ng’s 5-cut method to square the fence. The sled is ¾ BB ply and the fence is 2 ¼ wide laminated hardwood. There is absolutely no wiggle in the sled when pushed through. I set the blade to 90.0 degrees on my PowerMatic tablesaw before starting. The blade is a Forrest 40 tooth with not a lot of use so far.
My problem is that I am getting different readings each time I do the 5-cut. I have used ¼ BB, ¾ BB and ¾ MDF as the test piece. The height of the 5th cut is at least 9 ½ inches. I am using a Starrett thou calipers and a good rule the reads to 1/6 inch but it is easy to read to the 32. I check for dust on the sled before each cut.
My test result over a 21 inch fence points range from .006 to .014 to .020.
Does anyone have any suggestion as to what is causing this variation?
And yes, I was tempted to stop at the .006 but since this sled will be converted to a precise 45 degrees for making boxes I wanted those cuts to be as close to 90 degrees across the miters as I can. So I double checked and got this.
Replies
I sounds like your blade maybe out of alignment with your miter sots (not parallel with each other). I found these two videos helpful in correct this problem.
Table saw not working right? The proper way to adjust/align it..
https://youtu.be/tNfvtStKkFA
How to accurately set your table saw bevel angle
https://youtu.be/8JebPTBB8wI
Forget about using a rule; you need to be able to read down to thousandths.
When you measure with the caliper, make sure you rest it flat on the bench and squeeze the arms firmly against the test slice. A very tiny tilt to the caliper than make a big difference in this case.
When you say you got .006 over 21 inches, that corresponds to a difference of .0027" over the 9.5" test cut, is that what you measured? In other words, your fifth cut left you with a slim piece 9.5 long, and one end was .0027 wider than the other end, correct?
Also, are you sure you are moving the fence in the correct direction? Jumping from .006 to .014 sound like you corrected in the wrong direction, doubling the error instead of canceling it out. It the leading edge of the fifth cut slice is narrower than the trailing edge, then your adjustment should move the far end of the fence toward the saw table. So you would place your .006 shim between the fence and your stop, then remove the shim and move your fence against the stop. If the leading edge is wider than the trailing edge, you set the fence against the stop, loosen the fence and insert the shim.
Don't ignore williams advice about using a completely different hole every time you move the fence.
Finally, did you joint the face of your fence? A very slight bow in the fence will throw you off; that's why William laminated plywood for the fence against a very good known flat surface.
Carbide_Tipped's last item is one I would confirm before I moved ahead. You say it is laminated hardwood. Are you saying it is a glued up lam that yo made? Whether you made it or not was it face jointed and then edge jointed on the surfaces that count?
If the fence checks out, how is the bed? The base, that is. Does it set flat on the table top? Plywood is notorious for not being flat. That being said, my bases are all 3/4" BB ply and have been reliable for years.
If the base and the fence bottom and face are all OK, check your blade to miter slot alignment. If your miter slots are not parallel (I know, why would that be but, it has been reported more than once) you may want to switch to your left slot.
I say left since I assume that is where your miter gauge runs. Be sure to check blade alignment from the slot you use for your miter gauge. Let's get through all that and see where we are.
My sled covers both miter slots, has a runner in each and runs without any wobble whatsoever.
I removed the fence which is three lengths of 1x poplar glued together and re-jointed the face and jointed the base to 90 deg which I tested.
I re-tested the right hand miter slot to the blade (as Stubby recommended) and it was less than .002 front to rear which is very close to what I measured in April.
I re-ran a 5-cut and got -0.0228 over the 21 ⅜ fence points. I did a second run and got -0.0115676. I had not moved the fence.
I don't have any tool to measure the length of the sample to thousanths and the fence points to thousanths so I did the calculations assuming I was off 1/32 in both directions for both the sample length and fence points. My rules are good enough to tell me if I am off by that much. The difference to the basic calculated error doing long/short short/long was less than 0.0001 inch.
The first sample was 11.250 long; the A measurement was 1.126; the B measurement was 1.174; the fence points were 21 ⅜ apart.
The second sample was 10.625 long; the A measurement was 1.028; the B measurement was 1.051; the fence points were 21 ⅜ apart.
And I lay the caliper on the bench and press it as learned from watching William's video. Re-measuring the samples gave the same measurements.
Comments? Help?
When you have a set of information that doesn't reveal a weak point you have to get pretty detailed in your search. Sounds like you have a good solid build. How is the fence fastened once in place? Is it held at each end, 21" apart? Poplar is quite soft but, the lamination seems like it would add adequate stiffness for use between 21" centers.
You could eliminate this flexing possibility by clamping a section of "straight edge" to the fence; avoiding the blade path obviously. A piece of extrusion or a milled strip of hard maple or some such would do. These may seem like grasping at straws ideas but, your info states things should be OK and they are not so, something transient is happening somewhere ;-)
P.s. Blade runout would be revealed by a wider than expected kerf so you could check for that. Arbor bearings, etc.
This is almost embarrassing, but I figured out what I was doing wrong.
I have made several sleds so I was used to the process, but habit looses understanding. All my other sleds were strictly crosscut so were wide and deep. But this one is for miters so the left side is shorter than the right side since the bulk of the material will be on the right side of the blade.
So I did my 5-cuts on the right side of the blade. Yes, I then used the technique that I was used to from the left side: I turned the sample clockwise. That meant I was cutting random angles instead of consistent angles.
I just did two samples on the right side turning them counter-clockwise so that the fresh cut side went against the fence. They gave me a -0.032 and a -0.30. It is so big because my incorrect first sample sent me into la-la-land.
Thanks for the advice, it was all proper and reminds me to watch William again.
I made the needed adjustments and now have my sled error at 21 3/8 of +0.00106. Repeatable within in 0.001.
Excellent. Thanks for coming back to close the loop.
This forum post is now archived. Commenting has been disabled