How are you measuring... it may be your measurement method account from some of the error.
I did you use a standard AGD dial indicator (DI) or a Dial test indicator (DTI). A DTI is a better tool for measuring TIR but not many wood workers will have one.
How is the DI oriented to the arbor?
If the stem of the indicator is at even a slight angle to the base you've introduced what's commonly called Cosine Error. In simplest terms if the measurement is taken on an angle, the indicated values is larger than the actual value.
You can correct it by using this chart.
Correction Factor Table
Angle Degrees Multiply by
10 - 0.98
20 - 0.94
30 - 0.86
40 - 0.76
50 - 0.64
60 - 0.50
I'd start here:
What
The bearings on a Table Saw arbor are designed to allow it to rotate freely but not move side to side. As bearings wear, this side play becomes more pronounced and will cause vibration.
Why
Side play can affect the accuracy and reliability of meansurements needed to align the blade. Vibration from side play will produce a poor quality cut which will need further work (jointing) before glue-up. When the arbor bearings are in good shape, a well aligned saw with a high quality blade can produce an edge that is ready for glue-up without any further preparation.
How
Place the stylus of the dial indicator against the surface of the blade as shown in the photo above. The dial indicator is shown with an Offset Bar on the end of the plunger. This isn’t necessary, you can place the stylus anywhere on the surface of the blade plate. It doesn’t need to be near the table surface. Next, rotate the scale so that the large hand points to zero. Pressing on the sides of the blade will cause the reading to change. Basically, you are measuring how far the blade flexes as you press against it. Ideally, the reading should return to zero when you release the pressure. If it doesn’t, then you are measuring the amount of side play in the bearings. If it’s less than 0.001″ (one thousandth of an inch, one graduation on the dial), then there’s nothing to worry about. If it’s more than 0.001″ but less than 0.005″ then you’ll want to make note of it for future measurements (this play will influence the accuracy of your alignment). If it’s more than 0.005″, then you should replace the bearings (or forget about using this table saw for accurate work).
Assuming you've got tigh bearings, set us as before:
Start with the same setup you used to check for bearing side play. The Offset Bar on the end of the dial indicator plunger isn’t necessary, you can place the stylus anywhere near the edge of the blade. It doesn’t need to be near the table surface. Rotate the blade by hand and observe the reading on the dial indicator. If you are aligning a contractor’s style saw, then turn the blade by pulling on the belt. If it’s a cabinet saw, then you’ll have to do it by touching the blade. You have already seen how flexing of the blade can influence the reading so do your best to avoid it. Basically, you want to measure variations in the surface of the blade (its flatness) combined with any wobble in the arbor and arbor flange (run-out). You want to avoid changes in reading caused by flexing the blade or bearing side play.
When you locate the point with the highest reading, mark the spot with a felt tip pen.
Loosen the arbor nut so that the blade can rotate freely without turning the arbor.
With the arbor remaining still, rotate the blade half way around (180 degrees) and re-tighten the arbor nut.
Now, with the dial indicator stylus against the blade as before, watch the reading as you rotate the blade. Stop when you find the highest reading. If you find the spot that you marked with the felt tip pen underneath the stylus of the dial indicator, then the variations are being caused by blade warp. If the total change in reading by less than +/-0.001″, then your blade, arbor and flange are in great shape. If it’s more, then you might want to think about getting a better blade, but there’s no reason to be concerned with the condition of your saw.
However, if the highest reading is nowhere near the spot you marked, then the variations are being caused by wobble (run-out) in the arbor and/or the flange. The next steps will reveal exactly where the problem is.
Remove the blade and the throat plate so that you can place the stylus of the dial indicator directly against the arbor.
This close up photo shows the stylus resting on an unthreaded portion of the arbor. You won’t be able to obtain reliable readings if the stylus is on the threads. If your arbor is threaded all the way up to the flange, then you won’t be able to measure arbor run-out. Also notice that the flange is dish shaped. If the dial indicator were exactly vertical, then the plunger would rub against the outer edge of the flange. So, the dial indicator has been tilted slightly so that there is clearance between the plunger and the flange. Normally, your most accurate and reliable measurements occur when the plunger of the dial indicator is perpendicular to the surface that you are measuring. This is one of those rare instances when you need to break this rule. The tilt is slight so the effect will be corectable (see table above).
Next, you want to rotate the arbor and observe the reading on the dial indicator. It’s easiest to do this with the belt(s) removed. The change in reading should be less than 0.001″. If it’s more, then you should replace the arbor or forget about using this saw for precise work. This mode of run-out is caused by a bent arbor. It could be a defect in manufacturing but most often it is caused by a kickback (wood binding in the blade during the cut). The problem cannot be easily corrected in a machine shop. A shop which specializes in straightening crank shafts might be able to do it but the cost would easily exceed that of a new arbor.
In preparation of measuring flange run-out, tilt the arbor to 45 degrees. Place the stylus of the dial indicator against the outer rim of the flange. Notice how the dial indicator is tilted to match the tilt of the arbor. This arranges the plunger so that it is perpendicular to the surface being measured and will give you the most reliable and accurate readings. I’ve seen a number of people (authors, manufacturers, and other so called “experts”) demonstrate this measurement by leaving the arbor horizontal and tilting the dial indicator at a steep angle. This is a very poor practice that will cause the readings to be exaggerated. Now you know better than these “experts”.
The close-up photo shows the stylus tip resting on the center of the raised rim of the flange. This is the part of the flange that contacts the blade and is therefore the only part that matters. As before, rotate the arbor and observe the reading on the dial indicator. Removing the belt(s) will make the task a lot easier. You should see no more than 0.001″ variation. If it’s more, you can bring the arbor to a machine shop and have the flange faced on a lathe.
So, what happens when you do all these tests and find nothing wrong? The blade seems flat, the arbor isn’t bent, and the flange runs true but you still see excessive run-out when you measure at the edge of the blade. Look for a piece of dirt, saw dust, or grime on the flange, or on the blade where it contacts the flange. Make sure that all the mating surfaces are clean before re-assembling and testing again. The problem could also be a burr or ding on the edge of the flange. A couple of swipes with a file could clear it all up.
These photos show the tests being demonstrated on my mid-80’s vintage Delta Unisaw. It has seen really hard use during the past 20+ years. In addition to thousands of board feet of lumber, this saw has seen many tons of aluminum across it’s table. Even so, it’s still accurate and still passes these simple diagnostic tests.