Overly complicated .sbp file

[jstrasser]

As a new shopbotter, we are testing out our system (PRS Standard 96x48). We have our table level and square to where we have cut a 30 x 90 rectangle into the spoilboard and it looks good in terms of being straight and square.

However, we have just begun cutting 3/4" vc plywood for cabinet parts and we are finding that the pieces are not of equal size even though the AutoCAD rectangles are all of equal width.

I looked at the sbp file (FE) to see what I could dope out and was surprised to see that for cutting out 6 rectangles there were hundreds of lines of code but more surprised to see that when moving along both x and y axes the x and y coordinates do not remain constant when you would expect them to. That is, from 1,1 to 17,1 the x coordinate flutters all about 1. Should this remain constant? Might these wiggles be adding to my part inaccuracies? The cuts look straight but parts varied (2@ spot on, 2@ + 1/32, 2@ +1/16).

Thanks for any help.

[steve_g]

The old computer programmer adage likely applies here... "garbage in, garbage out".
If the rectangles were not parallel to the axis in question when designed, they will not be parallel when toolpathed. Did you import the vectors from another source? (you said AutoCAD) How were the toolpaths created?

SG

[jstrasser]

Thanks Steve,
The rectangles are all parallel to the axes. The Vectors were imported to partworks 3.5, and toolpaths generated there. The problem is that the rectangles though drawn to equal widths, cut out inconsistently. They are drawn as 15 7/8. 2 were good, 2 at 15 29/32 and 1 at 15 15/16. We have triple checked the AutoCAD file to make sure lines are straight, parallel and consistent. Drawn in AutoCAD 2002LT.

It was all one toolpath, 1/4" downcut, 4 passes.

[steve_g]

J
I'm sure you're right, but can you check one other thing... some times when DXF files are imported they come in as many many segments rather than straight lines or curves. You would like to see just four nodes in a rectangle. This can account for a very busy .sbp file but not necessarily inaccuracies.

SG

[jstrasser]

SG,
I'm sorry I forgot to mention that we did check that when I was talking to Frank at shopbot. 4 nodes each. Here's an image of some of the code with fluctuating coordinates:

[dana_swift]

That is, from 1,1 to 17,1 the x coordinate flutters all about 1.

This is not your problem, but the coordinates 1,1 to 17,1 are a change of 16 inches in X. Y remains at 1 during this cut. That simple command is actually more complex, as the shopbot accelerates from one corner up to a constant cut speed, then decelerates to the next corner. Thats with a single move command. I suspect that is what you meant, but I thought I would point out the flutter for this move would be in Y.

The complexity of the file is due to the increasing Z depth with each pass around the rectangle. That is controlled by your cutting parameters in Partworks when you select the bit properties.

It helps if you include the SBP file so we can look at it and help you understand what the file is telling the router to do. If the router does not cut straight lines given a straight line command, your pinion gears are probably not properly meshed with the rack gears. Especially if this is a new installation. Push on the router/spindle in X and Y (with drivers on) there should be no perceptible motion. If there is, you have something wrong somewhere and you need to tighten up something.

Shopbots are capable of very good precision if they are set up and maintained properly. When someone is new to the machine its not easy to know what proper is, but we all seem to figure it out.

The larger machines are rated at 0.002" of motion repeatability, and my own measurements on several Shopbots show this value is accurate and consistent. The desktop is even more accurate and repeatable. With a LOT of care on my BT32 it is possible to get 0.001". You should expect 0.002" out of the motion system, most materials are not that stable, and bit flex and other issues not Shopbot related will degrade that performance. That loss of accuracy would be the same for any CNC system and depends on experience to compensate for the errors if its that critical.

A typical "garden variety" cut should be 0.005-0.010 of variation depending on the material and bit combination. In any given situation you can improve the edge quality, chatter, and other factors if you want better.

If you are seeing variations greater than 0.010(ish), something is wrong. SB tech support is excellent, they are usually avaialble to help when you need it, and if time is not so critical, the forum members are usually glad to help but at a sightly less predictable reply rate. We are just fellow owners, and occasionally the advise quality is questionable.

When you get your machine dialed in correctly, you will be surprised at how good and reliable it is. How much "flutter around 1" are you getting?

Post the SBP file so one of us can have a peek at what you are doing. That may help us clear up at least some of the confusion and help you concentrate on the correct problem area.

Hope that helps-

D

[dana_swift]

The image of a text file you posted while I was writing is not legible. Post the SBP file (you may have to zip it), then we can look at it.

D

[steve_g]

J
Your code clip came in to small to read... But from what I can see it looks like you have about 14 lines of code between corners. I programmed six rectangles with 4 step downs just for fun... I have one line of code between corners.
SG

[jstrasser]

Thanks guys,
Much appreciated. You will see 8 parts on the sheet. Please let me know what you think. I used '1' as an example but I think you will see what I was calling flutter.

I did find a little play in my Y axis and will adjust presently. The variations were in the Y axis direction. So that's probably the problem. Though I want to know what you think about the many lines of code.

Thanks again

[srwtlc]

J,

Part of the reason for a lot of extra code is that you have tabs on the corners of your parts. If you move all the tabs to the edges, you'll see a bit less code. Along with that, all corners (toe notches included) are a rounded path around the corner which will insert a CG curve in the file at each corner. While this really doesn't matter for ply parts, it can help alleviate corner chipping with solid wood. You can change that by checking "Sharp external corners" when you set up your profile toolpath.

Your tabs are quite high/thick, and they are being slightly cut in the second pass also which also adds to the amount of code. I'd lessen the height them so that they are only in the last pass along with only being on the sides or ends.

[jstrasser]

Thanks all,
Y axis gear slop has been eliminated, and the code comments are helpful. I will run the file again as soon as some scrap material is delivered.

Best,
JS

[steve_g]

The first pass of the first edge required 13 lines of code... I plotted them on a Cartesian plane and got the following results...(see attachment) the seemingly wandering x and y are from stepping around the corner, after that they are dead on parallel and straight but yet more points than I would think necessary.

What post processer are you using?

SG

[jstrasser]

As my PW3.5 is set up at present I have only two post processors from which to select. They are Shopbot TC(Inch)(*.sbp) and Shopbot TC(MM)(*.sbp).

I'm working in inches. Are there additional post processors I should acquire?

[steve_g]

J

I would think you would want an arc inch PP. I don't have a tool changer and so don't know if one is available in the TC series... anyone else?

SG

[adrianm]

The Post Processors have recently been streamlined and as far as I know (I don't use the eCabinet link though) there are now only those two.

[srwtlc]

The TC inch ones do process arcs as arcs.