Virtual Leveler

[steven_seim]

Greetings. I have a suggestion regarding an enhanced table base coordinate system.

I built a hydraulic steel frame for my bot, and I use a laser optics bench as the working table. I can raise the table such that it touches the bot frame, and/or can be lowered to the floor.

Careful as I was to machine and level the frame, I still want that last 5%.

I would like to suggest a virtual leveler. Give me a keystroke combo that prompts me to touch the table at all four corners of a projects working area. This area could be the entire table or just a square foot. I'll have my keyboard with me, and use the arrow keys. I'll put a 1/1000 brass foil shim on the table. When I cant move it, thats the x,y and z limit for that corner. Repeat for the remaining three.

Presto! I have leveled the table and set the working limits.

If this message is inappropriate I apologize.

If this can already be done with combinations of VL, VA, ZT and the like, I'll figure it out by cracky!

[roney]

Steven,
By cracky is it...with all due respect, I'm not quite as impressed with your leveling scheme, since you can insert a 1" flat-bottom cutter and run it back and forth (via a file)to level the sub-table, as I am with the description of your actual table...sound pretty wild!!! Got any pics?
jrc

[Gerald D]

Hi Steven

I like your lateral (level) thinking! Roney is right if you want to level the TABLE, but sometimes we need to level the reference surface of the object to be machined. A very simple example: We cut small wooden signs with a V-cutter, where the depth of cut produces the "fatness" (width) of the letter strokes. When a customer brings us his favourite bit of old oak, it is often warped and we can spend a lot of time trying to pull it down flat. Your "virtual leveller" will help here.

The brass shims often referred to on this forum, are electrically connected to the spare inputs on the SB. (Simply testing if you can move it by hand, is more economically done with a bit of scrap paper)

[rgengrave@aol.com]

If you use any wood top? you will not get a level table. wood will move with any weather, I day level, the next day not, then again level, if man can bend light, then weather can bend wood, just look at it as a raised panel door, it will bow if it can't move, sheet metal will shrink .00010.02 with heat at 98.2 degree, lead .0001.0 at 87.2 and copper .00001.02 at 56.7, and wood will be .45.04 x 6 times that amount at any temp above 67.3
Nasa 2/6/53 and Nasa 9/6/99

But for a shopbot this is just right if you do not sweat while cutting.

Ron

[oneyra]

Yes Ron I agree with you 100 percent,
therefore our warranty of 30 min. or 30 ft. which ever comes first. It also helps to turn off most of the lights in the shop while showing the finished product.
Man, there I go giving away trade secrets again...
crj

[rgengrave@aol.com]

I new I was doing somthing wrong by leaving the light on..(-:

Ron

[Gerald D]

Ron & oneyra, I don't understand your brand of "humour" - would you guys care to comment on how to deal with distorted surfaces that cannot simply be skimmed flat?

[rgengrave@aol.com]

Gerald all I meant was if you have a metal surface instead of wood? you will never need to relevel.

Ron

[Gerald D]

Ron, I get your point now. I am a "metal" guy by profession and I know of no economical way to build a flat steel table suitable for a SB. Welding stresses cause distortion, and even the sun shining on one end of a table could have the same effect (realise here that I am talking of sheetmetal tables - not heavy cast iron or similar).

The ideal, economical way still appears to be a strong steel sub-frame, that can be welded without too much concern for flatness, covered by a renewable wood top that can be skimmed from time to time.

[rgengrave@aol.com]

I was looking at a cnc table at Ford Motor Co. last summer when I visited my buddy thay runs the cnc machine, they have a metal table with panels, the panels look like 1" solid wood? there were 6 panels, the panels just layed is slots.

He said this seemed to work the best because it could breath better and not warp,When i build my new metal table I think I might just try this.

Ron

[bill.young]

Ron,

Instead of solid lumber, what about using that recycled plastic "lumber" as a top material. I've put in some bulkheads made out of it, and it seems pretty stable, strong, and inert.

Bill

[davidallen]

woodn't MDF work as well and be cheaper?

da

[steven_seim]

Steven_Seim here.

Excellent responses. Thanks. Lets keep this going and focused please. Let Shopbot know multiple people can make real use of this, and this request may take a higher position in the DCR (Design Change Request) list.

Roney, thanks for your feedback. I can certainly use a 1" cutter sure, but I put it to you that its easier to do it in the software. Its common practice in game development to create a table against which polygon distortion is measured. Said another way, you allow the polygon -so- much distortion before you re-gen it. The same logic applies here. Declare your working area and surface, but only allow so much distortion before you correct. I will post picture of the frame to my profile sir, and send you a copy as well. If you have Rhino, I'd be happy to send you the file.

GeraldD is on the beam, at least on a similar frequency as my thoughts if I may suggest. Dont assume all is flat, and all will remain flat. Its cake to compensate with the software in this case.

GeraldD, let me comment on my table if I may. Its a 4x8 foot 3" laminated stainless steel laser optics bench, with tapped bolt holds every 2 square inches. It features 1/4" stainless top and bottom, with a steel honeycomb in the middle. It was planed -after- contstruction. Thats how it got so nice and flat. Given its thickness and weight, (500+ pounds) I'm not concerned with distortion, at least I dont think I have to be. As for cost, it was more expensive than the shopbot itself sir.

Oh man I could use this feature; I really could, believe me.

Is the software open source?

Thank you all. More to come. Imagine how powerful this feature would be, and imagine also your increased working options -and- safety gains!

Its a righteous cause.

[steven_seim]

Steven_Seim here again.

Drat Roney, I forgot to mention that I can raise and lower the table 4 feet. I can touch the bot frame, or lower it within 8 inches of the floor. Careful as I was in frame construction, the working table is not perfect all the time.

Yes, I can indeed plane the material in advance, no question. For me though, its a logical step to position the table, then declare my working area and surface. Secure my raw material and walk away.

If I may, I would like to stand and back GeraldD. Not only should we not assume all is flat and will be flat, dont assume the bot is equipped with a router! I use a router for the big stuff, then switch to a 120,000 RPM air driven dentist drill.

Comon Shopbot! Bring it on baby! Yeah! Uh! Virtual leveler! Yee haw!

[danhamm@abccom.bc.ca]

Well shopbot while your at it, make the table the
forth axis, and put a laser level compensator on
the x axis then we can just put anything on the table it will always be level..

[steven_seim]

Steven_Seim here.

I use a laser to level the table now.

I raise or lower the table to the working height I want, then put one of those spinning laser levelers at the tables center. I then measure/adjust accordingly.

[danhamm@abccom.bc.ca]

On the new co2 lasers, there is a laser compensator
that measures the distance between the head and the object on the table and keeps the head a set distance from the object on the table. so if you
put and uneven object on it and want to cut a
1/2 in. grove unevenly It will..In your case it would compensate for the difference in level from corner to corner..

[Ted Hall, ShopBot Tools]

As I understand Steve's first question, what this means is that assuming the table is already a perfect flat plane, but perhaps not parallel to his ShopBot's motion system, can we locate it's 4 corners and then compensate for the out of plane condition? Hmmm ...

Finding the location could easily be programmed now ... and could be done with touch plates or prox sensors to make it fully automatic. However, the software compensation can not be done automatically in the current software. Our plan has been to add a new feature to the [FP] Command that would allow rotation of a file before cutting (similar to the way scaling is now handled). What we would like to do is make this a full 3D function. Once implemented, this function among other things would allow an easy adjustment of the cutting/machining of the file in relation to the known plane of the table.

Now if you were brave ... you could go ahead and program this right now as an external function ... or, as another alternative, if your original file was in Vector, you could just rotate it to the right orientation before generating the SBP file ... (I kind-a like that last one as a nice challenge!).

And for Dan's suggestion, seems to me that if one wanted to put some screws on geared stepper motors, there would be no problem using the 4th channel to manipulate the table ...????

[rgengrave@aol.com]

Bill I called everyone around here and no one has it, do you have a number I can call to see if I can order it?

Thanks
Ron

[donframbac@aol.com]

Ron, They have it at my local Home Depot. Called "Trex".

Don

[rgengrave@aol.com]

Thanks Don, I will give then a call, is this in 4x8 sheets?

Ron

[bill.young]

Ron,

I'm not sure how they sell "Trex", but the material I've used is designed for bulkheads and is 2 1/2" thick by 8" wide or so and is tongue and groove. In that size it's over-kill for a table top, but I think that in thinner boards it might make a good length-wise first layer for a top; sort of a "sub-flooring". Then you would put the 4x8 sheet of whatever you wanted on top of it; maybe some sort of thick plastic for the ultimate in stable table tops?

The advantage to this I think is that it would be inert enough to not move with temp and moisture changes, and also not sag between cross-pieces on your table like MDF tends to do. MDF also absorbs moisture like crazy if you don't seal it well.

Bill

[rgengrave@aol.com]

Bill thanks for the info, I will check and see what the have, this just might might take care of the problem of bowing?

Ron

[rgrim]

Trex also come in the same size as a 2x6 up to 20' long. Trex is used to replace 2x6 treated decking . it also come in 1" also for decking.

Bob

[alton]

Ron,

I agree with Bill on the plastic.
However be careful. Many plastics have problems
with static electricity. If used for a table
top some will attract sawdust and virtually
be unable to wipe off clean. UHMW or plexiglass
may be undesirable. While styrene or HDPE
would be better. Check with the supplier on this.
I am sure that there is an ideal material for this.
There are hundreds to chose from.
They'll probably let you take samples for trial.

Alton

[alton]

By the way, I don't think styrene comes in
thicker sizes. I would look into HDPE.
It's cheaper than UHMW and seems very
stable.

Alton

[Gerald D]

Be careful with some plastics that absorb water. Nylon is understood to be one of the worst. The boatbuilders here should know that rudder-stock bushes of nylon need a huge clearance when new.

[rgengrave@aol.com]

I stopped by a sign shop near me and took a look at there $89.000 CNC machine, this machine will make coffee...lol, it has a tool changer, material feeder,5hp router,and it still took 2 people to run it.

There table top was 1/2 sheet metal and was tacked to the frame, he put a line on it and there was no gap, he said it has been level for the last 6 years.

I am thinking of doing this? since I am going to build a new table anyway I wonder if this would be the best way?. I can get a 1/2 4x8 for $239.00 and Metal Mart here in town, and weighs 262lbs.

If I was to do this? I could have them make me tracks and set some fences on it to hold material in place?, it would look like 1 big checker board.

I have 20 or so stop locks I use for me table saw and there is no moving them when locked.

Gerald do you think 1/2 is over kill?

Ron

[Gerald D]

Ron, you are asking a loaded question, but I will try my best:

1. A sheet of 1/2" steel by itself is still pretty flexible, unless it is supported on cross members at about 20" intervals. (my choice of crossbearer would be a 2.5" angle, 1/4" thick with wide flange at bottom (plate resting on thin top edges)

2. Of course the cross members must be connected to solid longitudnal rails with legs to the floor.

3. The plate should only be tackwelded to the cross members. 1/2" long welds at 15" intervals. More welds will cause high stress and bowing.

4. Any other fixtures to the top surface should be screwed/bolted (not welded)

5. The major problem is to find a flat sheet from a supplier. Often steel is coiled into rolls at the mill and the middle-men uncoil & guillotine cut to size (Yes, even for 1/2" thick!). Although the major part of the surface may seem to be flat, you often get some guillotine distortion within 3" of the edges.

6. The type of accuracy that I would expect from such a setup would be around 1mm (1/25") for hollows and bumps, but the twist over the whole sheet can a lot more - depends on your cross-members & a level floor.

7. If I wanted to follow this route, I would probably go for a 5/16" or 3/8" sheet, because it is stable enough and is flatter from the supplier. Also it can be "corrected" (with shims) more easily.

8. I would build the whole table, including the setting up of the SB, BEFORE welding the sheet to the cross-members. Then I would run a dial (clock) gauge in the SB and find the bumps/dips in the sheet and try to shim it accordingly before tacking it into place. If the sheet is not level enough, take it back to the supplier, or find another piece of material than can be surfaced by the SB.

Hope this helps - this is more or less how we build "welding" tables for our metalwork.

[sheldon@dingwallguitars.com]

I built my table top out of 1/4" steel (laser cut with jig and mounting holes in place), bolted to 2" square tubes on 12" centers. It took a lot of shimming to get it to within .010". I can easily deflect the areas between the supports .010" with one hand.

We'll see how this will affect the quality of cut (I haven't put it to heavy use yet).

I could see going to a slightly heavier plate like Gerald suggests, but I sure wouldn't want to try shimming 1/2"

[steven_seim]

Steven_Seim here.

Ted Hall. This thread has spun way off course and out of control.

I think you've got it. Imagine the bot frame as the letter U. Now draw a line anywhere between the uprights. The 'anywhere' represents my ability to raise and lower a 4x8 foot, 3" thick, 500+ pound dead flat steel laser optics table to the top of the uprights where the bot is mounted, or down to the floor.

Draw that same line at an odd angle. Let this represent the table as not perfectly in alignment with the shopbot mounted at the top of the uprights.

Now draw a dotted line above and -parallel- to the line you drew that touches the uprights. The dotted line represents the -corrected- tool path. This is a 2 dimensional representation of a 3 dimensional Design Change Request to your software.

I would like the have the option to pick any four 'Z' corners, and have the result be a matrix that makes sure the tool is always aligned with the tables surface.

Yes. I can write a routine to move and position, and turn the bot into a touch sensor by making a circuit between the bit and the table. Cool. No sweat.

Its the integration into your software thats the rub. This is where I can use advice.

I'd like to know more about your planned [FP] changes.

Bottom line. You; Ted, (Shopbot) last thanksgiving sold me a machine that exceeded my expectations regarding movement and accuracy. This is most excellent. If you can compensate for table position in your software, I (and we) can take full advantage of it.

I'll write your company address and see if we can pick this up offline.

Thank you

[longboard007@ev1.net]

Make a steel/ aluminum table.
now onto you steel top fasten a PVC sheet the size of your table to this surface. (Use wilson 800 spray adhesive; you can remove it later with xylene) You'll need to weigh the sheet down all over the place; I used full paint cans, a bunch.

When its all done, you jsut run a mill pattern to level the top. The pvc is waterproof, and has minimal variation with temerature changes.

[birdsofplay]

Assuming that ther is a linear change in level
ie a planar table that has some tilt, THEN ...

Do a Z zero at 0,0 and then take a reading at 96,48.
Now modify all Z plunges by
Z*(currentX*(offest/96))
ie
JZ,.05 * %(1) * &MO / 96

IF you're really masochistic do the same for the Y

Of course if the table isnt linear in the first place, then you're out of luck !

[steven_seim]

Steven_Seim here.

Responding to an off-line comment. it is in alignment with this thread.

It is still crucial to level the table in advance when the machine is assembled or when/if the table is moved. This point should always be up front and clear. I used a machinists level to get the frame within 1/5000" per foot. I tried using a laser, but the machinists level worked out to be more useful.

All said, once you've got the frame good and straight, it would be a most excellent final touch to have the software pull up the rest of the slack, little as it will be.

[curtismai@aol.com]

I am currently looking for a source of the extruded aluminum t slot channels that come on those other expensive machines for my table. After finding some I will then use machineable grade pvc not expanded 1/2 inch think and you have a stable machineable non-porous surface. Anyone know of a source for the t- channel stuff?

[jkforney]

Curtis
Try Eagle-America at this url. If it doesn't work just try www.eagle-america.com and look under router table accessories.

http://secure02.ecommercesoftware.com/cgi-bin/ncommerce3/ProductDisplay?prmenbr=29455&prrfnbr=87992&FROM_CA T_REF=56077

John Forney

[davidallen]

look in McMaster Carr. they have the 1 and 2 inch wide extruded stock. you could use several of these laid parallel to get the same effect.

da

[ron]

I just read the entire thread - I have a question.

Why are pool tables - pocket billiard, snooker and billiard tables for the technical here all made with wooden frames and slate beds? Of course the best of them are one piece slate.

There are also some 'modern' panels with flatness specifications that approach slate - or for the anal retentive granite - available for mere money.

When someone does this I would like to know the cost in % of a new shopbot

Ron

[steven_seim]

Hi. Steven Seim here. Just coming back to see whatever happened to this thread. Thank you for your comment Bob. Great suggestion. I will use it. Thanks for keeping the thread alive as well shopbot.

Nice comment too Bob If I may add. I'm not masochist though. I just want that extra .1%, since I know the bot can deliver on it.

Even with my table hydraulics installed and tuned, once again I could make use of a feature that virtually levels my tabletop once the table is positioned and locked. To repeat, I would like to touch off on the 4 corners of my work area, be it small or huge, and have the result be a table that compensates for any residual offset, tiny as it would be.

Running a pattern is just wasteful and silly, no offense or insult intended in any way. This should be handled in software.

Please consider this if you would. I can get the results ported into a nice app that runs under XP if that spices the pot a bit.

Thanks.

sseim@microsoft.com (mailto:sseim@microsoft.com)

[imserv@imsrv.com]

Steven, 3 points define a plane. Assuming that you table is flat, you can measure your Z variation at (0,0) (0,48), and (96,0)


Draw these 3 points in Vector.

Orient the Vector axis to the 3 points.

Construct or backplot your program into Vector as you would normally do.

Reset the Vector axis.

Generate your program, you are now going to compensate for the X-Y and Z error of your machine throughout the plane of your table surface, within the SBP program. If the Z depth is -.5 at (0,48) and 0 at (0,0), and you cut to a -.1 depth in the program, the sbp file generated by Vector will have a -.6 value when you command a cut of -.1 in Z at (0,48).

What were previously simple Z axis moves will now be 3 axes to compensate for the fact that your table is not perpendicular to the machine axes.

[dmdraper]

Fred,

I immediately cut and pasted your post into my "Vector References" file for future use - but for a different reason than you might think. I recently had a customer who wanted a sign cut into a very thick, very warped piece of hardwood he supplied (he didn't want it planed). I about went nuts trying to clamp the thing down to straighten it out (at best was off 3/8" across an end to end diagonal). Your trick would have allowed me to compensate for the uneven "thickness" of the workpiece!

[steven_seim]

Steven Seim here.

Now your talking Fred. You are so on the money. Yes, the table is flat. Dead nuts, 3 inch thick laser optics bench flat it is.

I can do exactly as you describe, though I use Rhino and Mastercam instead of Vector.

Problem is, (and by problem I mean only irritation) Every time I position the table I need to check and compensate for any residual offset, and adjust the world axis grid accordingly. Stuffing a dial indicator into the router holder and running it around gives me what I need to know. What a pain though!

I still; with respect, put it to the forum that this process would be better, quicker and likely more accurate if any residual offset can be declared through the software, simply by touching off on 3 (or 4) points of the needed working area.

(It would be nice as well if in keyboard mode, the motors could be moved in specified increments, Say .001 every time the PgDn key is struck)

This is evolving into an excellent discussion, especially when thoughts like yours Fred feed folks with good ideas.

Come on folks, bring sb.exe out of 1989 mode. Thats my wish to Santa this year.

Thanks

sseim@microsoft.com (mailto:sseim@microsoft.com)

[donchandler]

Steve,
You can move the motors a fixed amount with each press of the keyboard, as you wish. Just read the screen when you are in keyboard mode. Hold down the space bar while pressing the appropriate key. You can even set the amount of movement to your preference.

[Ted Hall, ShopBot Tools]

Hi Steve,

Since Don already pointed out the incremental moves in keyboard mode ... I'll go ahead and suggest that the target for a Windows ShopBot version is the first of the year ... so, it could be your Christmas present.

The 3d rotation is on the drawing board, but it is lower than the basic windows version and a USB interface in priority at the moment.

In principle, the rotation in drawing software (whether Vector or Rhino) would produce exactly the result you are after. I'm afraid it will be a few more months before it is really easy in a ShopBot Command.

I'm thinking that for some people with uneven surfaces, another option along these lines might be to create a map of the surface topography of their work surface (say with a probe) and use this to provide a look-up correction of a cutting path. This does not create a rotated cut, but would fit a cutting file to an irregular surface.

[steven_seim]

Ah yes; found that Don, Thanks. I've been so used to the windows UI that I missed the obvious in that primitive dos windows front end. Like, duh. My biff.

Thanks for listening Ted. I'll take your comment as one where a solution is pending. Dont hold back on it sir. My guess is that the developement hours required will not be a waste of time. It would be a most swank, often used tuning feature; count on it!

Manipulating the world axis is indeed the answer, but an awkward and temporary one if I may suggest. Imagine a 'botter' crawling around a 4x8 foot table to read the dial indicator, and you have a comical view of what it takes to declare the offset.

sseim@microsoft.com (mailto:sseim@microsoft.com)

[davidallen]

even more useful would be using the accessory axis to probe the surface while the z axis cuts based on that offset.
two contact switches connecting the accessory axis and a floating disk around the cutter could be connected to inputs. one would signal above and the other would signal below zero. the accessory axis would drive to maintain no signal and the z axis would use the accessory value as an offset.
very complicated, but if I wanted to scribe a sign on a natural rock face or tree bark, that's probably what it'd take.

[bwclark@centurytel.net]

David,

Why not digitize the surface and then apply your "cutting" to this surface. Programs like Rhino and Modelmill can do this.

Probably not as easy, but hey the ends are still the same.

Bruce Clark
bwclark@centurytel.net (mailto:bwclark@centurytel.net)

[davidallen]

with the accessory option, the machine would measure and cut at the same time. no intermediate step. it would also allow for non-planar variations without digitizing a lot of points.

[dmdraper]

I use the ShopBot to measure the table surface. It's a little crude but the accuracy is as good as a dial indicator. The process involves writing a little file which opens another .sbp file for output. I measure the surface with the z-zero plate and output the %(3) (current Z location) system variable, you can measure as many locations as you need - it works great.

I used method this to measure what I call the z-axis runout (i.e., how far from perpendicular the z-axis in the fore - aft and side to side directions). My original machine setup using a square was off 1/8 inch over 21 inches.

[toys]

Ted Hall,

Has there been any developement on this virtual leveler at least as a sbp file?

Is it possible to look up an information table that was written with the probe and modify the z-depth on the fly? How?

Any hints on some system veriables that might help? ie what do "start segment size"%(25) and "full segment size" %(26) return?

What is the difference between %(3) and %(33)?

Are there any undocumented commands that might help like perhaps a timer function for timed
adjustments to the Zaxis?

Has anyone taken up the challenge to write the code for something like this?

[toys]

Ok I figured out what what the differance between %(3) and %(33) are.

[toys]

Ok I figured out what what the differance between %(3) and %(33) are.

[toys]

I have been trying to write a routine to do this virtual leveling, and I need to open more then one file at a time is this not possible?

I've used the OPEN(filename) FOR INPUT AS #1 statement but I can't open a second file for input as #2. It reports it as an error.

What I was thinking is that I could open my cut file then sequentially compare it to a file generated by the probe. Then I would open a 3rd file for output and write the new cutfile with new Z values.

Is this not possible?

[bill.young]

Hi David,

The ShopBot programming language is just a sub-set of the BASIC commands...my guess is that you're not going to be able to do what you want (open 2 files at the same time) within the ShopBot language, at least not at this point.


What kind of surface are you trying to work on... are the variations irregular or is it something consistent like a slanted surface? If it's something like a slanted surface, could you change your file by multiplying your z-axis values by a factor (based on your x and y position), instead of reading a probed file?


If it's an irregular surface and you have an old copy of DOS, maybe you could write your program in Qbasic and convert the files outside of the ShopBot software. You could also try Microsoft's free version of Visual Basic, the "Controls Creation" edition. I don't know anything about it but the price is right.

Good luck,
Bill

[chip]

My ¢.02...

No matter how you face it, the table will go out of square eventually. Heat, cold, light, dark, floor makeup and the malleability of steel will vary the level of any table - over the course of time. Humans make machines and steel. Humans are faulty... not our fault %-)

The best way to guarantee "levelness" is to level the table using the ShopBot, and a large, flat bit once a quarter. We, when we can, buy 2" plywood for our tables. It is expensive, but we can operate on them for about two years before replacement - even longer if the foundation is strong enough.

Don't forget about the middle of the table, either... The corners may be in square, but the middle, or the sides, or two feet diagonal from the middle, may have sunk below tolerance.

The best way to keep things in square with a ShopBot is by using a large square and straightedge and "machining the machine" that way.

Usually, considering our flooring, there is not call to take more than 1/8" at the highest point from the table. We can count on dead on accuracy from the machine thereafter. (As long as we don't leave sawdust under the workpiece.)

The ShopBot is a great piece of engineering! I live by them and swear by them. But it is not one of the high end machines that you see in large manufaturing facilities - nor is it intended to be. It is pretty damn close, though, and the measure one takes to make it competetive is absolutley worth the cost of the effort.

Support the Troops!

Chip
www.marshallsmillwork.com

[toys]

Bill, Thanks for the info on basic. I have been looking into that possiblity. I do have old versions of gw-basic but there on 5 1/4" diskettes. I'll look into Microsofts free version. Thanks for letting me know about it, I've been looking arround for something I could download for free. My table is some what random. Its really not that bad for regular cutting. When I'm using my air-turbine it bogs down in some areas. I cut with two passes already and If I could get a more level cut I could possibly eliminate the extra pass.
I'm not using a spoil board system so I can't just
level the table. I've got a sheet of acrylic screwed down to several layers of MDF. So the idea of a virtual leveler would work great. It would at least hold me over until I could sink the funds into a real beefy steel table.

[toys]

I've been playing around writing a routine in basic and I've been able to detirmine the length of the line between points in my shopbot parts files using Pathagarium's theorum, but I need to figure out points inbetween my part file points so I can activate the probe on shorter line segments.

So does anyone know how to calculate the points inbetween two points? I can't clear out enough cobwebs from my brain to figure this out. Please help.

[bill.young]

Hi David,

I think you'll want to find the change in x, y, and z coordinates between your two points, divide each by the number of segments that you want to split your line into, and then add those values back to the coordinates of the start point.

As a simple example, let's say you have a line that starts at 6,10,0.25 and ends a 12,25,1 and that you want to divide it into three pieces. The change in X is 6 (x2 - x1 or 12 - 6) , the change in Y is 15 (25 - 10) , and the change in Z is .75 (1 - 0.25).

Divide each of them by the number of segments ( 3 in this case) and you get a value of 2 for your x step value, 5 for your y step value, and 0.25 for your z step value.

Now add these values to your starting point coordinates and you get these two intermediate points...

8,15,0.5
10,20,0.75

To do it in your program you will probably want to put this in a loop of some kind...just remember that the number of intermediate points is 1 less than the number of segments that you're dividing it into...if you want 100 segments you'll have 99 points in between.

Hope this helps,
Bill

[bill.young]

FYI, one of the panel discussion topics that we're planning for the Jamboree is "Writing part files and programming"

Bill

[toys]

Bill, Thankyou a ton. I spent a all of last night tring to figure the puzzle out and I came up with a solution, but It was quite complex and made my routine run very very slow. I changed my program to use your simple method and it increased the speed by a factor of at least 5.
Hopefully I'll have this thing done sometime next week. I choose to have the routine load my existing sbp files and write out a new file with probing information at a variable distance rate.
I think I'll then take the out put from that probing and generate a 3rd and final cutting file with adjusted Z values.

[bill.young]

Glad it helped.

Bill

[toys]

I finished my virtual leveler. I run my shopbot parts files through it and it generates a new file with an imbedded probing routine. When run on my shopbot this file probes the cutline of my file and generates a new file with corrected m3 moves.

Because of this leveler, I can now cut my 1/4 baltic birch in one pass with my air router. previously I broke a lot of bits and I had to do two passes. I'm so excited about this thing. Thanks to everyone who helped me write the program and Saint Joseph who's feast day it is today. I've just cut a record of 80 puzzles today. That is double my previous attempts.

Saint Joseph patron of wood workers, pray for us.

[stevenseim]

Then on to a 'virtual squarer' then.

I've been using the prt96 machine for small parts for a long time, again on a custom heavy steel frame and laser optics table as mentioned in the first thread. Now that I'm starting to use the entire cutting area, I'm becoming more tuned to the idea of rebuilding the entire YZ carriage.

Aside from a need for locking fine Azimuth adjustments for the z axis on its two axis of tilt, I'm coming to conclude that the software should be able to compensate or 'final tune' the integrity of the entire assembly, not just table leveling. Tightening the nuts and bolts for me always takes the assembly out of square.

A working virtual table leveler is good news, but as a machinist friend told me, "...You must do what you can to physically level the table with the frame, or be willing to accept aysmetric cutting loads on your bits.

With the frame assembly nuts and bolts positioned as awkwardly as they are, I'm now finding it difficult to accurately square the machine as well. As with leveling, its possible to get close, but the software should be the final tune.

So to square the machine, it would be nice to see a 'Virtual Square' routine as part of the software, in addition to virtual leveling.

For the PRT 96, I use the long axis (x) as a base, and assume it is straight.

It might work something like this:

With a line (machine) drawn or cut on the x axis at the far edge of the table, I then draw a line (manually) at a 90 degree angle. With this done, I place the mounted pen or bit at the intersection of x and y, and call a routine that squares the XY travel.

I should then move the carriage on the y to the other tables extreme edge. Disconnecting the y axis motor makes it easier to traverse the two points.

With a very slow move speed, (.05) I move the x axis forward or back until the pen or bit is aligned with the line drawn. With a 'save this' button, the table coordinates stored now knows how to compensate for frame deformation on the XY.

Ideally this routine should allow the selection of either axis as the 'calibration' leg, for those that use the desktop version of the machine.

David, would you mind sharing your results?

Many thanks.

Steven Seim
sseim@comcast.net (mailto:sseim@comcast.net) (address change)

[Brady Watson]

Steven,
Why not just get the whole assembly as square as you can and lay a weld bead on the intersections?

It will never go out of square again.

-Brady

[stevenseim]

Good Point Brady, thanks.

I have been tempted to just weld the entire carriage assembly, after replacing the rails with precision rails from Bishop Wisecarver. My assumption is that welding will produce no better results.

My fear with welding is heat deformation that alters the integrity of z axis positioning. Know what I mean? I am fearful of 'curling' at the edges of the table.

An alternative would be to flip the roller mounts and extend the carriage (which it should be anyway) such that the x axis motors are covered and protected, and weld at that longer edge.

Seems to me to be a lot of work, with again varying results due to temperature change.

It occurs to me that final tuning through the software is what is needed to produce consistent and repeatable results, under nearly any condition.

and again, what is easier, and more acessible to the wider audience, weilding a torch or pushing a few buttons?

Its probably my (retired thank god) time with MS that leads me to conclude a software solution would be easy to produce, and consumable to the larger number of folks that have invested in the equipment.

[Brady Watson]

I don't think that you are going to have any real warpage issues if you weld it. If you tack it up while it is well clamped, then lay a bead, there won't be a problem. I was thinking of a real MIG welder here, not a hobby/arc welder. You can weld 26ga automotive sheet metal without warping it if you take your time. 1/8" steel isn't a problem and readily soaks up the heat. In fact, take a gander at your Z-axis...It's welded together.

Sounds like a neat project for squaring up the table. I'd be interested to see how you make out with it.

-Brady

[stevenseim]

Your points are well taken Brady, thanks. A welded solution triggers 'rebuild' with me. If I'm going to whip out the welder, I might as well build a new one piece unit square to begin with, incorporating z axis azimuth adjustments, and use precision rails all the way round.

I've sent a request regarding what the final leveling solution actually is.

If I can figure something out I will post it for sure, but ultimately, these enhancements, presented in a public forum, should ship with 'dot' revisions of the software as it evolves.

[gerald_d]

Welding vs bolting is a bit of a red herring - the real issue is the "flexibility". We can flex our gantry +/- 0.25" simply by holding it out of square before we switch the motors on.

Once the motors are switched on, then the squareness is locked in for the remainder of the session. . . . . locked in much firmer than welding or bolting will ever do for you.

Elsewhere on this Forum, there has been talk of an auto-squaring routine that will sense markers on each of the x-rails and tweak the x-motor positions accordingly. This tweaking is possible (and necessary) because of the inherent flexibility - not because the corner joints are loose.

In our case, we have adjustable end-stops at the end of each x rail (4 stops in total). We have mechanically adjusted those stops for perfect square. We simply hold the gantry against the stops during the switching on when there is a critical job. End of story. See this post (http://www.talkshopbot.com/forum/cgi-bin/discus/show.cgi?tpc=26&post=2850#POST2850).

[Brady Watson]

I saw that before, Gerald. Looks like a good idea. What do you think about removing the utility strut and replacing it with 2X2X1/8 steel? That would eliminate the flex you are speaking of and with some gussets would be super strong.

-Brady

[stevenseim]

A buddy of mine, who has spent literally months to get his new desktop machine straight and aligned to the point where it is acceptably square mentioned your technique Gerald. Since his machine has the stepper motors with the gearbox, he has to account for gear lash as well as frame flexure.

I would suggest (again) that these are final, repeatable, required steps the software should support. Once the scaffolding is assembled, let the software tune the rest, with the ability to repeat as needed.

Here is what I do now:

I have the bot draw a line on the x axis. I move the gantry out of the way, and manually draw a line 90 degrees to the first.

I move the gantry and the 'chucked' pointer to the intersection of both lines, then I freeze that side by disconnecting the motor and clamping the carriage. (using MO) I keep the motor engaged with the track.

With the bolts loosened but not slack, and the y motor disconnected, I slide the z carriage to the other side, and move the x axis on the opposing side very slowly until the Y traverses that straight line.

Trouble is, when I get it where I want it, I have to hold the interior frame nuts (which a wrench wont fit around) in place as I firm the structure up. This process itself throws off the work to square it all up.

As you mention, alignment from that point forward is fairly fixed while the contoller is on, but it is a process that needs repeating, and made much easier by just responding to a few prompts, and storing the offset in the Table Coordinate base, not wrestling with a (natually) flexible structure.

If I 'ZX' when I freeze the one side of the carriage, the contoller can collect (and store therafter) the offset after I true it up and reconnect the other x axis motor.

I dont know what percentage of owners/users have the ways, means and desire to modify the structure of the bot, but I'll recon that the percentage is higher with regard to those that would take advantage of -software- tuning if it was available.

I could sure use this, and the thread was started in 2001.

I've had the machine a long time, and its been fun, but now that I am working the entire area, I'm really feeling a burn to see the software modernized.

Many Thanks.

[gerald_d]

Brady, there will be less flex, but you will never eliminate it altogether. Synchronised motors at each end will always hold square better than a lot of steel in between.

The question is how to synchronise the motors: Steve suggests to do it "softw.-electronically" - we do it "manu-chanically".

[stevenseim]

It may be also possible that adding steel adds weight that loads the drive motors unacceptably. I do not know this as fact. My (now former) background and history is in the software world.

My gut feeling is actually an attraction to an element of flexibility in the scaffolding. I'm unnerved by the idea of welding the frame. It seems too permanent, for too small a win, though I am not even remotely suggesting you are in the wrong Brady, not at all.

The idea of the customizing the bot over time is again attractive. Replacing the rails with better, using counterweights instead of springs, adding azimuth ajustments; all good; all things that can be done as the owner ramps up and gets savvy.

When I rebuild the carriage the spans the x axis rails for example, I intend to use quik-locks instead of nuts and bolts, sort of like a worm drive saw, whose depth adjustment is quick and easy.

The point (again) is, the software should always be there, ahead of the owner, ready for whatever level the owner wants to take the machine over time.

This same friend that spent months tuning and rebuilding his desktop bot spent hours playing with the gear ratios until it was right for him. That same ability to tune distance and speed should also be applied towards geometry correction, getting the machine to cut a simple right angle, once the frame is firmed up.

From what I've seen, the ability to collect and/or declare the offset amount is already present. Why then not just incorporate that value into motion correction that adds that extra and final bit of goodness?