Hi Oak,
do you really need the machine to be that accurate? Don’t get me wrong, I don’t think 1/32"(~0.8mm) would do. But given that there are still some ridges that need sanding after you cut the board and these ridges are considerably higher than .003" (~0.08mm) you are likely to wipe out a lot of the detail at that tolerance when sanding out these ridges.
regards,
Håvard
Foilednuts, obviously you have not used some of the over the counter surfboard 2D programs. Have you actually used or done any type of programming youself ? Again, algorithms are essentially a mathematical formula designed to CREATE surfaces. The algorithms are so predictable that if shaper A punches in a 6’ x 18” x 11 nose x 13 ½ tail at the same slices as shaper B you will get the same curve. EXACTLY THE SAME OUTLINE CURVE. What is not quite right about that? The algorithms are so predictable you can shift any of the slices .250-.500 and the shapes would still look similar.
I agree my methods aren’t for everyone. If accuracy is not a main concern and you are satisfied to shape a facsimile or a general interpretation of your designs, the 2D method is fine. It will do fine as most make the best of it. But this will come at the cost of generalizing your designs.
From my personal perspective, if you are going to use a CNC to mill the board why not make it as accurate as you possibly can to your finished model? Anything less seems like a compromise as you aren’t utilizing the cnc technology to the fullest capacity. So much potential/capabilities are wasted.
Kawika, do you know if this is an optical method of triangulation? I have some friends in the auto design studio and have used the optical method and mentioned that they have had some deviation issues. And have actually went back to the touch probe method off of an actual study model. I believe for solids it’s still the most consistent and accurated methods for scanning.
Or if it’s some sort of crt type of scan/imaging? Either way, the data collected would still need to be interpreted. I believe this is where the distortion may arise, if any. I am very interested in how it is done and what type of results you get. Please keep me posted on your results.
Haarvard, I dunno?.. do you really need anything? There are only a few things one REALLY NEEDS. For some, good is good enough. Is close, close enough? What’s close? Do you really need that new TYLER? Or will a knock off do? For most a knock off will do. For some, actually appreciate only desire the real deal, and a knock off will not do. It’s depends on where you are.
My methods seem to be appreciated by long time hand-shapers that want their shapes and designs replicated as faithfully and accurate as possible. Most all of these shapers have gone down differing paths to get their shapes so called scanned by differing methods and all have mentioned the degree of accuracy and the noticeable difference.
This is why I feel you need to have an accurate scan from inception. If you are out of spec to begin with once a finish shaper get’s their hands on it and doesn’t have his skills dialed he can easily ruin the shape. The more true and accurate the scan to the actual study model, the closer you will be in the end. The less the finish shaper has to do. Less human distortion.
A good finish shaper should have his finishing techniques down to where this isn’t an issue. Most of the shapers that designed the boards actually finish themselves so it’s still from the source. Other shapers that have hired finish shaper generally go through some type of training and schooled on proper finishing methods and key design areas to concentrate. In this way the finish-shapers know how to maintain the essence of the shape with out injecting their own
Surfding, sorry to hijack this thread. I was curious as to your methods and we end up here. Again, I aplogize.
Yeah… me too… sorry for the hijack.
I’ve played plenty…
don’t know any programing… I wish I knew.
Oak, I imagine your boards are so precise it’s not funny, I appreciate your efforts, I would love to see you doing it, sounds like good fun. I’m even afraid the .08mm bug might bite me.
For now I just want a good reliable board… I can get a true magic board with the over the counter programs, and it’s possible to reproduce it over and over again… that’s really good enough for me at the moment.
As far as scanning… I think one day we’ll be able to say that you can get a true scan using the OC2D (over the counter 2d programs)… Until then we’ll cope with the little deviations.
foilednuts, my process of full 3D scanning is very challenging to me even though I’ve been doing this for some time now. It’s not a simple process to say the least. I’m constantly making changes and trying to improve the process to make a better finish.
If you are local, and in need of a good scan I would love to scan one of your shapes. I would be more than happy to let you see the entire process. It would be great for you to bring by a scanned board and 2 milled plugs. You are welcome to see the process and compare. I think then you would understand the difference.
Foilednuts, obviously you have not used some of the over the counter surfboard 2D programs. Have you actually used or done any type of programming youself ? Again, algorithms are essentially a mathematical formula designed to CREATE surfaces. The algorithms are so predictable that if shaper A punches in a 6’ x 18” x 11 nose x 13 ½ tail at the same slices as shaper B you will get the same curve. EXACTLY THE SAME OUTLINE CURVE. What is not quite right about that? The algorithms are so predictable you can shift any of the slices .250-.500 and the shapes would still look similar.
If you do plug in the same numbers you’d expect the boards to be the same. It’s the same thing as scanning the same board twice, you’d expect it to be the same. If you don’t change a thing, you don’t get changes. Or would you like the program to throw in a little random change here and there? But IMHO it’s besides the point because that’s not how you (should) design a board in the first place.
You don’t design a board by just plugging in some measurements, you design it by modifying and adding to what you get as a starting point when you plug in those measurements. But you’re right, shifting a slice a quarter of an inch is not going to do a whole lot as the difference between the interpolation at that point an the slice is very small unless you have some extreme slices close to each other. But depending on what you want to achieve, moving a slice is probably not what you should do either. You change the curves, move and add control points, add more slices for more control, etc. If shaper A and B just plug in 6’ x 18” x 11 nose x 13 ½ and call it a day, they both did a poor job.
I think the point about what accuracy is good enough is valid. 0.003"(0.08mm) is less than 80 grit size (0.2mm) and I would think that even the best finish shaper are going to wipe out some detail at that scale. Not saying there is anything wrong with accuracy…
Don’t apologize. I find this very interesting…
regards,
Håvard
Oak, it’s all good!
I pushed a button when I made this post. It’s been a fun one and I think it has generate a lot of interesting discussion. I hope there is no miss understanding as to the way my scanner works? It runs off my servo drives and is controlled by my Win CNC program. I scan the top of the board - negative side: bottom side positive - So one side of the board is scanned. This give you a board more accurate than the original, both rails are equal as well as the outline and foil thickness. Rockers always come out perfect. My friends at Win CNC call it reverse engineering. You may not agreed with them. What do they know they went to MIT. I have worked with another company using Laser technology. However my touch probe takes a good hour to scan a board and the laser can do it in less than 10 minutes both sides. It’s not much better only faster. So I’m not sure it’s worth the extra expense in a under capitalized market? As far as CNC technology is concerned you can take it as far as you can afford. 5 axis is the way to go. No finish shaping to speak of. However there is no real money in surfing. So from a financial perspective the CNC technology for the surf industry has a ceiling it’s called lack of capital. Having said that I think that the new CNC machines and the software currently available are Ideal for medium size board companies. Also the smaller board builders can toy with the whole experience by bringing his design to a CNC cutting house and seeing what he comes up with. I have my company set up for helping shapers enjoy the craft big or small it doesn’t matter. As long as the overhead is paid it is for everyone to enjoy. I built my company to enjoy the art of shaping. By hand, computer design, CNC, XPS, EPS, PU, balsa it’s all fun. I just want those who enjoy it to experience it all.
With money you can take technology as far as you like.
Haarvard, In theory one would expect the same but it’s not. The OC2D method is so far out of tolerance from the actual study model that a shaper would have to reconfigure and “bring in” the milled shape considerably to the point that the finish shape has been considerably changed to the milled blank. You can then take that finished shape and inspect w/ digitizer and one can see how much the shape deviates from the actual board scanned. Honestly, the inspection tool w the digitizer would not be necessary. It would be very noticeable just by feel and eye.
I can then take that same exact board and using my 3D method to scan the board, mill it and finsh. Place said board in fixture and compare the two methods and it would be very clear to see the difference.
My offer to foilednuts to have him re-scan the second generation from the first OC2D scans due to bring in the shape so it would be even closer to the finish shape. As in theory the successive scans may help.
It would help you understand to actually do the practical aspects and do your own experiments. Then, I believe, you would see the light. I think you are still in the “box”. Once out of the “box” it will be clear as day. It all sounds good in theory, till you have to use your theory in real practice.
An accurate finish shaper can control the amount of finish material that get’s removed. Most skilled finish shapers remove .050-.060 of material. For most finish shapers and myself I write the program to compensate for this material that is going to be removed, an accurate and consistent glasser and sander will add .050-.060 (4+4 / 4oz). So in REAL practice you can control the amount and degree of finish. As with most shapers that use this method, I have designed and made my own of set of fine tuned tools just to control the finish. I know it’s hard to believe but there are some highly skilled finish shapers.
It’s sooo much more than just knocking down ridges, what kind of ridges and where? How wide? With the 3D method you can control infinately the amount of tool paths, where, how much of step over in certain isolated areas. each individually so, in areas like the rail you can make fine highly detailed cuts and areas that aren’t as critical such as long flat areas you can open up the cuts to make up on cycle time. You like good rail detail? Or the bottom tuck and edge detail? All these are areas are controllable individually. Also an added benefit is the ability to isololate and control differing parts of the boards in differing levels. The actual milled board will still maintain the 3D detail. With the OC2D methods what you see on the screen isn’t necessaily what is cut, so you can spend enormous amounts of time and detail, but the actual cuts do not reflect what is on the screen. It becomes a game of compromise.
An accurate finish shaper can control the amount of finish material that get’s removed. Most skilled finish shapers remove .050-.060 of material. For most finish shapers and myself I write the program to compensate for this material that is going to be removed, an accurate and consistent glasser and sander will add .050-.060 (4+4 / 4oz). So in REAL practice you can control the amount and degree of finish. As with most shapers that use this method, I have designed and made my own of set of fine tuned tools just to control the finish. I know it’s hard to believe but there are some highly skilled finish shapers.
It’s sooo much more than just knocking down ridges, what kind of ridges and where? How wide? With the 3D method you can control infinately the amount of tool paths, where, how much of step over in certain isolated areas. each individually so, in areas like the rail you can make fine highly detailed cuts and areas that aren’t as critical such as long flat areas you can open up the cuts to make up on cycle time. You like good rail detail? Or the bottom tuck and edge detail? All these are areas are controllable individually. Also an added benefit is the ability to isololate and control differing parts of the boards in differing levels. The actual milled board will still maintain the 3D detail. With the OC2D methods what you see on the screen isn’t necessaily what is cut, so you can spend enormous amounts of time and detail, but the actual cuts do not reflect what is on the screen. It becomes a game of compromise.
You missed the point. The point I was making is: does a 0.003" detail over say 1/2" matter when you remover .050" of material when doing the finish shaping? Because you’d hit the high spot harder with the sandpaper than the surrounding area and I’d say it’s a fair chance this detail gets washed out in the process (not that there should be a 0.003" detail like that in the first place, since the shape that was scanned was sanded in the first place…) but for me it’s a guessing point what level of detail is good enough. I’d be happy to beat my hand shaping which isn’t really a challenge.
If what you see on screen in the OC2D as you call these programs, that tells me that there is something wrong either with the program or the machine, but not necessarily with the theory used. You could easily have a scanned mesh being messed up by both program and machine too. You should have WYSIWYG with either methods. As far as I know, the OC2D program allow you to specify the level of detail in different areas of the board with the possible exception of surfcad.
regards,
Håvard
Harvard, the + -.003” accuracy is the machine repeatability. Over an ½” that amount is a small deviation. If you remove .050” of material over the ½” the .003” would get finished down. If you use the OC2D programs to scan, it will not hold these tolerances through the entire scanned shape. Hence your milled board is out of spec using the OC2D. Why? Machine? Program? Or both?
All I can say is for you to just try it and see for yourself. If hand shaping isn’t a challenge for you then I’m afraid you won’t really understand the complete methods of trying to maintain the accuracy in your scans and hand shapes.
Makes me thing though…
Mathematics is an exact science…
An equation defines a curve (bezier…nurbs…spline… whatever) with exactness… so you know exactly where the points of the curve are passing in a 2D plane…
Why would these programs not output a result (coordinates) identical to what you have graphically/numerically told it to do??
the akushaper program tells you really fine increments on the curve at the profile view, that must be accurate… isn’t it???
The machine seems to be an easier thing to blame for inaccuracy… Depending on how the machine was designed/build/maintained… worn out cutting tool… 101 reasons…
Anyway…
Just got a couple of shapes done at an aps3000… took the ridges off… rechecked the measurements… pretty bloody spot on, I was happy to confirm that.
I used a good quality rocker stick… my lufkin stainless steel ruler with 1/32" marks… For the rails I used my plastic vernier thanks… don’t laugh it’ll give you .5mm accuracy okkk.
I did use a blank bent to my exact deck rocker…
Little board looks clean… just beautiful.
Harvard, the + -.003” accuracy is the machine repeatability. Over an ½” that amount is a small deviation. If you remove .050” of material over the ½” the .003” would get finished down. If you use the OC2D programs to scan, it will not hold these tolerances through the entire scanned shape. Hence your milled board is out of spec using the OC2D. Why? Machine? Program? Or both?
All I can say is for you to just try it and see for yourself. If hand shaping isn’t a challenge for you then I’m afraid you won’t really understand the complete methods of trying to maintain the accuracy in your scans and hand shapes.
Hi Oak,
given that a .003" detail over a short distance is getting wiped out with the sandpaper anyway, why not interpolate over this distance? This is typical of the details you cannot pick up with interpolation, small detail over a short distance. The question is really at what point the interpolation deviates too much from the scanned surface. Ideally you’d scan the board and calculate the surfaces in the software and check the deviation, automatically adjust a little bit, add a cross-section at the point where the worst error is or a point in the outline or rocker, then re-iterate until all errors are below a set tolerance. Worst case scenario, you end up with a control point at every scanned point… No worse than a mesh.
Foilednuts,
it’s not as straight forward as it would seems. When you see a 2D bezier in the aps or shape3D or whatever, it’s actually composed of two functions. So the vertical y position is a function of t(time if you will) and the horizontal x is a function of the same t value. So far so good. But say you want the y value for a given x value, that’s not so straight forward. You first need to find the t value for the x value and use this to find the y value. This is ‘complex’ since this is a 3rd degree polynomial and so it’s solved numerically. However, the tolerance of the result can be extremely accurate, so I doubt any errors come from this. Next there is the interpolation between the slices that may not be ideal. There are some cases where it creates some artifacts. But my guess is errors in the software comes from inaccuracies in the compensation for the tool in the cutting path generation. Then the machine can have the wrong setup, the hold down system may allow for the board to be out of position or flex, inefficient cutting tool running at too high speed create too much force on the blank create distortion, etc. Plenty that could go wrong.
Sorry to go technical…
I’ve been holding back from butting in on this conversation, but now that it seems to be winding down, I wanted to add a couple things. First of all, considering the conversation on over all accuracy of the finished shape. Even if you are doing absolutely perfect copies, the glass work is still going to distort the finished product. Take a good look at the glass job on one of the boards you have laying around… you want .003" accuracy? No way, you’d have to put the board back on the mill after the glass work and re-cut it in order to achieve a goal like that. Around my parts we like to keep it to +/-.0625. You can’t expect a sander, making approx $15-25 a board to give you anything else… it’s not like they have a set of digi calipers in their sanding rooms anyway.
When I first started getting wrapped up in the cad/cam/cnc circus, I was infatuated with the beauty of the bezier curves I could draw in my 3-d cad program. One of the first things I started doing was taking a series of points off of my templates, and redrawing them in cad with nice clean curves. I thought I was onto something, was gunna mill me some perfect templates! But once I started cutting them, I realized that what seemed like a big deal zoomed in on my screen, translated to nothing in the real world, when I would lay the traced/milled template on top of the original, they were identical, minus a slight bump here and there that I never noticed... but nothing that I would consider design compromising. With that in mind, I would venture to say that todays hand shaper has been pushed to a higher standard, when you walk through a surfshop and see all those perfectly foiled CI's in the racks- you realized the bar has been raised, cnc or not. The customer expects a certain look, and it's usually the large, popular shops that set the tempo for what the "standard" is. All the large shops are utilizing cnc, so the look that is a product of that tool, is becoming the standard. Clean foils with no lumps or bumps :)
Finally, Oak mentions trying to find the “magic” by high resolution scanning, and reproducing of the shapes. What about the flexural characteristic of the stringer, fiberglass layup, fin placement, fin foils and everything in between? Are you applying your ridged standards to those elements of the design also? I think that unless you are building moulds of your designs and you are pulling the finished parts from those moulds, you will never be achieving the sort of accuracy that you desire. I think the magic is in the complete package. Using conventional methods, it cannot be reproduced perfectly, because of the amount of variables. But, close enough for the average Joe-pro? Sure no problem! But I can do that by hand… 
-Carl
Carl:
Thanks for your common sense.
I use surfers for R&D and Quality Assurance. The Lab is the finish board on the waves under the surfers feet. Proof is in the pudding. Were not making pen caps for Bic! Also after I have a model recorded in my program I print out a rocker profile and have the Foam company give me a Blank with that rocker. That is such a great feature. Even if you are hand shaping to have a blank with the correct rocker keeps you a head of the game. Start well finish well.
This blog has gone on for a long time and I think OAK made it fun and thank him for it.
Sorry fellas, I haven’t had the time to respond. Carl, you need to go back and re read my posts. Or you don’t have an true understanding of my methods. Making templates in 2d is one thing and this is a great way to design some nice full templates, but don’t confuse this with 3D scanning.
I would love to camp out on this post but this has taken way too much of my time.
Surfding, I like to think the surfboard as much more sophisticated than Bic pen caps. As from the sound of your post , maybe you should be surfing on pen caps instead. They maybe more accurate. J/K! Could make the difference between the QS and CT. Who knows?
Please don’t hate the messenger, I have done the homework and done the experiments. And stand by my results. If you don’t believe me take me up on my offer.
Surfding let’s make this REAL FUN! I will step up to the plate and will purchase all the materials and pay for all the glassing etc.to do the experiments to compare. Let’s do a friendly comparison. If you added all the deviations from the scanned model it would be in inches in just a few points. If you added all the deviations in the entire board in maybe in feets not inches. With no modifications to the orginal program once scaled who knows how far out it will be.
At the end of the comparison whom ever method has the most deviations pays for the the accrued costs including the scanning time. The proof IS the pudding.
Put up or shut up. Heheh… j/k.
Or anyone else that would like to get in the experiment speak up.
Let’s make it fun. Any takers?
Oak,
Sorry, I was using a 2D analogy to simplify my point. I assumed you were building point clouds. If that is true, any point in that cloud would be represented by 3 dimensions, simply 2-d with a “z” dimension added to the mix. That aside, I was making 2 points, the first being that the algorithms that drive the curves in CAD are “close enough” to be non-design compromising, in my opinion. If anything, they are usually cleaner anyway, so why not gravitate towards perfection of design? I personally love clean curves.
My other point was that you are concerned with achieving .003" accuracy, when in a practical reality, where laminators and sanders make a mere pittance, the shape is going to be distorted during the glassing process. And even ignoring that fact, the characteristics of the hand layup, stringer wood, fin placement, edge tuning are all going to affect the finished product. Once again, simply put, to achieve a .003" design tolerance with any kind of realistic production goals, you would have to resort to a moulding process producing finished parts. -Carl
Carl, You clearly don’t quite understand my methods and technique like Haarvard you are thinking in the “box”. Please re read by posts, no clouds used, no algorithms, no cross sections slices used. What is perfection? Outlines a program or algorithms renders for you? If that’s your bag then that’s perfection to YOURSELF. For, other shapers and myself is an accurate replication of my actual board that I actually shaped. Not something a computer had drawn or rendered or averaged out data.
The .003 is machine repeatability, this is for a decent machine. My personal machine is set-up to SCAN IN 3D to the 1/10.000 (.0000 that’s four decimals places) No where do I state that my boards are in this tolerance as this is the human element, but I regularly hold ± .005 -.010. constant. But the actual surface is 3D scanned and is digitized to 1/10000th.
The glassers that I use are very consistent with their laminations and the sanders are very consistent and hold regular tolerances. This is inherent in their methods and technique if you can’t determine what your glass jobs will be then you need to find another glasser, there are many thaT have it down, consistently.
Due to the human factor this is why it’s more important that the shape be in tolerance from inception of manufactoring. If the shape is out then it’s all down hill from there.
Contrary to your statements about moulding vs cnc the milling process is much more accurate to holding tolerances than moulding process with the materials being used.
Be aware, NOT ALL SCANS ARE THE SAME.
Carl, You clearly don’t quite understand my methods and technique like Haarvard you are thinking in the “box”. Please re read by posts, no clouds used, no algorithms, no cross sections slices used. What is perfection? Outlines a program or algorithms renders for you? If that’s your bag then that’s perfection to YOURSELF. For, other shapers and myself is an accurate replication of my actual board that I actually shaped. Not something a computer had drawn or rendered or averaged out data.
I have a hard time grasping this. Are you saying that a hand-shaped board is better because the curves are not as smooth as the algorithms and therefore you want you machine to reproduce the given imperfect shape? Kind of like why live music is better than recorded and recorded music is better than synthesized music? In music it’s those little imperfections, those tones that missed the beat by a fraction, the tone that changed by the way the guitar was moved, the human factor that makes the music what it is. Maybe that’s where the soul of surfboards are, in those little human imperfections?
If you are not using a clouds(which is a fancy word for a set of scanned points), no algorithms, how can you at all define a board? I find it really hard to believe that you don’t use scanned points(‘clouds’) nor interpolation. Given you scan to the tolerance of the machine (let’s say 1mm grid just for good measure) and use 32bit 3D floating point coordinates, that would make the files ~64MB for a 6’ board using a rough estimate of the area of the board. Somehow I doubt this especially if you have been doing it like this for while, are able to scale each axis independently and edit and analyze it. That amount of data would be slow to edit to say the least, double so if you’ve been doing it that way for 5-10 years. There would be methods to get around this though. It would be a little easier if just mass produce copy after copy without doing any form of editing though. I can’t help thinking you either have some really advanced custom software, or use off the shelf CAD software. But still, you’d be using linear interpolation between these 1mm grid coordinates, since the machine cannot magically go from on of the position to the next without passing through something in-between.