5 cuts method acceptable accuracy

[Francis_Beland]

I did a custom MFT workbench using benchdog fence and dashboard rail bracket. I'm calibrating it and I think I'm trying too hard to get the perfect 90 degrees cut. I'm wondering what is the acceptable deviation for woodworking. I was lucky as I achieved 0,08mm/m deviation but I tried to be more perfect than that and I messed it up. After some trial & error, I'm now at 0.2mm/m deviation. Is it acceptable or should I fine tune it again?

 

[semenza]

    You've got two tenths of a millimeter over a length of one meter?

    Roughly 1/128th of an inch.

  That is a small enough deviation that a few specks of saw dust could account for it. Or a bit of fuzz against the dog or what not. Or a slight hand movement on the saw.

    I'd leave it alone. Think you will be hard pressed to get any better. Although you had it better so I suppose ....................

Seth

 

[ChuckM]

0.2mm over 1000mm? 0.0002mm!  You sure?

Some squares' tolerance is less than 0.01mm over 10mm.

 

[Francis_Beland]

My last cut is 29,58mm front and 29,35mm back et the lenght of the cut is 281mm. If my calculation is right,

((F-B)/4)/L= 0,000205mm per mm

If I multiply by 1000 to get the deviation for 1m, I get, 0,205mm per meter.

Maybe my calculation is not good.

 

[Francis_Beland]

    You've got two tenths of a millimeter over a length of one meter?

    Roughly 1/128th of an inch.

  That is a small enough deviation that a few specks of saw dust could account for it. Or a bit of fuzz against the dog or what not. Or a slight hand movement on the saw.

    I'd leave it alone. Think you will be hard pressed to get any better. Although you had it better so I suppose ....................

Seth

Yes I had better but I was lucky. I should have let it like that. That's my problem. I want it perfect. I tried to use a piece of paper as a filler gauge to try to be more accurate but I changed it to the wrong side and when I tried to come back to where I was, I wasn't able to achieve the same measure.

 

[Steve1]

.2mm over a meter is nothing in woodworking.

(only rarely is it out of spec in metalcutting either)

Call it a day and have a beer.

 

[kevinculle]

You were definitely at an acceptable error level at 0.2mm/m and there is no perfect in the world of metrology and calibration, just the limits of the measuring tool you choose.  I recently calibrated my Kapex and found a method that helped me dial it in more systematically and quickly.  I made a wood spacer to use between the right end of the miter scale and the face of the casting next to it...the spacer needs to be a thickness that places the miter scale near but slightly to the right of the correct 90 degree setting. You can then use feeler gauges to better control the process of finding the best setting.

 

[Crazyraceguy]

My last cut is 29,58mm front and 29,35mm back et the lenght of the cut is 281mm. If my calculation is right,

((F-B)/4)/L= 0,000205mm per mm

If I multiply by 1000 to get the deviation for 1m, I get, 0,205mm per meter.

Maybe my calculation is not good.

What you are getting is actually 4 times the actual variation. By using the 5 cut method and measuring the taper of the last cut-off, you are adding the deviation of each corner.

Remember, this is wood, it moves more than that. It also has some flex and compressibility that can change things far more than the measurements you are seeing.

 

[Francis_Beland]

Thanks for your replies. I'll leave it like that.

 

[Cheese]

FWIW...this quote is from the Festool Kapex Manual Supplement:

"The calibration method described here is based on compounding an error by a factor of four. This makes it easier to detect extremely small calibration errors. However, care should be taken in over-using this calibration procedure because it has such a fine accuracy that it could be easy to get carried away and try to over-calibrate the saw. The factory calibration threshold is ±0.16°, but this calibration procedure is capable of measuring errors as low as ±0.001°, which is nearly impossible to obtain in actual practice."

So your .2mm/m measurement equates to an actual deviation of .002" per 39"...I'd be happy with that.  [smile]

 

[semenza]

FWIW...this quote is from the Festool Kapex Manual Supplement:

"The calibration method described here is based on compounding an error by a factor of four. This makes it easier to detect extremely small calibration errors. However, care should be taken in over-using this calibration procedure because it has such a fine accuracy that it could be easy to get carried away and try to over-calibrate the saw. The factory calibration threshold is ±0.16°, but this calibration procedure is capable of measuring errors as low as ±0.001°, which is nearly impossible to obtain in actual practice."

So your .2mm/m measurement equates to an actual deviation of .002" per 39"...I'd be happy with that.  [smile]

The OPs formula has the 'divide by four' in it to get the final .205mm  In either case I'd be happy with that too.

Seth

 

[Francis_Beland]

If I calculate the angle of the error, the result is 0,011 degrees. I just checked Woodpecker's website and it states that their 1281 square tolerance angle is 0.08 degrees. As Steve1 said, I'll call it a day and take a beer. Thanks

 

[woodbutcherbower]

Yep. Me too. And I’m picky and obsessive like you wouldn’t believe. Most of my squares are made by Joseph Marples in Sheffield, England. Even their top-dollar ‘Trial’ range only manages 0.025 degrees. Enjoy your cold one.

 

[Michael Kellough]

I made a large cutting table (a long time ago) with the MFT rail hardware and fussed around with the 5 cut method until it was nearly perfectly square to the fence. An imperceptible difference over 30 something inches. As soon as I simply changed the height of the guide rail the accuracy dropped by two orders of magnitude. Still very good but also a demonstration that the Festool wedge-like height brackets can't be depended on. They don't shift when clamped but can't be relied upon to stay square to the table while adjusting or locking.

The Dashboard brackets don't have the same problem. If you keep pressure against the same side of the slot while adjusting you can be reasonably certain that the rail will still have the same relationship to the fence when changing the rail height. There is the potential to knock the rail holding part of the bracket askew when raised high and there is less contact between the mating parts.

 

[Crazyraceguy]

The OPs formula has the 'divide by four' in it to get the final .205mm  In either case I'd be happy with that too.
Seth

It does and I definitely missed it.

 

[sherbert409]

Unless you are running a machine shop 0.2 mm per meter is stupid precise. That is actually 0.011 degrees, or 41 seconds of an angle. That’s the thickness of a playing card over a meter.

Sent from my iPhone using Tapatalk

 

[mino]

For a square-based cut, that is as good as it gets.

From my measuring, the FS/2 rail itself has about 0.1mm/m deviation/bend. Which is actually pretty good - engineer's straight edge needs to be DIN Class 1 to be better ... normal ones are worse.

The Makita rails I had were off by about twice compared to the Festool ones.

If you want/need better accuracy, the common way is to use a parallel guide system for ripping and then use the square for cross-cutting. That way you avoid any compound errors.

Then, for cross-cutting, using something like the TSO GRS 16 (or a fixed squared fence line on MFT/3) should get you better than an MFT dog system itself ever could.

That said, it should do as is. You will find yourself if is causes you trouble.
IMO unless you make big pieces out of laminated chipboard and the likes - which do not move - you should be fine.

 

[Francis_Beland]

For a square-based cut, that is as good as it gets.

From my measuring the FS/2 rail itself has about 0.1mm/m deviation/bend. Which is actually pretty good - engineers straight edge needs to be DIN Class 1 to be better ... normal ones are worse.

The Makita rails I had were off by about twise of the Festool ones.

if you want/need better accuracy, the common way is to use a parallel guide system for ripping and then use the square for cross-cutting. That way you avoid any compund errors.

Then, for cross-cutting, using something like the TSO GRS 16 (or a fixed squared fence line on MFT/3) should get you better than an MFT dog system itself ever could.

That said, it should do as is. You will find yourself if is causes you trouble.
IMO unless you make big pieces out of laminated chipboard and the likes - which do not move - you should be fine.

Thank you for your reply. I don't NEED that accuracy but I'm on the perfectionist side where I want everything to be as perfect as I can. I'm mostly mad at myself that I achieved better accuracy (0,08mm/m) and messed it up for trying to get even better. Now I've locked everything and I promise that I won't try again.  [big grin]

 

[sancho57]

dood, wood moves more than that.

 

[Oldwood]

I find when trying to get a very fine increment of movement like moving the rail bracket it is best to set a stop and use a shim like feeler gauges to move the stop before moving the bracket or use a dial indicator.

The increment of movement is so small when chasing the perfect square that going by eye is not an option.

This gives you a gauge so if you move the bracket .010" and it is twice as much as you needed you can use a .005" shim to put it where it needs to be.

I too am a little obsessive when chasing the perfect setup.