The Fifth-cut Method and the Dial Indicator Method
Newly constructed Cross Cut Sled
The crosscut sled is probably the most important accessory that you will ever make in your woodworking shop. It transforms a simple saw table into a multi-purpose tool, not just for making accurate cross cuts and dados but, with appropriate jigs, to do miters, box joints, tenons and joinery. Its main benefits are accuracy, safety and repeatability. For these reasons, it is worth putting in the effort to make sure that it will always cut square to the fence (against which your workpiece or jig will be registered). Furthermore, the miter gauges commonly supplied with table saws are usually not up to much, with runners that wobble in the track, and do little for accuracy, safety or repeatability.
I made a new sled recently for my table saw and it struck me that it might be worth doing a post on the methods used for calibrating the front (or near) fence of the sled. There are tons of tutorials online on how to actually make a crosscut sled so I didn’t feel I needed to go there. But beginners to woodwork may not fully understand the business of fence calibration so, hopefully this post will help.
I will deal with the 5th Cut method first (because it is the most laborious method and we might as well get it over with). If you happen to own a dial indicator, you can opt to skip this section and go straight on to The Dial Indicator Method.
Where the fence is positioned perpendicular to the blade (as it is when using a crosscut sled) the most conclusive method of calibrating the fence was generally, the “Fifth Cut” method. I say “was” because in recent years, dial indicators have been commonly used for this type of fence calibration but more on this later.
I do assume, by the way, that your table saw is properly calibrated: the blade must be parallel to the mitre slots and the blade perpendicular to the table. If your table saw is not properly calibrated in itself, there is little point in trying to calibrate the crosscut sled fence until the saw is put right.
Before beginning the Fifth Cut Test, square your sled’s fence to 900 with the saw’s blade as accurately as you can with a good quality, large square. No matter how accurately you do this with the square, you will find that the Fifth Cut Test will discover an error. Once you have squared up the fence as best you can, ensure that it (the front fence) is securely fixed or clamped in place for the actual test.
In this system of fence calibration, a board sample (MDF or plywood is good), approximately square or rectangular, is used in a number of sacrificial cuts in order to establish the extent of the error in the alignment of the fence of the crosscut sled. The sample should be as long (front-to-back) as the particular sled can handle. The sides are marked 1,2,3,4 in an anti-clockwise fashion and proposed cuts marked about ¾” in from the edge. The actual cuts are taken off each side in a clockwise direction, starting at side #1. The rule to follow when cutting is that the freshest or last cut side is the one that is registered firmly against the fence for the next cut. If you have numbered the sides correctly, this should mean that you are cutting the sides in numerical order.
A rectangular sample of plywood marked up for cutting.
Once each of the four sides has been trimmed in this way a further and final cut is taken (another ¾” in) from the edge of side #1. This is the renowned “5th cut”. Just before taking this cut the “away” end of the strip that is going to be cut off should be marked A. The inner end (closest to the fence) of the strip should be marked B. The offcut from this final cut is saved for measurement; the first four offcuts are discarded. [The reason for taking the fifth cut is that a digital caliper will not generally be large enough to span the width of the board, even after taking four cuts, but by taking a ¾” slice off the board and measuring it, the error will still be revealed].
Now the width of the 5th cut piece at the A and B ends are measured carefully with a digital caliper and the values noted. It is essential to place the sample and the caliper down on a flat surface while taking these measurements to avoid any skew in the contact between the caliper and the sample, which could lead to an erroneous reading.
Two more dimensions are required before we can calculate the error ratio. One is the length of the 5th cut piece which is, the “length of the cut”, which we call L. The other required dimension is the distance from the pivot point of the fence to the exact point at the free end of the fence where the adjustment will be taking place. This is usually close to or at the very end of the free end of the fence. Nominate this final dimension as D.
Armed with this set of values (A, B, L & D) we can now apply a formula to give us the error ratio that we require for the adjustment that is needed to the fence.
This formula, for calculating the error ratio in a fence that is perpendicular to a blade kerf, is based upon the geometry of cumulative error ratios. It is not difficult to understand that at each of the four cuts that we made on the sample, any error in the previous cut is increased each time; doubled in fact. Moreover, the longer the cut and the longer the fence, the greater the error (if there is one).
The formula is e = [(A – B)/4 ÷ L ] x D
e is the error ratio
A is the width of the outer end of the 5th cut sample
B is the width of the inner end of the 5th cut sample
L is the length of the 5th cut
D is the distance from the pivot point of the fence to the point of adjustment of the fence.
The value 4 represents the four sides (or, more particularly, the four angles) of the sample.
If e is a negative number, then the free end of the fence is too far back or too “low”.
If e is positive in value, then the free end of the fence is too far forward or too “high”.
To correct the error, we must move the fence either forward of or backward (depending on whether e is negative of positive) by the amount e. To do this, a feeler gauge is commonly employed.
Why not simply move the fence by (A – B)/4 ?, you might ask. Will not simply dividing the error by 4 give us our correction? Not quite; there is a little more to it. The term (A – B)/4, gives only the error of the cut without reference to the length of the cut or to the length of the fence (pivot point to the point where the fence is being calibrated). Thus, the requirement for the second part of the formula; the division by L, which gives the error per inch of cut, and multiplication by D which gives the error ratio for your particular fence length. As the error will increase with the length of the cut and/or the length of the fence, these two factors cannot be ignored and must be accounted for in our calculation.
To use the feeler gauge, proceed as follows.
First, mark the point on the fence where the correction is to be made with a pencil.
Where the calculated value of e is negative:
The fence must be moved forward (assuming you are making the adjustment on the LHS).
So, with the fence securely clamped, place a feeler gauge of the value e (or one closest to that value) at the pencil mark on the fence and then place a pointed hardwood batten snugly up against the feeler gauge. Clamp the batten at this position. Loosen the fence, remove the feeler gauge and move the fence forward tight to the point of the batten. Clamp the fence and then secure with a screw from underneath.
The full test can now be carried out again with a new board sample to check that the fence correction has been successful. Most woodworkers aim for a corrected value of e = ± 0.001 of an inch. (a 1000th), which occurs when the 5th cut piece is measuring the same at both A and B. This is achievable with some effort.
Where the calculated value of e is positive:
The fence must be moved backwards (assuming you are making the adjustment on the LHS).
Place the pointed batten snugly against the fence at the marked location and clamp it down.
Now loosen the fence clamp and place a feeler gauge of value e (or one closest to that value) between the point of the batten and the fence. Hold the fence snugly against the feeler gauge and clamp it down. Now secure the fence with a screw.
Note that where the feeler gauge does not exactly match the value e, we are left with no option but to use the feeler gauge of value closest to that value. This will result in a small residual error. How do we deal with this? The solution is to repeat the test and adjust the fence again. The adjustment this time will be smaller than it was with the first test. A number of tests may be required until we get to an e value of = ± 0.001 of an inch.
Once you are happy that the fence is properly calibrated secure it with a number of screws drilled and countersunk from below.
Let us do a hypothetical calculation.
We are at all times working in inches, decimalized.
Say A = 0.850
And B = 0.839
A – B = 0.011 divided by 4 = 0.00275
If the length of the cut is L = 24, we divide by this to give 0.00011458
Now, assume fence length of D = 28.5, so multiplying 0.00011458 by 28.5 gives 0.00326
Or just over 3/1000ths. As we do not have a feeler gauge of 3.26/1000ths, we can only adjust the fence (backwards in this case) by 3/1000ths and a further run of the test will be required to correct the small residual error.
The Dial Indicator Method
It is also possible to calibrate a crosscut sled fence using a dial indicator. There is a certain elegance and simplicity to this system but of course, you do have to own a dial indicator to do it. It obviates the need for repeated 5th cut tests and will usually bring a faster result. There is also no need for math. The setup is shown below. The dial indicator is held in place on the saw table with a magnetic base. Using the longest engineering square that you have (that will fit in the sled). Wipe the edge of the blade of the square to ensure there is no grease or dirt on it. Touch the indicator off the square while holding the square firmly against the fence. Use the bezel to zero the indicator and then move the sled forwards. If the dial moves, the fence is out of adjustment. Ideally, the needle should hover closely around zero.
Bird’s-eye view of the Dial Indicator set-up. Magnetic base fixes the unit to the saw table.
An engineering square is held against the fence while the sled is pushed forward.
A longer square than this would be better.
Note: the construction of the pivot point of the sled is clearly visible in the picture.
Note also, the hardwood face on the fence.
It is necessary to build a simple micro-adjuster into the sled behind the adjustment point, that is, the point furthest away from the pivot point to complete this test. Without a lead screw to make tiny adjustments you will find yourself trying to move the fence back and forth manually time and again and continually miss the target. Nothing fancy is required; just a small, firm bracket drilled and threaded to take a machine screw. The micro-adjuster helps in making the tiny adjustments required and can be used again in the future if the fence should fall out of alignment. I usually include one when I am building a new sled.
For both type of test, the required pivoted end to the fence can be achieved by tightly fitting a bolt through the fence and bed of the sled (countersunk underneath). This pivot point must not be loose or sloppy or it will not be possible to make an accurate correction to the fence. Make it tight enough that the bolt must be screwed into the pre-drilled hole. In the event that the fence needs to be re-calibrated in the future, the wood screws that were driven in to permanently secure it will have to be removed but the pivot point and the micro-adjuster will still be there to be used again.
Since wood can move in tangential and radial directions, seasonal tests on the fence are well worthwhile. For this reason, I never glue the front fence to the base of my sled and as stated before, fit a micro-adjuster and a bolted pivot point on any crosscut sled that I make. Moreover, if the fence becomes distorted in any way through seasonal movement, it can be replaced. I usually construct such a fence from a series of layers of MDF (because it is relatively stable) and then glue a prepared hardwood face onto the face of the fence to achieve a good registration surface.
One of the main problems experienced by new woodworkers when making crosscut sleds is slop in the mitre slots. There may be side-to-side wobble in the sled as it moves forward on the table. This causes the sled to rack as it moves across the blade leading to off-square cuts. It is usually caused by inexperience or lack of care in making the runners; they may be a smidgen too narrow for the slots. If this is a problem, you will not be able to calibrate the fence properly and your sled will continually produce cuts that are off-square.
If your sled has slop there is a way of correcting this problem. Disconnect the bed from the fences. Lower the saw blade. Now place the two bed halves on the saw table with their runners in the slots. Gently clamp the two halves together with a pair of long clamps so that the runners are riding snugly against the blade side (or inside, if you like) of the slots. Not too tight. Now glue down the away fence to the bed halves and allow the glue to dry. Remove the clamps and carefully pull the sled back until the near end is overhanging the table a few inches and place the fence on top. Re-bolt the pivot point. Put a few screws in from the bottom to secure the fence. Push the sled back and forth (with the blade lowered) to ensure that it is moving sweetly without slop.
Assuming it is, you can now raise the blade and re-rip the kerf. Now remove the screws from the near (or front) fence and square it to the blade as best you can with a large square. Once squared, clamp the free end of the fence. You can now go through the procedure for calibrating the fence before inserting screws again to secure it. Don’t use the same screw-holes as before (they may tend to pull the fence out of alignment). Drill new countersunk, pilot holes from underneath and screw the bed to the fence.
This procedure will probably result in a wider than acceptable kerf in the bed of the sled, that is, it will no longer be zero-clearance. To solve this problem, you can glue an auxiliary hardboard bed onto the top of bed of the sled and when the glue has set, re-rip the kerf by running the sled across the blade.
The corrective procedure detailed above will also work if you are having the opposite problem – the sled is binding in the mitre slots when you push it forward. Clearly, if the sled is binding, fence calibration will be impossible. In this case, the runners were probably not aligned exactly parallel. Don’t beat yourself up over this; it does happen. In this case, you will definitely need to fit the auxiliary hardboard bed after the corrective work as the kerf in the sled bed will likely be skewed. The new hardboard bed camouflages the cock-up and allows a new zero clearance kerf to be ripped. There may be a little tidying up to be done on the outside of both fences as the bed may now slightly overrun the fences at one end. If this is the case a few strokes of a hand plane should sort that out. Nobody will ever know…