Attempting to use CW as cross-section 3D printer, but plagued by X-axis issues

[IHaveThatPower]

Hello, CarveWright users!

I am attempting (and have been for many months now) to use my CW as a sort of 3D-printer, using 3D modeling software to create a virtual representation of the thing I want to make, then taking cross-sectional slices of it, converting those into grayscale height maps, and importing those as patterns into the Designer software. I've seen others use this method with a CarveWright to great effect. Unfortunately, I've continually run into severe-enough inaccuracy along the X axis that I have, to date, created no usable cross-sections. The X axis dimension of each cross-section is always too short and never by the same amount (i.e. I can't just overscale something to compensate for this). I have done a great deal of searching through these forums' archives, the machine manuals and guides, YouTube videos, and so on to attempt to self-diagnose the issue, but after today's unsuccessful carve (four hours with great hope that I had finally resolved the issue, only to see a 7mm difference between expected and actual X-axis dimensions), I have to admit that I simply am out of ideas.

I have tried the following:

• Checking, re-checking, triple-checking, and quadruple-checking my source measurements. They're right in my 3D software, they're right in my image editing software, they're right in Designer. It definitely ain't the source.
• Leveling the head. I initially also had issues with varying depth along the Y axis, but leveling the head resolved this (along with bullet #4).
• Cleaning and re-greasing everything. I also measured low head pressure at first. After cleaning and re-greasing, this seems to have gone away, but now I frequently see high (as in > 115 pounds!) head pressure if I allow the "clutch" to click more than two or three times. If I just click it a few times, I measure ~84 pounds.
• Removing, cleaning, and re-seating the sandpaper belts. They were, initially, partially rolled under on one side.
• Changing the type of board. I started out using 1/2" MDF, but noted that MDF tended to "shred" when it got too thin. I switched instead to 3/4" (really, 11/16") plywood, which did away with the shredding problem.
• Manually calibrating the X axis. I bought a 36" board specifically because it claimed at 36" length that I verified with two separate tape measures. The CW measured its length as 36.27925" (average of four measurements, each of which was slightly different). Using the instructions for manual X axis calibration, I adjusted the offset from 890.213 to 899.521, which resulted in a measurement of 35.958", which I was happy with (0.042" = 1.07mm, which seemed pretty accurate).
  
  • I subsequently did a test carve that seemed to come out more-or-less correct with ~2mm total inaccuracy, but on all axes. Since the 1/16" carving bit itself is nearly 2mm across, so I figured this was probably simply within expected tolerance.
  • Today's carve came on the heels of that test carve and resulted in the aforementioned 7mm variation (specifically, a piece that should be 19.66cm wide came out at 19cm even; that's far too much slop for what I'm trying to do).
  • I have since reset the offset to default (I haven't carved since doing this, though). In all of these tests
  • All of these tests and carves were done with a board 14.5" wide.

I've attempted to carve some unpleasantly large number of these cross-sections at this point, using different configurations on the board. All of them always come out wrong along the X axis (and only the X axis) and never in ways that are consistent enough to simply "smooth over" after the fact.

I've also variously encountered the following errors along the way:

• X axis stalls. These were resolved after the thorough machine cleaning.
• Tracking roller errors. These seem to happen only on longer boards (the 36" one caused it a couple of times) and I hit a spate of them at one point until I realized that I simply hadn't firmly re-tightened the screws attaching the tracking roller assembly to the guide plate. Derp.
• Back roller stuck. No idea what's causing these. They've started happening in the last day or so when attempting to measure short (~15") boards and they only occur once the board has left the back roller. Is there a minimum board length?
• Power fluctuation issues. Resolved by changing outlets.

I am somewhat suspicious that there is a problem in the X drive motor. I ran the X-axis "user test" five times in a row and each time got different values (which, unfortunately, I have not found any information on interpreting, short of "send this in to the techs if they ask for it," which is not terribly enlightening).

A=125.37 M=181.63
A=124.34 M=178.41
A=123.66 M=175.96
A=123.22 M=178.23
A=122.93 M=176.16

My assumption is that these numbers should be much closer together than they are and that the disparity between them is indicative of at least part of the problem.

I inherited this machine from a friend moving cross-country. As an inherited device, I don't have the option to simply "send it in" for fixing; I'm kind of on my own (along with whatever kind help the folks here on the forums can offer) when it comes to repairs. And, as a cherry on top of that cake, I'm not really in a position to lay out a bunch of cash for replacement components just now, either. I'm hoping there's some kind of home repair or calibration or cleaning or cheap hardware store replacement part that will be the magic bullet to all my woes here.

Additional miscellaneous info that may be relevant:

• Machine type: A (yep, old)
• Power-On time: 265 hours
• Servo time: 192 hours
• Cut time: 185 hours
• Firmware: Rev: 1.187.10870, Build: 0417:090305, Secondary Rev: 7

Any advice anyone could offer would be very much appreciated!

[bergerud]

Could you post a picture of a carve or carve setup which came out too short?

[fwharris]

Like Dan said, pictures are worth a 1000 words to give us a better idea of whats going on.

I did notice that you said
"two separate tape measures, no less! -- measured 36.27925". Using the instructions for manual X axis calibration, I adjusted the offset from 890.213 to 899.521, which resulted in a measurement of 35.958", which I was happy with."

I would try to calibrate the X axis to get closer than 0.321" with a 36.278" board. Also to get a better reading make sure you run it 2 to 3 times to get a better average as well as looking for any possible tracking issues, use a wider board (do not know your width of the board you used).

Do not get to out of sorts because you have an old A machine as there are plenty still in service. You should try to verify what upgrades have been done (if any) to better determine what you might be looking at needing. One of the first upgrades was the A907 Z Pack. This includes a new Z motor, circuit boards, cables leads and the 14 pin FFC cable. To verify if you do or do not have it, the wires from the black cover (back left side corner behind the y motor) plug into the top of the circuit card mounted on the back of the Y rail with the upgrade. With out he upgrade they plug into the bottom of the card.

http://support.carvewright.com/a907-...nnector-board/

Other items would include new cut motor brushes and Y gear box bearings bases on your number of hours cut time and the CarveTight chuck if it still has the original Quick Change chuck.

[IHaveThatPower]

Could you post a picture of a carve or carve setup which came out too short?

Like Dan said, pictures are worth a 1000 words to give us a better idea of whats going on.

I will try to post some when I get home from work this evening.

I did notice that you said
"two separate tape measures, no less! -- measured 36.27925". Using the instructions for manual X axis calibration, I adjusted the offset from 890.213 to 899.521, which resulted in a measurement of 35.958", which I was happy with."

I would try to calibrate the X axis to get closer than 0.321" with a 36.278" board. Also to get a better reading make sure you run it 2 to 3 times to get a better average as well as looking for any possible tracking issues, use a wider board (do not know your width of the board you used).

I phrased this poorly (have just now edited the post to clarify). The board claims a length of 36", and it is that length that I verified with independent measurement. After having verified exactly 36", I then fed it into the CW to measure, and the CW measured 36.27925" (that's an average of four measurement attempts, hence the extra digits). The difference between the real 36" and the CW 36.27925" is what I used to calibrate the offsets, after which the CW provided an average measurement of 35.958", which is within 0.042" rather than 0.321". I realize my phrasing didn't make this at all clear.

In just about all of these cases, the board width has been 14.5", so nearly the full width of the belt.

Do not get to out of sorts because you have an old A machine as there are plenty still in service. You should try to verify what upgrades have been done (if any) to better determine what you might be looking at needing. One of the first upgrades was the A907 Z Pack. This includes a new Z motor, circuit boards, cables leads and the 14 pin FFC cable. To verify if you do or do not have it, the wires from the black cover (back left side corner behind the y motor) plug into the top of the circuit card mounted on the back of the Y rail with the upgrade. With out he upgrade they plug into the bottom of the card.

http://support.carvewright.com/a907-...nnector-board/

Other items would include new cut motor brushes and Y gear box bearings bases on your number of hours cut time and the CarveTight chuck if it still has the original Quick Change chuck.

I'll take a look tonight and post pictures of these areas as well. Thanks!

[Digitalwoodshop]

Not a clue?

AL

[eelamb]

1)When you place the slices on the board, are you doing them on one board in designer, or on multiple boards, where designer has to measure each board?
The multiple board method does introduce some errors, such as board slippage in measurement. You need to use best methods to help minimize the errors, like tape on bottom for brass roller to track on, and flat square boards, are two best method.

2) you say you are using grayscale images, here again this can give some errors, as edges are difficult to read, thus smaller scale of the slice. Do you have the STL importer? I feel it is a better method of importing of 3d models, plus you can import the 3d model and slice it inside designer, giving you a higher accuracy of each slice. I think most if not all use the STL importer to import their 3d models and slice them.

3) calibration is a necessity for what you are doing, the more accurate you can get the x-axis the better your results will be, and the amount of error in reading should be minimized.

My feeling is item #2 is your problem.

[mtylerfl]

As long as your slices are correct (using Pierced in the Designer software layout), AND when you are placing the slices onto your layout you UNCHECK the option to "Scale to Fit Board" (which is ON by default...turn it OFF by unchecking), AND you are positively NOT allowing any Scaling when at the machine, your slices should fit together perfectly (regardless whether the X <board length> measurement is off by a small amount).

There are several other possible variables we cannot know about.

Just a single example: We don't know what software you are using to create your slices...I would recommend using the CarveWright STL Importer add-on. It will create perfect slices for tooling with your machine (also allowing you to save the slices with automatic "Pierced" turned on to be sure the slices will fit properly after machining). Using greyscales for importing may not be the best choice for assuring trouble-free results. (I know you are trying your best to do this on the cheap, but you could sure make your life a lot easier using tools designed for the job.)

Now, you may indeed have some kind of X-axis foolery going on there that is part of the problem (although I strongly suspect that's not the only issue). Both my machines are A machines and perform flawlessly, so don't think because your machine is one of the early models that you can't achieve superior results - you certainly can. You may need to send the machine in after you have checked everything suggested thus far and not gotten it up to normal performance standards. (eek! you might have to spend some money on that free machine!)

[IHaveThatPower]

Here are the requested images:

• Cut motor circuit board (a bit dusty!). I'm not sure which wires I'm supposed to evaluate to determine whether or not an upgrade has been done, though.
• Classic chuck, not upgraded to "quick chuck" as far as I can tell
• Brass roller, not specifically requested, but feels like it's ultimately going to be the culprit somehow
• Sample carve in Designer (with Pierced checked, which it was not when I did the carve), showing the pattern scaling and then comparing it with images from measuring the final piece. A cookie to anyone who guesses what it is.

1)When you place the slices on the board, are you doing them on one board in designer, or on multiple boards, where designer has to measure each board?

To date, I have placed multiple slices (not necessarily related slices, though) on a single board, to the maximum carvable extents of the board and the machine, so as to minimize the number of boards I have to purchase. That said, I have not done a manual X calibration on each board before each run, so perhaps that's something to consider...?

The multiple board method does introduce some errors, such as board slippage in measurement. You need to use best methods to help minimize the errors, like tape on bottom for brass roller to track on, and flat square boards, are two best method.

Yep, I always put masking tape on the roller-side of every board. I also try to make sure every board is as square as possible and have increased the quality of boards I purchase since my initial attempts for just this reason (much to my wallet's dismay).

2) you say you are using grayscale images, here again this can give some errors, as edges are difficult to read, thus smaller scale of the slice. Do you have the STL importer? I feel it is a better method of importing of 3d models, plus you can import the 3d model and slice it inside designer, giving you a higher accuracy of each slice. I think most if not all use the STL importer to import their 3d models and slice them.

I do not have the STL importer. For $200, I will probably never have the STL importer.

3) calibration is a necessity for what you are doing, the more accurate you can get the x-axis the better your results will be, and the amount of error in reading should be minimized.

This was my hope. I thought calibration was going to be my "magic bullet" in finally resolving my X-axis issues. It did not seem to be, based on this most recent carve.

As long as your slices are correct (using Pierced in the Designer software layout)

I confess, I did not know about the Pierced option until you mentioned it! I have long wondered why my carves were leaving ~0.5mm of board thickness rather than carving all the way through. Now I know. Unfortunately, while this could fully explain the ~1-2mm slop I see in the Y axis, I don't think it's enough to explain the ~6+ mm slop in the X.

AND when you are placing the slices onto your layout you UNCHECK the option to "Scale to Fit Board" (which is ON by default...turn it OFF by unchecking)

Always.

AND you are positively NOT allowing any Scaling when at the machine

Not to my knowledge. I always say "No" whenever prompted to scale to board, I always stay under rollers (+7" of board on either side of my pattern), I always tell the machine to center the pattern on the board...as far as I know, these are all the correct choices.

your slices should fit together perfectly (regardless whether the X <board length> measurement is off by a small amount).

That'd be nice.

Just a single example: We don't know what software you are using to create your slices...

The original model is created in Blender, as are the slices. The slices are then given a black-to-white linear (NOT gamma-corrected) gradient along their "view" axis and rendered in linear (NOT gamma-corrected) color space. From there, they're brought into Photoshop for proper pixel ratio setting (scale I use is 100px = 1cm), allowing Photoshop's measurements to correctly correspond to the Blender (and expected real-world) measurements, which I can then use as a touchpoint when importing the patterns into Designer.

I realized the issue with the non-linear colorspace early on, when angles that should have been relatively flat were coming out highly curved. Switching to linear space completely resolved this particular issue. In retrospect, it was obvious: I was not rendering for display, I wanted exact values to represent height!

I would recommend using the CarveWright STL Importer add-on. It will create perfect slices for tooling with your machine (also allowing you to save the slices with automatic "Pierced" turned on to be sure the slices will fit properly after machining).

See above. This is a non-starter for me.

Using greyscales for importing may not be the best choice for assuring trouble-free results. (I know you are trying your best to do this on the cheap, but you could sure make your life a lot easier using tools designed for the job.)

If I were seeing other issues in the patterns, I would agree at once. The patterns, other than the shortened X dimension, however, are coming out exactly (off by ~1mm along each side of the Y axis) as I expect, which leads me to believe it's not a methodological problem, but a mechanical once of some kind. I realize I could probably increase my precision with the STL method (for example), but that would be an issue if I were seeing slop everywhere, instead of along a single axis.

Now, you may indeed have some kind of X-axis foolery going on there that is part of the problem (although I strongly suspect that's not the only issue). Both my machines are A machines and perform flawlessly, so don't think because your machine is one of the early models that you can't achieve superior results - you certainly can.

That's good to hear!

You may need to send the machine in after you have checked everything suggested thus far and not gotten it up to normal performance standards.

I suspect the shipping alone for something so large and heavy is out of the question, let alone service fees. Ultimately, though, this is a machine driven by (relatively) simple electronics. It should be diagnosable and repairable by a reasonable savvy owner. It's just a matter of knowing what to test and how to test it. As mentioned in my original post, I'm sure I could interpret some meaning from the X axis user tests if information relating to those test was available somewhere.

(eek! you might have to spend some money on that free machine!)

My "inheriting" the machine was not free. I did pay aforementioned friend for it (and not a small amount, either!).

[mtylerfl]

I think you are doing a very good job at asking questions and methodically checking everything you can think of, etc. I hope you'll be able to get the machine and your sliced projects working to your satisfaction, without having to ship the machine to CarveWright for a thorough going over. Lots of help available here (as you can tell), so your chances are better than good.

[bergerud]

I think you are biting off more than anyone could chew. You need to do some simple cut outs to trouble shoot. Forget the Blender gray scale stuff until you can cut out a rectangle the right size. Are, for example, you doing full depth cutouts? Maybe you are asking too much from the x drive and something is slipping.