Question about the Shopbot Gantry.

[chris@brtguitars.com]

Hello, I will be purchasing either a Techno or Shopbot within the next 3 weeks. I am trying to decide which would be a better machine for me.
I like the Shopbot and the simplicity but I am a little nervous about the gantry. By that I mean, with the system the way it is, does it sag over time? Is it sturdy enough not to flex when cutting/plunging? I ask because I need acuracy within a few (3-4) thousandths of an inch. The machine will be used daily but not for high production work. Probably only 15-20 parts per day.

[richards]

The Shopbot Specs page gives this information:

Step Resolution

[gerald_d]

Chris, if you really need 3-4 thousandths accuracy(for guitars?) then you probably won't be happy with the ShopBot that gives a 30 thousandths wide tolerance in the spec.

[chris@brtguitars.com]

Ok, Ok, I had a brain fart. I was thinking that I need to have no more than a 1/32" play and I was thinking hundredths instead of thousandths. Which would make 3-4 hundredths right around 1/32", if my math is correct.
Anyhow, 30 thousandths should be more than sufficant. Has anyone had problem with the gantry sagging over time?
Thanks,
Chris

[olecrafty]

The older of my 2 machines is 2 years old and has 2 z axis mounted on the gantry. I use the machine in production work with close tolerances required. After 2 years I haven't had any sag.

Kaiwa
olecrafty@charter.net (mailto:olecrafty@charter.net)

[dvanr]

Chris,

Even if there was sag , surfacing the table would remove the effect of sag. Any spot on the table with the router over it would have the same distance from Router bit tip to table surface.

[mikejohn]

Dick
Only if you have vacuum hold down, and the material 'bends' to the shape of the table.
It's the distance from the router tip to the material that is important.
..............Mike

[chris@brtguitars.com]

Thank you Kaiwa. That's the answer I was looking for. I have been wanting to get a cnc for years and now that I have the cash in hand, I wanted to make sure I get the one that will work for me. The gantry was my only concern with the Shopbot.
Thanks,
Chris

[dvanr]

Chris,

Out of curiosity I checked what the flex of the Y gantry and the X -rails was on the Alpha96 that I use.

Several botters have already checked the Y gantry from the middle to the table surface, I've added a couple more tests. The dial gauge is located at 24.5 and the different measures are the different positions (y=) of the y car and the effect it has. X=16.5 inches, the gauge is zeroed with the Y car at 49.
y=z= 2.50 8-.0024 13.5-.0027 19-.0031 24.5-.0031 30-.0031 35.5-.0031 41-.0015 46.5-.0007 490

Second test is done with the dial gauge close to the X rail but still measuring the bottom of the Y car. Dial Gauge is located at 2.5"

The numbers start to look interesting!
y=z= 2.5-.0040 8-.0047 13.5-.0043 19-.0039 24.5-.0354 30-.0028 35.5-.0024 41-.0012 46.5-.0004 490

Moved the gauge down on to the U-channel side rail and measured up to a tooth on the rack at 18" and moved the X car to the middle and the y car away. Zeroed the gauge.
x=y=z= 50490 1849-.0039 1824.5-.0060 180-.0079
I was confused at first that the last number (-.0079) is bigger than the first number of the second set (-.0040), almost identical load in each case. The X-rail in the second case was already deflected -.0039, adding the -.0040 gives the -.0079.

It seems that the X rails play a bigger part in the change in Z than expected. Moving the Y-car from one side to the other not only changes the flex in Y but the flex in X too.

Mike,
I took sag and flex as meaning the same thing. We are all happier with things that flex, rather than sag.
My point was that with the flex and the levelling routine you will virtually have a flat surface and that most materials will conform closely to the surface with out the need of vacuum. The end result being parts with in the machines specs.

[jsfrost]

Dick,
Interesting tests.

Are the data from a single test? Is the data similar or significantly varied wheen returning to the same point? Curiosity, not nit-picking.

Jim

[gerald_d]

See also this post (http://www.talkshopbot.com/forum/cgi-bin/discus/show.cgi?tpc=7&post=21745#POST21745) (scroll down to pics with weight & dial gauge)

[dvanr]

Hi Jim,

The data is repeatable to .005mm , I ran the tests a number of times and the numbers agreed with what had been read previous.

There is a glaring error. In the second set z=-.0035 not -.0354. ( I converted my measures from metric to imperial)

A fourth set of numbers based on the third set should have been included.x=y=z= 18490 1824.5-.0021 180-.0040

If you do a subtraction you get an idea of how much Y flexs compared to X.
x&y.......x...........y
-.0035 - (-.0021) = -.0014


What I'll do is run the test again and collect X flex for the same y car points as Y flex. Subtract the two and see what Y does independant of X. I think it will be a close approximation.

You can see that the y-car moving across has two effects.
A full move across (leftside,middle,rightside) will have:
leftside X rail max deflected , Y rail minimum deflected.
both X rails half deflected , Y rail max deflected
rightside X rail max deflected . Y rail minimum deflected.

Gerald,
Y car carries a router that weighs 13.5lbs, so I looked around the shop for some weights to repeat your test. In my present state, head cold and sneezing I couldn't do it. Contented my self with cold meds and running the ycar , back and forth , back and forth .....zzzz

[mikejohn]

I'm a little confused in how you are measuring the deflection.
When the table was 'surfaced' (levelled) then presumably the deflection was present.
So if,as Dick says, you are measuring deflection to the table surface, what has happened since surfacing the table?
If you were to place a very straight edge acoss the table, we would expect, because of sag, a dip in the middle. But why isnt the deflection constant to when the surface was levelled?
.............Mike

[dvanr]

Hi Mike,

In order to measure deflection I placed a gauge much like Geralds picture in the linked post above.

I moved the y car all the way to one side and zeroed the dial, then I moved it back over the dial and read of the deflection of the Y-car rails. Its not perfect, because the flex in the X rails changes as well.

I am not measuring flatness of the table , just the change in height of the gantry over a point on the table when the Y car is as far away as possible and then when it is over the measured point. The deflection should be constant to when it was levelled.

What happens IF the Y rails have less flex than the X rails and you surface the table? Will the surface be a tiny bit concave or convex?

I'm not real happy with how the measurement was done. The best way would be to take the Y-car off each time, zero the dial at a point and then put the Y-car back on at that point and measure.

[gerald_d]

Mike, if you want to look at it this way, my little test was more or less looking at what an up-cutting spiral would do versus a down-cutting spiral. (If the spiral cutter makes 25 lbs axial force, the job will move either away or towards by 0.009" depending if the spiral is up or down. Beware, the 25lbs was sucked out of my thumb - actual spiral cutting forces are probably much lower, but a blunt plunging/planing cutter might get there...).

Dick's test looks at how the system flexes purely under its own weight while the gantry/car moves around. For this, one can flatten/level the table, but cutting forces are still the major causes of flex.

[mikejohn]

Gerald, Dick
Now I understand.
Measuring from the same point on the table 'remove' the Y car, take a reading, then 'replace' the y car.
I imagined you were taking a reading from the y car at point a, moving the y car to point b, and taking a second reading at point b.
Thanks for clearing it up for me.
A little more information than on how many routers to a z carriage.


..................Mike