For example, let's say that you want to make a simple half-lap joint with two workpieces, so that they join with both faces flush with one another. Hmmm ... you can feel a 0.001" mismatch, and easily see slightly more than that. If the faces are off just a little, then you've a chore ahead in scraping, sanding, and cussin' until it's right. But if you can cut to depth with precision, you can hit it "dead on", and very quickly.

To "feel" that joint into near -perfection so the faces are flush, you'll need to make at least 3 or 4 trial joints - unless you get really lucky - before you get it right by trial and error. Lotsa time and lotsa sawdust.

Worse, absolute measurement of blade height with some precision device may be - is most likely to be - misleading because the throat plate reference surface probably isn't within 0.001" of actual table height, and commercial blade tolerances may induce some wobble or may combine with a slightly undersized arbor to give you some vertical oscillation of the blade(if an arbor wasn't custom-machined to tight specs and retrofitted to the saw, you can bet that it's not full diameter, and only the most expensive blades have precise bores) . Did you measure at the high point or the low one? Maybe somewhere in between? Who knows? For those reasons, it's easier to cancel out all of those variables and just get within spittin' distance with a ruler or depth gauge, make a test cut, and then measure and correct for the error actually delivered from the cutting operation.

Good News - Measurement of degrees of rotation for the height crank can give you the ability to adjust height to within less than 0.001" in a couple of seconds - do a little simple arithmetic, crank in a few degrees, and you'll hit it dead-on.

The same is true for any blind cut that doesn't go through the wood. It's especially important for making curves or arches, so you can raise the blade into the wood exactly, rather than trying to fumble with lowering the wood onto a moving blade.

An aside:
Unless it's guaranteed to have been precision-manufactured, a height-setting gauge is most likely not accurate to within better than several thousandths - maybe much worse. If it's precision, then it's metal. For every good intention, there's an uh-oh lurking somewhere - do you really want to have metal tools on your saw table when the blade is exposed?

Note the scribed line at the left in the photo. Use it to read whatever the protractor says - any degree value. The protractor will turn 360⁰ for every turn of the blade height adjustment crank (or wheel, or whatever makes the blade go up and down), and it makes absolutely no difference where "0" is.

OK, on to methods:
The first thing you need to know is exactly how far the blade raises for each turn of the crank on your saw.
These next six steps will only be done once, so don't panic!

Determine the actual height change for your height adjuster for ten turns (a bunch of turns increases accuracy, and the number 10 is easy to work with.) To do that,

1. Place a straight piece of stock over the insert at the blade centerline, and raise the blade until it just touches. Back down and come back up slowly to be sure that the blade's height is exactly flush with the table. It's better not to simply feel for the "bump" - that's pretty rough. Instead, you can get close by the bump method, then back down 30⁰ and begin to creep up on it. Go back up about 20⁰, and then, going up in about 2⁰increments, raise the wheel a tad, place your hand on the wood, and gently turn the blade by hand to see if it makes contact. If not, try another 2⁰ - - repeat the turning, etc., until you've sneaked up on exact zero blade height.

2. Note or write down the degree value indicated by the protractor - doesn't matter what it is, just get the number.

3. Raise the blade ten turns - count accurately, and stop at exactly the same number of degrees that you noted at zero height. If you overshoot, back down 30 degrees and come back up again to the right mark.

4. In a piece of scrap 2x4, make a cut at that height.

5. Use your dial caliper to measure the exact depth of the cut - it should be something close to 1". However, it's probably NOT going to be exactly one inch - your dial caliper is probably going to read 1.062", or 0.947", or some such - it don't make no never mind - just read it and get the number.

6. Let's just make up a likely measured depth (be sure to use your own value when you do the math!), and call it say, 0.960"

   We know now that 10 turns delivered exactly 0.960" inches of blade travel. Well, 10 turns is a lotta degrees, and we only went that far to gain accuracy.

   To get the amount of blade travel for one turn of the crank, just divide by 10. For our made-up example, it would be 0.960" divided by 10, or 0.096" - that's 96 thousandths per turn - about a tenth of an inch.

   Now the fun: If we divide that 96 thousandths by the 360 degrees on the protractor, that's 0.096"/360, we get 0.00027". That is, each degree of turn raises the blade less than one third of a thousandth of an inch. Interesting, but now what?

   Divide 0.001" (the basic precision we're shooting for) by .00027, and learn that you must turn the crank exactly 3.7⁰ to raise the blade 0.001". Don.t worry about the tenth of a degree - just go to the nearest full degree. If you're in a hurry cranking the blade and miss it by all of 2⁰, you're still only off by a half thousandth!

   All of that is only done ONCE for your saw - it's all downhill from there.
   Be sure to write down the value you found - how many degrees per thousandth of height change. That's the one you'll use for every precision correction you need to make in the future.

1. Calipers told us that the stock was actually 0.812" thick. For a half-lap, then, the perfect dado depth is exactly 0.812/2 = 0.406"

2. Since it's 3/4" stock, we'll make a trial cut by setting the blade with a good ruler or depth gauge to 3/8" - that is, to half the nominal stock thickness.

3. Cut's made, and it turns out it measures (let's say) 0.386" deep. But we really want 0.406"...

4. How far off is it? That's 0.406 target minus 0.386 actual, or 0.020" - twenty thousandths off - ten times what would be visible!

5. Fix it fast. Raise the blade 0.020"

   Let's see - that's 3.7 degrees per thousandth times the 20 thousandths error = 3.7 x 20, or 74 degrees. The blade's gotta go up by 74 degrees of rotation. Turn the crank in about 3 seconds, and you're done and right on the money. No trial cuts, no excess sawdust, no compromise, and no fixing the joint after it's made.

   Ooops - turn the crank, eh? ... Remember, we said it made no difference what angle was indicated - it just was what it was. When you set for the trial cut, you paid no attention to the protractor wheel - you just raised the blade to trial height. So now the wheel says something weird, like 136⁰. No sweat - starting at 136⁰ and adding 74⁰ is 136 + 74, or a correction target indication of 210⁰. Go for it!
   (In other words, it's just a matter of figuring how far the crank needs to turn from wherever it happens to be right now for the desired change in height. Take where it is, add or subtract degrees as required, and turn to the new number.)

Let's choose the easy way. I'll share findings on my own saw, and discuss them.

The raising mechanism may or not be linear, depending upon the saw's age and manufacturer. I checked mine by simply finding zero blade height with a stick, and then raising exactly 10, 20, and 30 turns. At each stop, I used the depth gauge extension on a dial caliper to measure blade height. Figuring in only "close-to" perfectly vertical measurement at the saw's highest tooth, and my own clumsiness, I estimate that the results are within about 0.015".

At 10 turns, height = 0.982, and Change = 0.982, and Rate = 0.00027" per degree of turn
At 20 turns, height = 2.031, and Change = 1.049, and Rate = 0.00029" per degree of turn
At 30 turns, height = 3.053, and Change = 1.022, and Rate = 0.00028" per degree of turn

Now, I'm using this only for correction, not for direct setting, and I can measure rough setup within 0.060" if I'm falling down asleep...;-) So, let's say that there's a 0.060" correction needed.

For 0.060 correction at .00029 per degree, I need to turn 207⁰
At the worst, I'll be off by 0.00002" per degree according to 20 turn versus 10 turn above, so my error may be as much as 207 x 0.00002 = .0041"

That's too much - now what do I do?
Well, since the error is additive and gets larger and larger as I turn the wheel, I choose to limit corrections to no more than 100 degrees. If it's more than that, then I'll go most of the way and make a second setup cut as the new basis for correction. Note that if I'm more realistic and rough-set the blade only 1/32 off, the probable error falls to about 0.002", and I'm likely to be happy with the result.

I would never just take the whole on faith. Rather, I'd make the correction expecting good results, but would in any case make one more cut to double-check that all was in line. At that point, any final correction will be extremely small, and not accumulate significant error - I'd make it with considerable confidence ... and I've still only made a couple of cuts to arrive at a near-perfect joint.

I hope that this has been helpful for those who care to implement the method.
Glad to hear comments or questions via Email!