Trunnion Alignment


Just because you align the blade at 90 degrees doesn’t mean it will stay that way when you tilt it to 45 degrees.  On most table saws, the entire blade carriage rides in semi-circular tracks called “trunnions”.  The blade is tilted as the assembly travels along these tracks.  The “axis of rotation” for the blade tilt mechanism is supposed to run along the surface of the table where the blade comes through the table insert (throat plate).  This way, the table insert won’t interfere with the blade even when it’s tilted.  However, inaccuracy in machining, or stress relief in the castings often causes this axis of rotation to shift so that it no longer runs along the surface of the table.  So, as the blade is tilted, it gradually gets out of alignment with the miter slot.  Fortunately, this process should only be required once for the life of your saw.


The following symptoms are common when making bevel cuts on a saw when the trunnions are not properly aligned:

  1. Burning of the cut edge
  2. Board wandering away from the fence during a rip cut
  3. Sawdust being thrown up at the operator
  4. Kickback

These are the exact same symptoms that occur for normal (90 degree) blade alignment.  If you see any of these symptoms when you tilt the blade (making bevel cuts), then you need this procedure.


First, make sure that the standard blade alignment (blade at 90 degrees) is accurate.  Then, tilt the blade to 45 degrees and check the alignment. 

The measurements

The best way to do this is to tilt the dial indicator as shown in the photo above.  The most accurate readings always result when the plunger of the dial indicator is at right angles to the surface being measured.  Use the same method that was described for standard blade alignment - mark a dot on the blade and take all measurements with the stylus of the dial indicator on that spot - rotating the blade as necessary.  The error measured in this step will represent a combination of horizontal (90 degree) and vertical (tilt axis) misalignment.  If you don’t perform the standard 90 degree blade alignment accurately, then it will influence your readings when the blade is tilted and invalidate the procedure.

Measurements can be made with the dial indicator horizontal as shown above but I don’t recommend it.  Many sources of error can creep into your readings making them inconsistent and very frustrating.    The dial indicator will have a tendency to lift during the measurement process so be sure to check that the jig remains flat on the table.  Since the dial indicator is not perpendicular to the surface being measured, the readings will be exaggerated by 1/cos(45).  This turns out to be equivalent to the vertical (tilt axis) misalignment so no further correction is needed.

Make a note of the change in reading between your two measurement points as well as the direction of the change.  In this case it’s about 0.006 inches and it’s higher at the trailing edge of the blade.  To determine the vertical (tilt axis) component of the misalignment, multiply the measured error by 1.414 (the square root of 2).  The result, 0.0085″, is the amount of tilt axis error influencing the misalignment of your blade.  When measured with the dial indicator horizontal, the reading is a little more than 0.008 inches.  Correcting the error will involve inserting some shims and you will need this information to calculate the proper thickness of the shims and where to install them.

The second number you will need is the distance between the two measurement points.  In this case, it’s about 8 inches.  When measuring with the dial indicator horizontal, the distance between the measurement points is also about 8 inches.

The third number you will need is the distance from front to back between table bolts (cabinet saw) or trunnion bolts (contractor’s saw).  For my saw it’s about 20 inches.

The Calculation

To figure out how thick the shims need to be, first calculate the amount of change in reading per inch of measurement.  The vertical (tilt axis error) we calculated above is 0.0085″.   This was calculated from a measurement over an 8 inch span.  That’s 0.0085/8 or about 0.001 inches change per inch of measurement.

The thickness of the shims will be the distance between front and rear table bolts (cabinet saw) or trunnion bolts (contractor’s saw) multiplied by the “change per inch” calculation.  So, 20 x 0.001 = 0.020 inches. 

Installing the Shims

Make your shims out of metal.  If you make them out of wood, plastic, or cardboard, then they will compress over time and you will lose your alignment.  If a large amount of change is required, you will want to check the thickness of washers.  Soft drink cans make good material to cut thinner shims from.  Aluminum foil can be used for making very fine adjustments to the overall thickness. 

Shape for shimms

Cut your shims in the shape of a “U” so that they can fit around the mounting bolt.  Dial calipers will help you to measure the thickness of the shims so that you can start out with something close to what you need.  But, as you install them and tighten the bolts some compression may occur. 

Location of table bolts

If you have a cabinet saw, then the shims have the effect of raising the table.  They go between the table and the top of the cabinet underneath the bolts that hold the two together.  Place them under the two bolts at the front of the saw if the reading at the leading edge of the blade was higher than at the trailing edge.  If the reading was higher at the trailing edge of the blade, then the shims should be placed underneath the two bolts at the back of the saw. 

Monitoring shim thickness

I’ve found it handy to put a dial indicator on the table surface above one of the mounting bolts.  This will allow you to monitor the amount that the table is being raised.

Shimming trunnions

If you have a contractor’s saw, then the shims have the effect of lowering the blade carriage.  They go between the trunnions and the table.  If the reading at the leading edge of the blade is higher than at the trailing edge, then you need to shim the front trunnion.  Working on the rear trunnion is much easier than the front trunnion.  So, to save your knuckles from extreme torture, turn the saw upside down and deliberately over-shim the front trunion.  Then, turn it right side up and figure out how much shimming needs to be done to the rear trunnion to correct the misalignment. 

If the reading was higher at the trailing edge of the saw, then place the shim under the trunnion at the back of the saw.  It’s best to locate the shims directly underneath the bolts.

Checking the results

All of this shimming is likely to mess up the standard (blade at 90 degrees) alignment.  So, start by checking it and make any adjustments that are necessary.  Then, tilt the blade to 45 degrees and check for any alignment error.  If you did a good job making measurements and installing shims, the error should be significantly reduced or even eliminated.

Ed Bennett


16 Responses to “Trunnion Alignment”

  1. Art Says:

    With the indicator aligned at 45 degress to the table top [trunnion-alignment2a.jpg] why isn’t it necessary to correct [times root(2)] for the correct shim thichness?
    Alternatively, why not adjust the indicator perpendicular to the tabletop and get the shim thichness directly?

  2. admin Says:

    Hi Art,

    The geometry is a mind twister to be sure!

    Unfortunately, you can’t put a dial indicator on the tilt axis itself and measure it directly because it’s an imaginary line. So, you measure the effect it has on the blade alignment (relative to the miter slot). Eliminate the misalignment and you solve the problem.

    When you measure the error with the dial indicator perpendicular to the blade, you get the actual alignment error imposed by the tilt axis not being parallel to the table surface. The alignment error will always be perpendicular to the blade surface (parallel to the arbor). If you could tilt the blade so that it was parallel to the table surface (90 degrees), then you would still need measure the alignment error with the dial indicator perpendicular to the blade.

    Measuring with the dial indicator vertical (perpendicular to the table) is exactly the same as measuring with the dial indicator horizontal (parallel to the table). In both cases it will be at 45 degrees to the blade and you will suffer the cosine error on the reading. Correct for the cosine error (factor of 0.707) and you get the same reading that you have when you measure the error perpendicular to the blade. That’s because you’re not measuring the tilt axis, you’re measuring the blade alignment error.

    Is the alignment error at 45 degrees the same as the tilt axis error? Empirically I would say yes. In my experience the calculated shim size pretty much eliminates the error. And, using the horizontal measurement (1.414 times bigger with the cosine error) tends to result in over correction. I have yet to sit down and work out a geometric proof (which would be far better than empirical evidence).

    In any case, you don’t get the shim thickness directly because your measurement distance is a lot smaller than the distance between mounting bolts. You still need to figure out the slope (change per inch) and extend it over the distance between the mounting bolts.

    Ed Bennett

  3. Art Says:

    You haven’t convinced me.

    In the case of adjusting the trunnion, what is in error is the rear of the tilted blade is closer (or further) from the tabletop than the front is. In geometric terms the intersecting planes of the blade, the tabletop, and the vertical plane thru the miter slot and perpendicular to tabletop (ie the fence face) form a funnel instead of an infinite triangular tube.

    As the adjustment mechanism (shims) raises or lowers the tabletop (or trunnion) perpendicular to the tabletop then the dial indicator axis must also be perpendicular to the tabletop. If not, then the cosine error must be corrected for. By measuring as you state one would have to simultaneously adjust both the tabletop (or trunnion) elevation and rotation which would be considerably more difficult.


  4. admin Says:

    Hi Art,

    I understand that you want to measure how much the trunnion (or table) needs to be raised (or lowered). And, it seems logical that measuring vertically will deliver this value. But, I don’t think you are getting the result you want.

    The triangular funnel that you describe is an isosceles right triangle. If both of the sides have length 1 unit, then the hypotenuse (blade) has length 1.414 units (Pythagorean theorem: a^2 + b^2 = c^2). You can’t have unequal length sides unless you change the angles, and we know it stays constant at 45 degrees. So, any change in length of one side (the vertical measurement: blade to table surface) requires an identical change for the other side (the horizontal measurement: blade to miter slot) in order to maintain an isosceles right triangle. In other words, measuring vertically is identical to measuring horizontally.

    The only thing we can measure is the surface of the blade, and we’re going to measure it in relation to the table surface. So, if both sides of the triangle are being changed by the same amount at the same time, then what direction is the blade traveling? Simple vector analysis: it’s moving parallel to the arbor, or perpendicular to the surface of the blade.

    The motion perpendicular to the blade is basically the height of your triangle. For every unit change vertically, (and horizontally), you’ll get 0.7071 (1/2 the square root of 2) units change perpendicular to the blade. So, it would seem logical that multiplying the reading on the dial indicator by the square root of 2 will give you the vertical component of motion - which is what you originally asked about. But, I’m not convinced that this describes the error in the blade tilt because in my experience this overcorrects the error. I can’t help but feel that I’m missing something.

    The cosine error comes into play whenever the plunger on the dial indicator is tilted in relation to the surface being measured. The reading itself is distorted (exaggerated). It no longer represents the surface that you intend to measure. If you could measure vertically and have the indicator stylus contact a horizontal surface, then the readings would represent the surface and you wouldn’t need any correction for cosine error. It’s difficult to obtain reliable results when measuring an angled surface so it’s generally discouraged.

    Ed Bennett

  5. Art Says:

    A tablesaw in perfect alignment will have the blade tilt axis parallel to both the fence face/miter slot and the tabletop (It actually lies within the plane of the tabletop).

    To adjust the axis for parallel to the fence the blade is set parallel to the fence ( 0°) and the error is measured front to back using the miter slot with the indicator perpendicular to the blade (horizontal). This measurement contains only the horizontal component which is precisely what is desired.

    Likewise to adjust the axis for parallel, or coplanarity, to the tabletop the ideal would be to put the blade horizontal (90°) and again measure front to back with the indicator perpendicular to the blade (vertical). As this is impossible the best we can do is have the blade at 45° to the tabletop.

    Assume a right tilt Unisaw with the rear of the blade tilt axis lower than the front and that the axis is perfectly aligned with the fence/miter slot. Tilting the blade at 45° will put the rear of the blade both lower than the front and closer to the fence than the front.

    Placing the indicator perpendicular to the blade puts it at 45° to the tabletop. This measurement will now have both vertical and horizontal components to it and it is necessary to extract the vertical component by multiplying by 0.707. Taken to the extreme a blade tilt of 0° (vertical) would give a measurement having no vertical component which would be totally useless in making this adjustment.

    On the other hand placing the indicator at 90° to the tabletop and making front and rear blade measurements yields the vertical component directly. However, if there is any side play in the indicator this could give a false reading as you state. So it is probably better to measure perpendicular to the blade and make the 0.707 correction.


  6. admin Says:

    Hi Art,

    I understand exactly what you are saying. Ideally, alignment with the miter slot should done with the blade vertical and the dial indicator horizontal. Alignment with the table surface should be done with the blade horizontal and the dial indicator vertical. Vertical blade we can do. Horizontal blade is not possible. In both cases, a major source of systematic measurement error (cosine error) is avoided by orientating the dial indicator so that it is perpendicular to the surface being measured (the blade).

    What it really comes down to is one question: can you accurately extract the vertical and horizontal components of blade misalignment from a tilted blade and calculate the necessary corrections in both directions? I don’t think it’s that simple. I don’t believe that you isolate the horizontal component when you orient the dial indicator horizontally. And I don’t believe that you isolate the vertical component when you orient the dial indicator vertically. I wouldn’t want to correct horizontal misalignment based on measuring a tilted blade. So, I’m equally skeptical about correcting vertical misalignment with data from a tilted blade. In theory (isosceles right triangle) the error in both components are going to be equal at a 45 degree blade tilt. It’s only when the blade orientation departs from 45 degrees that you begin to see differences. It’s the orientation between the objects being aligned that matters, not the orientation between one object and the measurement device.

    I don’t believe that you can measure the surface of one object by placing the stylus of the dial indicator on a second object that is tilted in relation to the first object. It doesn’t matter that the dial indicator is perpendicular to the first object unless the stylus tip is in contact with the first object. In other words, I don’t believe that you can measure changes in the table surface by placing the stylus of the dial indicator on the tilted blade. The cosine error occurs between the travel of the plunger and the surface being measured. You are not free of cosine errors just because you have orientated your dial indicator perpendicular to some surface that you are not measuring (the table).

    Instead of trying to extract the blade tilt axis error directly, my approach has been to treat the symptom: measure the actual blade misalignment and correct for it. I’ve been careful to avoid introducing systematic errors (like the cosine error introduced by tilting the dial indicator). In practice, my approach has been very successful. So, I don’t mind sharing it publicly. And, I don’t mind having it scrutinized and challenged. My goal is to share the best practices, not just the ones I think are right or those that should be right according to some theory.

    Here’s what I’m going to do: I’m going to perform some tests on my table saw comparing shim thickness calculated using the blade alignment error and shim thickness calculated using the vertical component of the blade alignment error (using the correction factor, not tilting the dial indicator). It won’t be anecdotal evidence because I’ll collect actual data and report the results here. I would encourage you to do the same!

    Ed Bennett

  7. Art Says:

    That sounds good. I don’t have anything quite as nice as your aligner gizmo but I think I can cobble something together.

  8. Art Says:


    I have results from testing on my RT Unisaw.

    To simplify testing I reasoned that the measurement could be done by adjusting the height of the indicator using the shims directly under it instead of under the tabletop. This would also remove the need to compensate for differing measurement baseline and tabletop mounting distances. If you think there is a problem with this let me know.

    For the first test I tilted the blade to 45°, set the dial indicators axis at 90° to the blade and the indicator tip touching the blade directly over the arbor about half way up the blade. I zeroed the indicator at this point. Then I placed metal shims under the indicator, being careful to keep it’s base parallel to the tabletop, and made the reading.

    For the second test everything was the same except for the indicators axis being vertical, 45° to the blade.

    All numbers are in inches except as noted.
    My dial indicator is metric.
    My alignment jig is all metal. I can take a photo & send it if you’d like.

    Indicator = 45° [90° to blade]
    Thickness Indicator (mm) Indicator Indicator * 1.414 Error
    0.006 0.120 0.0047 0.0067 0.0007
    0.023 0.420 0.0165 0.0234 0.0004
    0.104 1.860 0.0732 0.1035 -0.0005

    Indicator = vertical [45° to blade]
    Thickness Indicator (mm) Indicator Error
    0.006 0.120 0.0047 -0.0013
    0.023 0.550 0.0217 -0.0013
    0.104 2.610 0.1028 -0.0012

    This data supports both of our statements.
    Your statement re the indicator needing to be at 90° to the blade is validated by the data in method 2. While the absolute error is only about 0.001″, it is 22% of the reading in the most applicable range.
    Method 1 validates my initial statement that the indicator reading must be corrected by multiplying by 1.414. My June 22nd statement re multiplying by 0.707 was in error.


  9. admin Says:

    Hi Art,

    I’m not sure I follow exactly what you did. It seems as if you have confirmed the Pythagorean Theorem. And, you’ve demonstrated why it’s important to keep the dial indicator perpendicular to the surface being measured.

    I’ve been very busy this week so I haven’t had a chance to do the testing I promised. I will post my results when I get the testing done.

    Ed Bennett

  10. mike Says:

    I’ve heard different opinions on the importance of the adjustment of parallel rods on the carriage of a contractors saw. the most realistic one says to adjust them by placing a piece of glass on the rods and adjust them so the glass doesn’t wobble. what are your thoughts.

  11. admin Says:

    Hi Mike,

    I’ve heard this too. Since I haven’t had a chance to try it myself I decided to leave it out of the initial publication of this article. Certainly, the rods do need to be parallel. Plate glass isn’t guaranteed to be flat but it’s not unreasonable to assume that its fairly flat. Placing it on the rods and adjusting so that there isn’t any rocking would line them up in one plane but not guarantee that they are parallel.

    I’m looking forward to rebuilding the Delta contractor’s saw and testing techniques like this. I will update this article with photos and information as they become available. Stay tuned!

    Ed Bennett

  12. admin Says:

    I’ve finished my tests and have reviewed some additional information that Art sent to me. I believe that Art is correct when he suggested that the dial indicator reading needs to be corrected to reflect the vertical component of the tilt axis alignment error. The difference was extremely small and it could be easily overlooked if you are not very careful. But, in the end, correcting the reading provides better alignment.

    Measuring horizontally or vertically is not the answer. It’s just too difficult to obtain consistent and repeatable readings. It’s best to tilt the indicator to match the tilt of the blade and then apply the correction (multiply by 1.414).

    I’ve re-written the article to reflect this and have added some additional information which might help people to obtain good results. Let me know what you think!

    Ed Bennett

  13. John Says:

    I have the misfortune of buying a Sears saw which, has aluminum trunnions. Set to extreme I am out of align by .009. Nearly 1/10th of an inch!

    I have ground out the holes of the trunnions to allow a further adjustment but it is all for naught. And, the trunnion holes now have very thin walls. It could be my imagination but I get the feeling the lock downs have turned to mush.

    Is there any place a person can buy replacement trunnions?

    I have my hopes set for a Delta but until that happens I have to get this POS working. Assist appreciated.

  14. admin Says:

    Hi John,

    I think you mean to say that your alignment is out by nearly 1/100 of an inch. 1/10th inch is 0.100″. Still not so good.

    It is possible that your saw was machined so badly that it cannot be aligned. I would have to get a look at it to know for sure what is wrong. Sometimes people don’t realize that it’s important to loosen both trunnions (front and rear) to align the saw. It’s also possible that there is something other than the size of the mounting holes that is preventing you from making enough adjustment to get proper alignment. Sometimes the best thing to do is turn the saw upside down and have a good look at the assembly. Loosen up the trunnions and check to see just how much adjustment you have and what is restricting it.

    Sears does stock spare parts for most items. Give their parts department a call to find out if you can purchase replacement trunnions.

    Ed Bennett

  15. John Says:

    Thanks Ed. I disconnected the “tilt” screw from the body and everything popped into place.

    For those following Craftsman, changing the tilt angle did not effect the problems in alignment at all. However when disconnected all went into place.

    When the tilt was mounted again everything worked as it should and the alignment stayed on.

    Also I loosened off the 3 upper cabinet bolts to allow flex when I did up the tilt again. Then, tightened these up as part of the operation.

    Thanks again Ed.

  16. admin Says:

    Hi John,

    You’re welcome! Glad to hear that you got it figured out! This is a first for me. I’ve never heard of the tilt mechanism preventing proper blade alignment. Good job!


Leave a Reply