MFT3 table not cutting square--not even close

[Michael Kellough]

First off, like Steve said, you should make sure the fence is straight. On of your comments that "the shorter the piece the more off the result" is symptomatic of a bent fence. If not a bend there could be a bur that pushes the work off the rail.

Also as Steve said, make sure you understand the 4 cut procedure. (irrelevant is the fence is bad)

The newer style MFT is simply not as accurate as the old style for several reasons.

1. the fence does not fit under the guide rail so the work under the guide rail is not supported
    If either the fence or that side of the work are less than perfectly straight the problem is compounded.

2. the guide rail is wider to even less of the work is supported than would be with the old rail

3. the fence pivot is much farther from the end of the fence than the old style.
    In many installations it is nearly in the middle of the fence.
    In that case the fineness of adjustments is halved.

4. the feather stop is too crude to allow fine adjustments even on the old style fence.
    The feather guide is sufficient to keep the rail in place after it is set to the angle you need.

To use the 4 cut method effectively you need to make very fine adjustments of the angle of fence to the guide rail.

To work around the deficiency of the new style setup use a clamp instead of the the feather stop to secure the end of the fence.

At the initial fence setup clamp a clean smooth ended block of wood so it contacts the fence. Make subsequent adjustments by changing the distance between the block and the fence.

Use a feeler gauge to move the fence the amount indicated (and re-clamp) by the 4 cut calculator in the Rick's Kapex manual.

If the fence is straight, and the cuts you make are straight (there is no chance the work shifted during the cut), and you follow the procedure correctly you should be able to get extremely square cuts consistently. When you achieve that cut another piece of ply to keep as a try square so you can quickly reset the fence to the rail in the future.

 

[tjbnwi]

My fence slides under the rail in the low position.

You can set the pivot anywhere along the fence, from right next to the guide rail, all the way out to the end of the table profile.

I find with once the miter head is adjusted to zero it is easier to use a feeler gauge to adjust the key. If you adjust the miter head off zero it will affect the miter angles due to no separate adjustments for the miters on the angle head.

I've always heard the older MFT's were worse then the 3 for the fence and miter head system.

Tom

 

[Michael Kellough]

The protractor head is better on the newer style but they're still only good enough to use as a rough guide.

I cut a lot of thin stuff and even the old style rail fence is too thick for me sometimes.
But, when it comes to using a bevel gauge to set the fence to the rail I wish the rail fence were taller.

 

[tjbnwi]

Do you have a backer sheet you could use to raise your thin products? I use a 3/8"" piece of plywood under 1/4" and 1/4" plywood under 3/8".

Tom

 

[grbmds]

"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 hope I'm not going off-topic here but I'm trying to understand the reported problem and, actually whether it really is a problem or not.

Is the above quote from a previous reply correct? I don't know since I've never done the 5-cut test.

If this is correct it would appear that the difference shown between one end of the cut to the other is just over 1/128 inch (excuse the use of imperial measurements but that is what helps me put this in perspective since that is what I'm used to). If I'm not right about this, please tell me. I'm trying to understand this so I can apply it to my work if it is important.

If I'm correct, would that level of accuracy be below an acceptable level of squareness or accuracy for most of you? That seems acceptable to me. I don't believe I have ever been able to consistently get that level of accuracy with other tools. I recently made drawers for a workshop table using the TS55 and Domino. The first drawer I assembled matched any other drawer I've ever assembled in squareness with much less effort. In fact, it might even be squarer, but the difference is not perceptible. (Just so you know, I am and always have been a perfectionist and hate it when things aren't square or accurate.)

I have never used the 5-cut method because I have found that my accuracy and squreness is acceptably achieved with squares I have. Since I have been using Festool, I find it is much easier to achieve a very high level of accuracy and squareness. I do have square up the MFT fence and guide rail once in awhile, but I find that a small price to pay (plus I bump the rail once in awhile so I could be less clumsy).
Thanks for any input any of you can give. I might be looking at this all wrong.

 

[Qwas]

Yes, it is true but you should run a test yourself to prove it. Put your test board in place on the MFT. Go to the far corner of the board touching against the fence and insert something ⅛ inch thick (you can use the base of a drill bit) and tape it in place. Now make your cuts and measure the error afterwards. You should have ½" error (plus whatever was in your setup before the test).

The problem with saying 1/128 of an inch is fine, is we don't know how long the test piece is. We're trying to measure an angular error by measuring linear errors. Let me explain by using some extremes. If the test piece is 1 inch long then 1/128 error indicates a ½ degree error which is a very noticeable error. If the test piece is 48 inches long then 1/128 error indicates a .001 degree error which I don't think anyone would ever notice with their eye.

Also the error depends on what you're trying to accomplish. If you're cutting a top for a table then it's no problem.  But what if you're routing for an inlay? Some tasks may require that level of precision, although I would think you're right at the limits of the Festool tools. You have inaccuracies in the guide rail, slop in the guiding mechanisms for traveling down the rail, saw blade runout, spindle runout, etc.

 

[grbmds]

Yes, it is true but you should run a test yourself to prove it. Put your test board in place on the MFT. Go to the far corner of the board touching against the fence and insert something ⅛ inch thick (you can use the base of a drill bit) and tape it in place. Now make your cuts and measure the error afterwards. You should have ½" error (plus whatever was in your setup before the test).

The problem with saying 1/128 of an inch is fine, is we don't know how long the test piece is. We're trying to measure an angular error by measuring linear errors. Let me explain by using some extremes. If the test piece is 1 inch long then 1/128 error indicates a ½ degree error which is a very noticeable error. If the test piece is 48 inches long then 1/128 error indicates a .001 degree error which I don't think anyone would ever notice with their eye.

Also the error depends on what you're trying to accomplish. If you're cutting a top for a table then it's no problem.  But what if you're routing for an inlay? Some tasks may require that level of precision, although I would think you're right at the limits of the Festool tools. You have inaccuracies in the guide rail, slop in the guiding mechanisms for traveling down the rail, saw blade runout, spindle runout, etc.

All good points. Plus, you're right that it depends on the project, but I'm trying to put this all in perspective of how accurate I could expect. I don't have any problems with the accuracy of the MFT/TS55 after squaring the fence up (using one brand of dogs or another) and then squaring the rail to that.

The first post in this thread indicates 2 mm over 1000 mm which would indicate .5 mm (2 divided by 4). Other information indicates subsequent tests yielded more or less than that I think. .5 mm is just over 1/64" and that would be over 1000 mm (40") according to the original post.

Obviously if this becomes and additive problem, as in miters for picture framing, I suppose it would be a problem. However, there are many instances when, even this, would be very accurate. The problem with me is I've never thought of it in terms of these numbers and comparisons. I have merely checked squareness with my reliable squares and measured with tools as accurate as possible and have not have any real problems. It just seems as it, for most things, the level of accuracy is, not only acceptable, but at a very high level compared to most tools. Plus, when parts are assembled, many times they need to be adjusted and trimmed to fit anyway. If you are within 1/64, and escpecially 1/128, that isn't much trimming. On occasion, I wish I had more to work with.

 

[Qwas]

Acceptable error is one of those things that varies from user to user. It's a tough argument because each person sees it slightly different. I see many guys spend $200 for a square and they will use it daily, while others will only pull it out to test and compare to their other squares that they use daily. Both ways are okay, it's just a matter of personal preference.

I don't know if Paul was really complaining about his accuracy so much as pointing at that his errors increased as the test piece got smaller. At least that is what caught my attention. I do agree that an error of ½ mm over a length of 1000 mm is what I would consider acceptable.  [smile]

 

[Tinker]

I have dogs with pedigree Tom from Tool Improvement (I think that's where i got them) and a few sets from RMW.  I find there is a very slight difference between the two suppliers.  I have no idea (don't own an accurate set of calipers.) which is most accurate.  All of RMW's dogs seem to be same, so those are the ones I use all the time except for the lock dogs from tom that I use permanently to set my fence.  Richard has several different dogs in his supply and I have no problem if I use the lock knobs to hold them permanently in place. If I am using the rail to fence for squaring, I check with my 12" triangle whenever I need extreme accuracy.  If using dogs for fence AND for setting rail, I find that, for my work, it is always square. 

I do make test cuts whenever I intend to make frames where all of the corners need to match up.  Often, a touch with block plane of sanding block will be needed to bring the 4th corner to purficshon.  There are times when I am only able to catch a few minutes to half hour to work on a project. I might make all of my cuts for a frame, check them out by clamping on the MFT.  they might come out perfect.  I number all of the pieces, wrap them into a pile using blue tape, set on another bench and not get back to glue for several days, maybe even weeks.  By the time I do set up for final gluing, that last corner does need some touching up as a result of uneven shrinkage.  If I were like my dad when he was alive, I wood probably start all over, or not accept a piece unless it measured absolutely perfection.  For me, a little touch up with sand paper is fine. 

Qwas is right on with his comment.
Tinker

 

[grbmds]

I have dogs with pedigree Tom from Tool Improvement (I think that's where i got them) and a few sets from RMW.  I find there is a very slight difference between the two suppliers.  I have no idea (don't own an accurate set of calipers.) which is most accurate.  All of RMW's dogs seem to be same, so those are the ones I use all the time except for the lock dogs from tom that I use permanently to set my fence.  Richard has several different dogs in his supply and I have no problem if I use the lock knobs to hold them permanently in place. If I am using the rail to fence for squaring, I check with my 12" triangle whenever I need extreme accuracy.  If using dogs for fence AND for setting rail, I find that, for my work, it is always square. 

I do make test cuts whenever I intend to make frames where all of the corners need to match up.  Often, a touch with block plane of sanding block will be needed to bring the 4th corner to purficshon.  There are times when I am only able to catch a few minutes to half hour to work on a project. I might make all of my cuts for a frame, check them out by clamping on the MFT.  they might come out perfect.  I number all of the pieces, wrap them into a pile using blue tape, set on another bench and not get back to glue for several days, maybe even weeks.  By the time I do set up for final gluing, that last corner does need some touching up as a result of uneven shrinkage.  If I were like my dad when he was alive, I wood probably start all over, or not accept a piece unless it measured absolutely perfection.  For me, a little touch up with sand paper is fine. 

Qwas is right on with his comment.
Tinker

Thanks. Also, good information. I really have not had a problem once I used the MFT and TS55 a little while. I find that, overall, the two together give me much more accurate and square cuts. Once I put the numbers in perspective, it would be hard for me to notice such a small amount out of squareness unless the nature of the project added the tiny differences into one larger difference. I feel better now. I don't feel the need to really do the 5-cut test as I realize I'm doing OK squaring it up with any of the dogs I have and my squares. Thanks again for the information.

 

[paulhtremblay]

5 more tests today, and I still can't get acceptable accuracy.

First, there seems to be some confusion on how I measure the inaccuracy. First, I subtract the difference between the two measurements. Then I measure the perimeter of the board I cut. So for my first post, I got a difference of 2mm. My perimeter was 570 mm X 4, or 2280. So my inaccuracy is 2mm over 2300 mm. Put another way, if I could fit a 4 X 8 sheet of MDF under the guide and cut it the long way, one end would measure 2mm shorter than the other.

This error may seem small, but it means I will have a hard time fitting together square joints. I am going to fit together MDF for cabinets with dowel joints. If the cut isn't square, you will see it.

Back to today: I did my first test with the qwas dogs (instead of the fence) and got a result of .2mm over 2,000 mm. Perfect, I thought. But when I repeated the test, I got 2mm error over about 1,800 mm. I made one more test and got 2 mm off again.

I then aligned my fence and did a test and got 1.2 mm off over 2,200 mm. Repeated the test, got 1 mm off over 1,800 mm.

The saw is not cutting consistently. After each cut, I put a square against the new cut. For example, for one test, I got nearly perfectly square cuts for the first three cuts. Then for the fourth, the cut was way out of square for no reason.

Finally, I ripped a piece of MDF around 450 mm long. I moved the board over about 30mm and repeated the cut, and then measured the difference. There was a .2mm difference between the top and bottom. I repeated the test, and got .1mm difference, then .3 mm difference.

 

[semenza]

Finally, I ripped a piece of MDF around 450 mm long. I moved the board over about 30mm and repeated the cut, and then measured the difference. There was a .2mm difference between the top and bottom. I repeated the test, and got .1mm difference, then .3 mm difference.

    So at least from this one test you are basically getting  +/-  one tenth of a millimeter for the consistency of the saw cut? Roughly +/- 1/250th of an inch over about 16" length. I am thinking the problem of the out of square is not with the consistency of the saw.

Seth

 

[Qwas]

There's no need to use the perimeter, just the length of the side. So you have 2 mm error over 570 mm length, or .5 mm actual error (.2 degree error before dividing by 4, or .05 degree actual error).

Your rip test at the end of the post shows some pretty good results. I would adjust for a .2 mm error and then you have a +/- .1 mm error. I really doubt you can get much better than that. Here are some things you can check for to remove any errors.

It sounds like something is moving between tests. We can rule out the fence because you are seeing movement when using dogs too. The first suspect should be the MDF top, verify it cannot move around inside the aluminum frame that holds it. Then check the clamps that hold the guide rail to the table's side. Make sure they are locked down and not slipping. Check that pin on the front one and make sure it is fitting the guide rail slot with no slop (if there is slop, try taping it for now). Next make sure the soft rubber under the guide rail is clean and gripping the test piece good. Then check the jib adjustments on the saw and be sure it is tight enough that the saw can't move sideways on the guide rail. Also verify that your saw base is sitting solidly on there guide rail and it is not rocking.  Check with a square and verify your saw blade is coming down at a true 90 degrees. Check and be sure your saw blade is mounted tight.

If all of that checks good, then I'm out of ideas. You can check with Festool to see if they have any other ideas.

 

[paulhtremblay]

There's no need to use the perimeter, just the length of the side. So you have 2 mm error over 570 mm length, or .5 mm actual error (.2 degree error before dividing by 4, or .05 degree actual error).

Your rip test at the end of the post shows some pretty good results. I would adjust for a .2 mm error and then you have a +/- .1 mm error. I really doubt you can get much better than that. Here are some things you can check for to remove any errors.

It sounds like something is moving between tests. We can rule out the fence because you are seeing movement when using dogs too. The first suspect should be the MDF top, verify it cannot move around inside the aluminum frame that holds it. Then check the clamps that hold the guide rail to the table's side. Make sure they are locked down and not slipping. Check that pin on the front one and make sure it is fitting the guide rail slot with no slop (if there is slop, try taping it for now). Next make sure the soft rubber under the guide rail is clean and gripping the test piece good. Then check the jib adjustments on the saw and be sure it is tight enough that the saw can't move sideways on the guide rail. Also verify that your saw base is sitting solidly on there guide rail and it is not rocking.  Check with a square and verify your saw blade is coming down at a true 90 degrees. Check and be sure your saw blade is mounted tight.

If all of that checks good, then I'm out of ideas. You can check with Festool to see if they have any other ideas.

I did all those tests already. (The saw went back already because of an out-of-alignment bevel; the bevel is now perfect.) I'm out of ideas, and don't want to start making cabinets until I am sure my cuts will come out square enough so I won't have to sweat during glue up.

 

[tiralie]

I'm out of ideas, and don't want to start making cabinets until I am sure my cuts will come out square enough so I won't have to sweat during glue up.

...you're kidding...unbelievable.
Just make a cabinet and you will see that .05 degree is inconsequential.

Tim

 

[Qwas]

Hopefully others will jump in and verify but I think you will be fine for making cabinets.  [smile]

 

[tjbnwi]

You should be just fine.

Tom

 

[grbmds]

If you have done all of the things that have been suggested, you're probably right, there would be nothing else to check. However, the differences you were reporting appear to be tiny and would not make perceptible difference when making cabinets. To provide yourself some perspective, use your digital caliper and open it .2mm (or even .29 mm which was one of the end out-of-square differences you reported). It is a tiny crack of an opening. Anyway just try making 1 cabinet or a face frame before being concerned any further. I think you will be happy with the result.

 

[Michael Kellough]

"The saw is not cutting consistently. After each cut, I put a square against the new cut. For example, for one test, I got nearly perfectly square cuts for the first three cuts. Then for the fourth, the cut was way out of square for no reason. "

You know that can't be true.

There are only two ways the saw could cut inconsistently.

1. The blade is loose. In that case you'd never get a good cut so it can't be that...

2. The jibs are loose so the saw is not snug on the spine of the rail allowing you to change the toe-in
    as the saw cuts. If the jibs are loose you can torque the saw clockwise or counter as you go.

If the toe adjustment is off too much you'll get measurable differences in the width of the piece if the entire blade did not pass all the way past the work. If the blade was not fully plunged before entering the work you can also get measurable differences. Keep in mind measurable does not necessarily mean significant.

 

[paulhtremblay]

If you have done all of the things that have been suggested, you're probably right, there would be nothing else to check. However, the differences you were reporting appear to be tiny and would not make perceptible difference when making cabinets. To provide yourself some perspective, use your digital caliper and open it .2mm (or even .29 mm which was one of the end out-of-square differences you reported). It is a tiny crack of an opening. Anyway just try making 1 cabinet or a face frame before being concerned any further. I think you will be happy with the result.

Yes, .2mm would be excellent over 2200 mm. Well tuned sleds don't get much better.

Unfortunately, I got that result only once. I am getting a full 1mm over 2000 mm. For a 24'' panel cross cut, that means about 1/4 mm. If I make the same error on the other panel and compound the error, that means about 1/2 mm. The top and bottom panels could further compound the problem.

1 mm is 1/32 of an inch. So .5 mm is only 1/64 of an inch, and put that way, the error does seem small.

More disconcerting is how I can make 3 really good cuts in a row, and suddenly, the 4th is bad, where I can probably see 1/16 of an inch between the wood and my square.