[Understory]

I recently picked up a used Powermatic 66 3HP in generally pretty good shape. Based on the S/N, I think it's roughly a 1987 vintage. Once I got it in the shop, I adjusted the table so the miter slots were parallel to the blade and then adjusted the rip fence to be parallel to the miter slots. It rips great and generally seems to be performing well.

However, I've just installed an aftermarket sliding crosscut table (Harvey Compass if curious) to the left wing and have been chasing my tail trying to get it to make a square crosscut on wider panels. I tried making all of the calibrations I could think of to the table itself and kept getting the same result. About a 32nd of runout over 12". I then turned my attention back to the saw itself, which I had assumed was true and square.

After double checking the blade for parallel to the table, it was indeed parallel to the miter slots, but only with the blade raised up all the way. If I lower the blade down to say 1" or so projection, I notice that the blade is no longer perfectly parallel to the table. It seems that when adjusting the blade height, the blade somehow swings slightly out of parallel. When measuring from the right miter slot to the blade, I'm getting a small gap (less than a 64th, but visible between my combo square ruler and the blade) on the back side of the blade(farthest from operator). The blade does stay vertically square as the blade is lowered though. I'm pretty sure this is the culprit for my non-square cross-cuts with the sliding table since I always had the blade lowered while cutting a 1/2" ply test piece. But I haven't been able to test it out yet by resetting the table with the blade lowered.

So my questions is what might cause the blade to go out of parallel with the table as the blade height is adjusted? Has anyone seen this before? Especially on older Powermatics?

 

[Steve Neul]

There must be some slack in the trunnions for it to move enough to notice. You might turn the saw over and see if you can physically move the arbor by hand.

 

[woodnthings]

On a cabinet saw the trunnions are mounted to the cabinet.
No need to turn that monster over!
You can still remove the table, no easy task either, and check the trunnions for the play that Steve mentioned.
Mount a magnetic base dial indicator so you can measure the blade movement or the equivilant.
And do snug all the mounting bolts!
Some saws change parallelism when the blade is tilted and that means shimming under the trunnion bolts.
That's not your issue here, though.

 

[B Coll]

Silly but basic question, what is on the saw when you are checking it? If it is a blade then it could be a blade issue. I see you are using a dial indicator which is the way to go. Ideally you should have a plate on the saw, dead flat, designed for calibrating the saw. I have always used a fresh Forrest blade as I know they are always dead flat.

 

[yomanbill]

However, I've just installed an aftermarket sliding crosscut table (Harvey Compass if curious) to the left wing and have been chasing my tail trying to get it to make a square crosscut on wider panels. I tried making all of the calibrations I could think of to the table itself and kept getting the same result. About a 32nd of runout over 12". I then turned my attention back to the saw itself, which I had assumed was true and square.

After double checking the blade for parallel to the table, it was indeed parallel to the miter slots, but only with the blade raised up all the way. If I lower the blade down to say 1" or so projection, I notice that the blade is no longer perfectly parallel to the table. It seems that when adjusting the blade height, the blade somehow swings slightly out of parallel. When measuring from the right miter slot to the blade, I'm getting a small gap (less than a 64th, but visible between my combo square ruler and the blade) on the back side of the blade(farthest from operator). The blade does stay vertically square as the blade is lowered though. I'm pretty sure this is the culprit for my non-square cross-cuts with the sliding table since I always had the blade lowered while cutting a 1/2" ply test piece. But I haven't been able to test it out yet by resetting the table with the blade lowered.

So my questions is what might cause the blade to go out of parallel with the table as the blade height is adjusted? Has anyone seen this before? Especially on older Powermatics?

I am far from an expert on the mechanics of the Powermatic table saw. However, my observation is that you are explaining two different and only slightly related issues. First. A small parallel alignment issue with the miter slot, blade, and rip fence will have a slight effect on ripping, but almost none on cross cutting. If you say that you are having trouble getting a square cross cut with your sliding crosscut table, that is an issue with the crosscut table, not the ripfence or blade alignment.
Of course, the blade should remain parallel when raised or lowered, but that is not related to your lack of square crosscut.

 

[woodnthings]

Silly but basic question, what is on the saw when you are checking it? If it is a blade then it could be a blade issue. I see you are using a dial indicator which is the way to go. Ideally you should have a plate on the saw, dead flat, designed for calibrating the saw. I have always used a fresh Forrest blade as I know they are always dead flat.

I didn't find where he said that?
All the measurements he has give are in fractions of an inch....

 

[woodnthings]

After double checking the blade for parallel to the table, it was indeed parallel to the miter slots, but only with the blade raised up all the way. If I lower the blade down to say 1" or so projection, I notice that the blade is no longer perfectly parallel to the table. It seems that when adjusting the blade height, the blade somehow swings slightly out of parallel.

The blade does not remain parallel to the miter slot.
This will affect the crosscut squareness in my opinion.
It may only "scrub" in the kerf and make it slightly wider, but who knows?
It's still not a good setup.

I am far from an expert on the mechanics of the Powermatic table saw. However, my observation is that you are explaining two different and only slightly related issues. First. A small parallel alignment issue with the miter slot, blade, and rip fence will have a slight effect on ripping, but almost none on cross cutting. If you say that you are having trouble getting a square cross cut with your sliding crosscut table, that is an issue with the crosscut table, not the rip fence or blade alignment.
Of course, the blade should remain parallel when raised or lowered, but that is not related to your lack of square crosscut.

I would test the sliding table for parallelism to the miter slot as it's moved front to rear also.
That may be difficult, but I envision a mag mount indicator on the table that will track along the miter slot

 

[Understory]

Thanks for all the ideas, everyone. It does seem like it must have something to do with the trunnion, so I'll have a closer look at all the mounting points and look for any wobble or wear. I have already had the top off to get in there to lubricate everything and improve duct collection. I didn't notice anything loose or looking amiss, but this was before I found the parallelism issue and knew to look.

Some clarifications / added detail:
I did set up a dial indicator on the blade and observed that it moved a few thousandths as I lowered the arbor. This was with a thin kerf blade installed and done somewhat hastily, so I'll try with several full-kerf blades to reproduce.

My method for checking parallel is a combo square in the miter slot and checking the same tooth at both the front and back and getting the slightest graze against the end of the ruler. I can't be certain, but I've always been under the impression that checking against the same tooth (I mark a tooth with a sharpie and rotate the blade back and forth) should help avoid errors due to a not perfectly flat blade.

The parallelism issue may well be unrelated to the crosscutting issue. I'm just focusing on getting the blade and saw table dialed in before moving back to that issue. I'm crossing my fingers that it does improve the crosscutting situation, but not holding my breath.

 

[_Ogre]

Your wood must be slipping on the cross cut sled.
Or the cross cut sled is set at an angle.

Even with the blade out of parallel with the miter slot, the cross cut sled will still push the board in a straight line. Think in extremes. If the blade was 20° out of parallel and you raised the blade ¼" in a ¾" board, the results would be a plowed cove cut in the underside of the board at 90°.

 

[Understory]

@_Ogre That sounds about right, but I'm wondering if the fact that I have to square the crosscut fence in the sliding table relative to the blade complicates your example and introduces some of the error I'm seeing. If the blade were off by 20 degrees and the crosscut fence we're perpendicular to the blade, wouldn't you end up with an angled cove cut in your example? I'm having trouble picturing that whole scenario without being Infront of the saw, but I'll experiment some more when I'm in the shop tomorrow. I will say that I'm fairly confident the sliding table is running square to the table itself since I did test that similar to how @woodnthings suggested above and saw less than a thousandths runout over 36".

 

[DrRobert]

Better to use a dial indicator. There are gadgets just for this, or a strip of wood in the slot to register base.

 

[Steve Neul]

Thanks for all the ideas, everyone. It does seem like it must have something to do with the trunnion, so I'll have a closer look at all the mounting points and look for any wobble or wear. I have already had the top off to get in there to lubricate everything and improve duct collection. I didn't notice anything loose or looking amiss, but this was before I found the parallelism issue and knew to look.

Some clarifications / added detail:
I did set up a dial indicator on the blade and observed that it moved a few thousandths as I lowered the arbor. This was with a thin kerf blade installed and done somewhat hastily, so I'll try with several full-kerf blades to reproduce.

My method for checking parallel is a combo square in the miter slot and checking the same tooth at both the front and back and getting the slightest graze against the end of the ruler. I can't be certain, but I've always been under the impression that checking against the same tooth (I mark a tooth with a sharpie and rotate the blade back and forth) should help avoid errors due to a not perfectly flat blade.

The parallelism issue may well be unrelated to the crosscutting issue. I'm just focusing on getting the blade and saw table dialed in before moving back to that issue. I'm crossing my fingers that it does improve the crosscutting situation, but not holding my breath.

In woodworking your smallest measurement should be 1/64", not thousandths. It's not necessary for woodworking machinery to be under that tight of tolerances. In actual use if you didn't put a dial indicator on it you would never know it was moving. With a tape measure you would never know the part you cut was a thousandth of an inch bigger or smaller than any other part. You can use hand pressure holding the parts you cut and the difference in pressure will make a bigger difference.

 

[woodnthings]

My method for checking parallel is a combo square in the miter slot and checking the same tooth at both the front and back and getting the slightest graze against the end of the ruler. I can't be certain, but I've always been under the impression that checking against the same tooth (I mark a tooth with a sharpie and rotate the blade back and forth) should help avoid errors due to a not perfectly flat blade.

A tri square is a very common device for checking parallelism among woodworkers, since not all that many will have a dial indicator.
The marked tooth is recommended also. A good full kerf blade will be better than a cheap thin kerf blade.
This is essentially a "once in a life time" adjustment so no need to purchase specific tools for it.
Some use a feeler gauge between the tooth and the tri square rather than going by feel. It won't matter that much.
I've used the tri square for setting up all of my saws for years, but by now I own about 5 dial indicators!
By now, I also own two South Bend metal lathes and Grizzly vertical milling machine, so dial indicators are necessary.

Grab the trunnions and give them a tug once the dial indicator is set up and see how much it will move the needle.
A wood pry bar against the cabinet wall will also work, but don't use too much force.
Some woodworkers recommend using star washers under the trunnion bolts to prevent movement.
If the castings are rough around the holes grind or file them flat.
You might remove one bolt at a time just to see how much larger the trunnion hole is than the cabinet hole is.
This will help determine how centered the trunnion assembly is on the cabinet edge.
I'd start with it as centered as possible and go from there.

 

[_Ogre]

@_Ogre I have to square the crosscut fence in the sliding table relative to the blade complicates your example and introduces some of the error I'm seeing.

i think you've just identified the problem. the only constant to a table saw is the miter groove. set everything off the miter groove. cross sled should be 90° to the miter slot, blade should be parallel to the miter slot, rip fence parallel to the miter slot.
when i set my miter gauge to 90°, i loosen the lock and use the front edge of the saw to set the miter gauge. it's guaranteed to be at 90°. when i check for parallel to the rip fence for a critical cut, i set the width off the blade, then check front and rear of the rip fence off the miter slot.
if the blade is slightly off of parallel, it'll still cut square. seriously off parallel it'll send off a rooster tail of sawdust on one side of the cut. i've only set the blade parallel to the miter slot one time. when i assembled the table saw back in 1982

 

[Understory]

A few updates after testing some things out:

I just tried the crosscut slider again while the blade was fully raised and parallel to the miter slots and I got the same off-square result. So I'm now fully convinced these are separate issues. I also realized that my test for checking the crosscut slider for parallel to the table was flawed. I had a dial indicator mounted on the table and a straight edge lodged in the track on the sliding table. Moving the table back and forth actually only told me that my straight edge was straight, and perhaps that the sliding mechanism on the table also runs straight. Checking it from only one point relative to the saw table wouldn't tell me whether the sliding table is mounted at an angle. 🤦‍♂️

So I still need to resolve the squareness issue with the sliding table, but I still want to get to the bottom of the issue with the blade parallelism. @Steve Neul, I'm not sure I agree about tolerance as it pertains to tool setup. In my experience, a blade out of parallel by even less than a 64th over the length of the blade can cause a burning or binding while ripping. I'm not too concerned about a 64th variance in a piece I'm cutting, but a 64th variance in certain tool setups can magnify the error over longer cuts. In any case, my combo square method relies on the faint sound of the teeth brushing the ruler end which I'm confident is far greater than a 64th of precision since it's beyond what I can see with my eye. And the difference in parallel that I'm seeing after lowering the blade is very visible to the eye. Probably in the realm of a 64th or more.

Regardless of the measuring technique or blade (I've tried 3 different blades including a full-kerf) I'm still getting a repeatable result of the blade going out of parallel as the blade is lowered. I've jostled all of the trunnion components around with a dial indicator on the blade and nothing appears to be loose. The most movement as measured towards the outside of the blade was only several thousandths.

I did get maybe a quarter turn tightening the large hex nut at the end of the shaft where the saw arm pivots to raise and lower the blade. This has thrown the blade out of parallel with the blade fully raised, but in the opposite direction of where it goes off when lowered. In the process of resetting it to parallel to see if it did improve anything. 🤞

 

[woodnthings]

A few updates after testing some things out:

I just tried the crosscut slider again while the blade was fully raised and parallel to the miter slots and I got the same off-square result. So I'm now fully convinced these are separate issues. I also realized that my test for checking the crosscut slider for parallel to the table was flawed. I had a dial indicator mounted on the table and a straight edge lodged in the track on the sliding table. Moving the table back and forth actually only told me that my straight edge was straight, and perhaps that the sliding mechanism on the table also runs straight. Checking it from only one point relative to the saw table wouldn't tell me whether the sliding table is mounted at an angle. 🤦‍♂️

So I still need to resolve the squareness issue with the sliding table, but I still want to get to the bottom of the issue with the blade parallelism. @Steve Neul, I'm not sure I agree about tolerance as it pertains to tool setup. In my experience, a blade out of parallel by even less than a 64th over the length of the blade can cause a burning or binding while ripping. I'm not too concerned about a 64th variance in a piece I'm cutting, but a 64th variance in certain tool setups can magnify the error over longer cuts. In any case, my combo square method relies on the faint sound of the teeth brushing the ruler end which I'm confident is far greater than a 64th of precision since it's beyond what I can see with my eye. And the difference in parallel that I'm seeing after lowering the blade is very visible to the eye. Probably in the realm of a 64th or more.

Regardless of the measuring technique or blade (I've tried 3 different blades including a full-kerf) I'm still getting a repeatable result of the blade going out of parallel as the blade is lowered. I've jostled all of the trunnion components around with a dial indicator on the blade and nothing appears to be loose. The most movement as measured towards the outside of the blade was only several thousandths.

I did get maybe a quarter turn tightening the large hex nut at the end of the shaft where the saw arm pivots to raise and lower the blade. This has thrown the blade out of parallel with the blade fully raised, but in the opposite direction of where it goes off when lowered. In the process of resetting it to parallel to see if it did improve anything. 🤞

What are the instructions for mounting the sliding table and do they mention shimming for parallelism?
I certainly would not want my sliding table "dog tracking" with respect to the miter slots.
Which is what I understand to be one of your issues here.

If it were my saw, I'd like to solve the blade height issue first since blade height adjustments are a primary function of the saw.
I'm not familiar with how those trunnions attach to the carriage, maybe like the Craftsman where it's a half round male to female intersection?
As I understand at this point, the height of the blade skews one end of the carriage which contains the arbor, out of parallel.
There was a common issue on the early hybrids sold by Craftsman and others where the blade did not remain parallel at a bevel setting ....
if memory serves?

 

[builtinbkyn2]

OP the saw is a recent purchase that you obviously had to move. It is possible that your top is out of plane with the trunnion - meaning it's tilted. This could have happened in the move or it could always have existed this way. If you, your mover, the original owner moved the saw by grabbing/levering the top, this could have caused it to go out of plane. The only way I know of to check for this is to raise the blade to full height, tilt it to 45 and get a dial reading at both the infeed and outfeed ends of the blade from the underside of the blade to the table top. You would never visibly notice if the top is out of plane, but it can me measured and then adjusted. A few thousandths off will compound during bevel cuts.

 

[_Ogre]

you guys and your dial indicators. other than jointer blades, i've never used a dial indicator in my shop. using a dial indicator will always show movement on a 1/8" sawblade, cuz they aren't flat. i did millwright work for dana corp for years. wood moves more than metal.

 

[woodnthings]

OP the saw is a recent purchase that you obviously had to move. It is possible that your top is out of plane with the trunnion - meaning it's tilted. This could have happened in the move or it could always have existed this way. If you, your mover, the original owner moved the saw by grabbing/levering the top, this could have caused it to go out of plane. The only way I know of to check for this is to raise the blade to full height, tilt it to 45 and get a dial reading at both the infeed and outfeed ends of the blade from the underside of the blade to the table top. You would never visibly notice if the top is out of plane, but it can me measured and then adjusted. A few thousandths off will compound during bevel cuts.

I like that thinking!
If it were me, I think I'd shim the trunnions where they mount to the cabinet.
If the cabinet is the reference??? then any measurements can be referenced off the that.
I'd like to see what the manual says for leveling out the top.
Seems like there's no way to reference that, maybe GPS?
If there is ZERO play in the trunnion to carriage attachment, then your thinking could be spot on.
I wonder what would happen if the table were removed and measurements were taken to the cabinet edge or some other fixed reference point, as a test?

 

[woodnthings]

you guys and your dial indicators. other than jointer blades, i've never used a dial indicator in my shop. using a dial indicator will always show movement on a 1/8" sawblade, cuz they aren't flat. i did millwright work for dana corp for years. wood moves more than metal.

YUP, that's for certain.
At this point however, it's become somewhat of a challenging "mystery" why it's going out of alignment.
It would be great if we can figure it out!