DIYbanter - View Single Post

[email protected]

Hi Tim,

wrote:
Better not look at that video, Ed. Unless the guy tapping the fence
into a new position between measurements has calibrated knuckles, he's
certainly doing trial and error. ;-)

I'll have you know that those knuckles have been certified tracable to
NIST standards! ;-)There is no "trial" of the fence setting in the
video. Just measurement during an adjustment process. The "trial"
comes when you make the test cut with the machine. The error is
reflected in the accuracy of that test cut. There is no error if there
is no inaccurate test cut.

My approach was to do the five cuts, then measure the taper on the
strip with a digital caliper. I put an indicator with magnetic base on
the outboard end of the fence, then try an adjustment, say 10 or 20
thou. Then repeat the five cuts, and measure the new taper value. If
10 thou fence adjustment reduced taper by X thou, then adjust the fence
another (current taper/X) x 10 thou. One more test cut, and Bob's yer
uncle, it should be right on.

Lots of trial and lots of error Tim. Let's see here, the first five
test cuts all have error. Then you make a measured adjustment and five
more test cuts with error. This "calibrates" your adjustment process.
So, then you make your final adjustment and another set of five test
cuts (not "One more test cut") and only after all 15 test cuts is "Bob
yer uncle". I'm not saying it's wrong. I'm not saying it's
inaccurate. I'm not telling anybody to avoid it. I'm just saying that
I don't like it. It's not something that I find productive.

This morning I tried a modification to this. Instead of putting the
indicator at any position along the fence, I put it L inches away from
the fence pivot, where L is about the length of each side of the test
panel. With my 10 thou fence adjustment, I was basically finding a
calibration factor for the system with the indicator wherever it was
and the test panel whatever size it was. By keeping these dimesions
the same, the calibration factor is just 4 (the amount of the
squareness error is increased going around the four corners of the
panel).

That's very good.

I roughly squared the fence with a $6 combination square. The first
test on a ~24" panel gave 0.180" taper. I put the indicator 24" from
the pivot, adjusted the fence 0.045" ( = 0.180"/4) and the second test
gave me 0.005" taper over about 24".

I'm sure that with a little math you could figure out the exact spot to
put the indicator. This provides you with a predictable mechanism to
use for monitoring the adjustment of your fence. But, you would do one
better to have a mechanism to monitor the actual setting of the fence
(it's actual angle). That's what you get when you use the square with
the dial indicator. Direct feedback on how close the fence is to 90
degrees.

Although I've heard of a guy who
aligned his contractor saw to within 0.000050" of true, this squareness
is good enough for me. (I wonder what ever happened to him? ;-)

He probably went to a very hot place for lying. There is nothing on a
contractor's saw which is stable to 50 millionths. Nothing. There are
people who buy these low cost digital indicators which can read to 50u"
and suddenly they become a Metrologist and all around expert on
machinery setup. Don't you believe it!

I expect some cosine error since the fence pivot point is offset from
the fence face, and my magnetic indicator base isn't terribly rigid,
but even if a third test cut was required it would be no biggie. I
didn't time it, but this probably took me ten minutes to square the
fence from scratch.

Not bad. You see me do it real-time in the video. No CGI; no cuts, no
time lapse video, and no stunt double. How long do you think that is?
The whole video (all three procedures) is less than 5 minutes (4:66).
Squaring the fence took about 2 minutes and I was deliberately going
slow so that people could follow what was happening.

Now, if I had a TS-Aligner and an 18" precision square to square a
sliding table fence, for instance, rather than measuring, tapping,
measuring, tapping, etc., until I was happy, I'd try a little different
process:

1) Measure out-of-squareness. Call this X.
2) Move the TS-Aligner to the sliding table. Use one hand to hold it
against the fence at a position 18" from the fence pivot.
3) Use the other hand to tap the fence a distance -X and lock the fence
down.
4) Re-check out of squareness. It should be zero, or pretty close to
it.
5) Done.

Sure enough. Yep, I'm sure it would work. But, if you're already
going to do 2 or three iterations of the adjust/check cycle, then the
extra effort doesn't really save you much.

What I show in the video is a procedure which is easy to understand and
follow for those who haven't done it before. "Step 1, step 2, step 3,
repeat as necessary, etc." However, after a while you realize that you
don't really want to bring the indicator back to zero. You actualy
want to go a little bit past zero. And, with a little practice it can
be done without much thought on the first adjustment. But wait! The
real expert discovers that the change in reading on the indicator can
be slowed and even halted by adjusting the fence while it is moving. I
do this all the time. There's not even a real need to establish a
reference (set the indicator to zero). So, for me it's just too much
bother to try and quantify the amount of correction needed in the fence
angle. I just do it.

There's no magic to a test panel -- it's just something of convenient
size that comes out of the offcut bin for a few minutes and goes right
back slightly smaller.

No problem. I'm just saying that it's not free.

Actually, the right setting with the least investment in time and money
is my definition of better. I don't have the square, so test cuts is
better for me. You do have one, so no test cuts is better for you.

Try it with a smaller square. All woodworkers must have a square,
right? Like I said, 0.001" at 6" is the same as 0.003" at 18" - or
0.004" at 24" which rivals what you described above.

A corollary is that if your customers have 18" precision squares, the
method in the video will help them, if your customers don't have such
squares, the video won't help them.

Not quite.

It sounds like you don't have a rotating stop for the zero position on
the fence. I'd strongly suggest looking into one. I set mine two
years ago, and after resquaring the fence this morning, the stop was
still dead on. This is with moving the fence between the front and
back of the table very frequenly over that time. If someone was
resquaring the fence from scratch each time it was moved, I could
certainly see why they'd avoid using the 5-cut method to square it.
With the rotating stop, set the fence against it and you're squared
without any measurement or adjustment.

As you can see from the video, I have the large Excalibur. It also has
the rotating stop but I just don't use it. Too many things go wrong
with stops. After a few years in the machine shop, you learn to check
everything all the time.

The scribed line would then be a nice back-up to periodically check the
rotating stop hasn't moved.

Stops and lines are great for quick and rough setups. But, when I need
something to be accurate, I trust my square and indicator.

Thanks,
Ed Bennett