MFT3 table not cutting square--not even close

[paulhtremblay]

Here are the picts from my latest test. (The last picture should actually be the third one down, to reflect the order of my process):

Unfortunately, all 3 tests still resulted in an out of square cut. The results from the picts, 1.2 mm difference over 1200 mm is about what I got for the first 3 tests.

What really makes the matter worse, I did two additional tests using the rail and Qwas dogs. I expected these tests to come out square, but they resulted in out of square cuts as well. The first one was a full 2mm off over about 1800 mm, the second, about .5 mm over 1500 mm.

 

[bkharman]

In the last pic... Did you do another 5th cut with the dogs?  Or did you do all 5 with the dogs as well. If so, similar results? 

I have read and re-read this post a hundred times racking my brain as to what it could be. I believe you said you got Qwas dogs as well as the rail dogs, right?

If so, use the fence or the dogs to act as a fence and for the rail, use another or take the hinge off and try it with the rail dogs. It might be a bit tricky to get it all to line up (and you will definitely end up with different kerf marks) but it will completely be using the holes for alignment. See how that cut test works and work from there.

 

[paulhtremblay]

In the last pic... Did you do another 5th cut with the dogs?  Or did you do all 5 with the dogs as well. If so, similar results? 

I have read and re-read this post a hundred times racking my brain as to what it could be. I believe you said you got Qwas dogs as well as the rail dogs, right?

If so, use the fence or the dogs to act as a fence and for the rail, use another or take the hinge off and try it with the rail dogs. It might be a bit tricky to get it all to line up (and you will definitely end up with different kerf marks) but it will completely be using the holes for alignment. See how that cut test works and work from there.

I got rid of the last pict because it was confusing. I also updated the post.  The last pict shows how I did the last two tests, completely independent of the first 3 tests. So I have already done what you suggested, with the same results. Yes, I am puzzled.

 

[bkharman]

Try this with the fence and the dogs:

Do a cut, the slide the board out 30mm or so and do another parallel cut. Caliper both ends... We should expect the same measure. (Fingers crossed)

 

[tjbnwi]

What and how are you adjusting to get the unit squared. Is it always off the direction?

Tom

 

[paulhtremblay]

What and how are you adjusting to get the unit squared. Is it always off the direction?

Tom

As I have stated, both squares (themselves tested), and, as the pictures show, dogs.

 

[grbmds]

In your pictures I noticed you were using a digital caliper. Two things:

1. Do you have fresh batteries installec?
2. When you do your measurements, did you zero out the digital caliper each time you make a measurement (both before the first measurement and before the second measurement)?

 

[tjbnwi]

What and how are you adjusting to get the unit squared. Is it always off the direction?

Tom

As I have stated, both squares (themselves tested), and, as the pictures show, dogs.

Sorry I did not ask my question properly.

Now that you know it is out of square what are you moving on the MFT 3 to set the unit to square?

Tom

 

[paulhtremblay]

What and how are you adjusting to get the unit squared. Is it always off the direction?

Tom

As I have stated, both squares (themselves tested), and, as the pictures show, dogs.

Sorry I did not ask my question properly.

Now that you know it is out of square what are you moving on the MFT 3 to set the unit to square?

Tom

I guess you are asking if I am then moving my fence or guide to correct an error after a 5 cut test? If so, I am doing nothing at this point. I guess I could move a feather key the needed distance, but I don't know if that will really be a long term solution for MFT3. If I have to move my fence again, I would have to re-do the 5 cut test each time, no? If you bolted the fence to the MFT3, then I suppose you could tune the feather keys just like a table saw sled.

 

[paulhtremblay]

Try this with the fence and the dogs:

Do a cut, the slide the board out 30mm or so and do another parallel cut. Caliper both ends... We should expect the same measure. (Fingers crossed)

Good suggestion. I will do this tomorrow. That I can't cut square even when using dogs makes me think the problem lies either with my technique (but how?) or the saw itself. I already had to send the saw back once to get the bevel set correctly.

 

[bkharman]

Try this with the fence and the dogs:

Do a cut, the slide the board out 30mm or so and do another parallel cut. Caliper both ends... We should expect the same measure. (Fingers crossed)

Good suggestion. I will do this tomorrow. That I can't cut square even when using dogs makes me think the problem lies either with my technique (but how?) or the saw itself. I already had to send the saw back once to get the bevel set correctly.

I own an analytics company, it is in my nature to take one step and over analyze it...

I don't know that I would bump the feather key just yet. Let's rule out what it isn't before we capture what it is.

Bryan

 

[tjbnwi]

You need to do something to correct the error. I moved my key. I use my MFT in the field and remove and reinstall the rail often. The only time I had a problem the rail must have got bumped hard. I had to loosen the bolts that hold the hinge to the rail and retighten them.

Clean the blade. MDF is rough on them.

Moving the feather key is no big deal. It's the primary adjustment feature on the MFT.

Tom

 

[paulhtremblay]

In your pictures I noticed you were using a digital caliper. Two things:

1. Do you have fresh batteries installec?
2. When you do your measurements, did you zero out the digital caliper each time you make a measurement (both before the first measurement and before the second measurement)?

Yes, the batteries are fresh. Yes, I zero out the caliper. I make multiple measurements, so the probability that the caliper would show the measurements off in the same exact way is < .01. (My job requires me to know stats!)

Plus, I use a square on the board after the five cuts. One of the corners reflects the accumulation of the 5 cuts, and is terribly out of square.

 

[paulhtremblay]

You need to do something to correct the error. I moved my key. I use my MFT in the field and remove and reinstall the rail often. The only time I had a problem the rail must have got bumped hard. I had to loosen the bolts that hold the hinge to the rail and retighten them.

Clean the blade. MDF is rough on them.

Moving the feather key is no big deal. It's the primary adjustment feature on the MFT.

Tom

Right, I agree moving the feather key is no big deal. I have moved my a number of times for these tests. But the issue is why, when guide and fence are aligned square to dogs, am I getting such large errors? Likewise, why am I not getting square results when using just the dogs? I don't won't to cover up a problem. The fact that the saw burns plywood really has me wondering.

I cleaned the saw blade just last night. The blade was brand new as of about 5 days ago, but I have done so many 5 cut test, I can't call it that any more.

Thanks for your suggestions.

 

[tjbnwi]

I must have missed your saw is burning plywood. Check your cut piece a few inches closer to the free end side. I wonder it the toe is off on the saw barking off the end of the test piece.

Tom

 

[Qwas]

I looked at your pictures and I think you are doing the cuts wrong. When it is done correctly, the just cut side goes to the fence for the next cut. So your numbered sides should be going clockwise, not counterclockwise.

 

[paulhtremblay]

I looked at your pictures and I think you are doing the cuts wrong. When it is done correctly, the just cut side goes to the fence for the next cut. So your numbered sides should be going clockwise, not counterclockwise.

Yea, you are exactly right on that. Never even occurred to me. I wonder what happens when you do the test wrong like I did?

Tomorrow I will do the test right. During the day, at work, I am going to ponder the geometry of doing the test backwards.

Thanks for having such good eyes! (Glad I numbered the board, too.)

 

[Qwas]

Yes, I always number the board too so I don't get confused halfway through. The board should get turned 90 degrees counterclockwise with every cut. If you go the other way you are only testing how square side 4 was to side 1, none of the other cuts mattered.

The other concern I have reading through this thread is whether the fence is actually straight. It sounds like it might have a very small bend in it.

The other to remember is the final caliper reading comparing cut 1 and 5 is a 4 times multiple of the actual error. So don't get too carried away with these numbers.

 

[grbmds]

Yes, I always number the board too so I don't get confused halfway through. The board should get turned 90 degrees counterclockwise with every cut. If you go the other way you are only testing how square side 4 was to side 1, none of the other cuts mattered.

The other concern I have reading through this thread is whether the fence is actually straight. It sounds like it might have a very small bend in it.

The other to remember is the final caliper reading comparing cut 1 and 5 is a 4 times multiple of the actual error. So don't get too carried away with these numbers.

I have not felt the need to use the 5-cut method to determine squareness on my MFT. I merely use a couple of squares I have confidence in and the end result has been excellent.

If what Qwas says is correct, that the ending difference is 4X's the actual error off of square, your measurements would indicate about a .29 mm difference over whatever length the piece was. I opened my digital caliper as close to .25 mm as possible and I could  barely see the opening between the jaws. It seems as if Festool users expect complete perfection in the cut and squareness. I fell into that way of thinking at first. Now, I know that my cuts and joints are better than before Festool; not perfect because I'm not sure that is even possible. Perfection is made up of a lot of pieces, including how good your eyesight is. To me, the true test is what the end result looks like. If the inaccuracy is so small that the naked eye cannot see it, then it doesn't matter, does it? Right now, that is how small my inaccuracies are using the TS55, router, and Domino. To go through the process of cutting a piece on 5 sides to show that the table and saw are a tiny bit out of whack wouldn't add anything to my woodworking experience. Why don't you just try to make something that you need. If the result is what you want once the project is together and done, then why worry. Just use the tools and enjoy them.

 

[shakenmartini]

I looked at your pictures and I think you are doing the cuts wrong. When it is done correctly, the just cut side goes to the fence for the next cut. So your numbered sides should be going clockwise, not counterclockwise.

Yea, you are exactly right on that. Never even occurred to me. I wonder what happens when you do the test wrong like I did?

Tomorrow I will do the test right. During the day, at work, I am going to ponder the geometry of doing the test backwards.

Thanks for having such good eyes! (Glad I numbered the board, too.)

Just to be clear, you are not putting the freshly cut side against the fence? This is what your pictures suggest, but I am not 100% clear on this.

If you are not putting the freshly cut edge against the fence after rotating 90º, you will have error creep into your calculation based on how square the best piece is. If you are using a factory MDF edge, it very well could be off by quite a bit and would mess up your error measurement. This probably explains why the error has not been systematic and instead random since when you cut the reference edge off, you are adding an unknown error to the mix, that is then multiplied by 4X (I think that is right).