good job
On steel surfaces I routinely sanded long beams to about 0.02 mm flatness/straightness to mount linear guides in 25 / 35 mm sizes.
A 50 mm wide steel flat, sanded with a *wide* (Festool 105) powerful belt sander and 40 grit paper will easily remove material in a controlled manner, to about 0.01 mm control in depth.
The edges of the 50 mm flat get rounded, but the center part is flattish to better than linear guide specs of about 0.03 mm.
No scraping needed, nor shimming.
Using a ground machine-tool builders sharp straightedge, and a backlight, it is easy to see low/high spots on the surface.
Then mark the steel with a sharpie, one line for small removal, more parallel lines for more removal.
Then sand until the marks are faint, check, adjust.
It is fast, and very accurate, you get excellent control.
It is also noisy, dusty, unpleasant, hard work.
The wide belt keeps the sander from digging in.
If gross material removal is needed, a Hilti/similar small angle grinder and 45 grit flap wheel will remove metal fast, in a controlled manner.
Then belt sand to finish.
The grinder and flap wheel can cause tilts, so use judgement.
good job
Hi Peter. The beam is built and the proof will be in the pudding. Regarding mass and moving fast (accelerating), this is one of the primary design considerations of course. I fully understand the relationship between mass and acceleration. That is why I built a composite gantry beam. Conversely, a component that does not move (the main frames) are made of steel filled with concrete and weigh a great deal. And the floor slab is 200mm of concrete. In any case all the design values are calculated long ago for the purpose of specifying equipment. Aluminium was not considered for the gantry beyond initial stages due to it's thermal expansion coef. The presence of mass in the form of a small quantity of core material is hardly having a major affect on overall mass, especially considering it relative to a steel design with similar stiffness (and probable residual stresses, have to heat relieve a 5.0m component somewhere). Steel and concrete have similar thermal expansion coefs. The steel laminated into the structure to receive the fasteners adds far more mass than the core material. And in any case, as I have said earlier in the thread, this beam is some what experimental. Hopefully I will be installing the gantry in the next short period. Your comments are all very good and I do hope that you won't take this the wrong way, but I do know all that stuff already, and with composites there is always many many ways to skin the cat. At the moment I am fighting with the mundane and time consuming issue of preparing flat and straight machine beds, so the higher level stuff is not at the front of my mind. J
Last edited by jono5axe; 03-16-2019 at 09:04 PM.
Jonathon Clarke
www.solpont.com
Awesome feedback, thanks very much. I had already considered this, and back myself to be able to get precision results with an angle grinder, sander, scraper, etc, but goddam it, it is a lot of work...
I also need flatness across the bed, at the edges as well.
When you quote those figures "to about 0.01 mm control in depth" , over what distance?
When you quote "part is flattish to better than linear guide specs of about 0.03 mm", what is this linear guide spec? over what distance? is that the stated rail straightness, or do you have something that specifies a bed flatness requirement?
Jono
Jonathon Clarke
www.solpont.com
One of advantages of epoxy is that you do get a coplanar surface for your rails quickly and easily.But it may introduce unacceptable flexion with what would seem to be heavy gantry and z.
I rarely wade into this sort of thing because of the flak I get in forums but I find comments like I can hand grind something with 40 grit to 10 micron flatness to be unacceptably optimistic. I regularly have to specify turned and milled product to tight tolerances and even using a $500k dollar machine it's tough getting to 0.010mm flatness over a reasonable distance. To achieve this sort of thing you have to grind or use a multi million dollar mill that's good for 0.001mm. The way to achieve this sort of thing by hand is laborious but doable. Its the 3 surface trick. You get three long flat objects (say a suitable aluminium extrusion). . You then use a contrast like engineers blue on the required surface. You then wipe the surface with board 1 until you remove the blue. You reblue and use board 2 then you use board 3. You work down the grits until you get to where you want. Along side of this you blue board 2 and wipe with board 1. You then do same with board 2 & 3 and rotate the boards. In this way the boards get flatter and the surface gets flatter. Once your happy with the surface your done. You can probably get away with 2 boards. This is how this sort of surface has been done for centuries by stonemasons and then engineers before machinery existed to do this sort of thing. Plus you use a scraper for the high spots so you don't dip the low spots. . Do the best you can with the dollars you have. Keep making. Peter
Last edited by peteeng; 03-17-2019 at 06:39 AM.
That's a big number .02 for that type of linear Bearing just figure how much clearance is in the Bearing and that is the max tolerance you can have on the flatness of the mounting surface, A preloaded Linear Bearing you don't have anything to play with for rail flatness tolerance, using epoxy it is not an ideal mounting surface as it will deform when you torque down at each Rail mounting point, any mounting tolerance you had will be lost just with the deforming of the surface
Mactec54
Don't worry about the flak, mate. At least not on this thread. I am interested.
The reason for trying less conventional methods, i.e. not machining these beds, is due to their size and the logistics of it, and the fact that I generally like to look at the less conventional / innovative ideas. Possibly a mistake in this case. But possibly not, it will work out one way or another.
I am quite familiar with how to manually work a surface with bearing blue and a scraper, and also with various other tools. And over the years I have made very many composite patterns and molds manually (not CNC) and am very used to working with surfaces manually, not as exact as what we are talking for machine beds, but very fine surfaces nonetheless. I also researched this area earlier in the design process and from that research I determined that, not so much for linear bearing beds but certainly for sliding beds and larger tables, that hand scraping was the way. It is all depending on the skill of the operator, of course. Experience. Hand & eye, care & patience, discipline. Good measuring tools.
From what I have done myself I feel that I can work to flatness of 0.03 or 0.02 mm over 500mm with a combination of tools and methods. But after that then my measuring tools errors probably come into play. And the time required.
But my original post on this matter was to ask the question about what tolerances I should apply to bed flatness for medium precision/preload linear rails (face-to-face) (20mm-45mm rails)
Jonathon Clarke
www.solpont.com
Hi Jono - What preload are you using in this machine? For an accurate router or mill machine bearing manufactures recommend their heavy preload spec. http://www.pmi-amt.com/en/data/Catol...EN_GW_MD08.pdf download the PMI design catalogue it has most of this data in it including parallelism, level delta, etc. Your brand of bearing should have same info. They are made to an international std so should all be the same. Cheers Peter
I'm enjoying this build thread, looking forward to seeing the next update! I used HIWIN brand rails on my machine and they publish the specs in their catalog. See attached image. This should give you a rough idea, but the manufacturer of your rails should have documentation for you.
Jeremy
http://www.diycncdesign.com/
Thanks for that, a number of people have referred me to that type of information, but as previously posted - I have that information already and have looked at lots of other manufacturer technical guides and while they do provide parallelism and relative height (level) tolerances of one rail to another, they do not provide flatness/straightness information. For example, what deviation is allowable in a rail of a given size up or down over say 300mm. If, say, a 35mm rail is all good to toque down with 0.003mm low point between two level points 1.0m apart? This is flatsness of the mounting surface, and is not specified in the Hiwin installation guide. It will obviously affect operational life of the bearing (if the bearing moves at all). The Hiwin catalog only states "mount on a flat and straight surface".
I have found some references to allowable deviations for 'out of alignment of blocks' on the same rail (ie blocks bolted to a carriage out of alignment) but this is only in a horizontal plane, and in any case is somewhat a different situation to 'mounting surface flatness and straightness'.
I suppose another way to consider this, in the absence of other information, is to ask the question "what is a typical flatness and straightness achieved by a larger milling machine cutting operation" and presume that that is acceptable to work with.
The flatness and straightness of a surface should be expressed as a deviation per unit distance, for example +/- 0.002mm per 300mm [0.002/300mm] or similar.
Jonathon Clarke
www.solpont.com
Hello Jono - The bearing companies specify the running parallelism of their bearings and rails. This is the tolerance of the height of the bearing land to the bottom of the rail along the rail and their width. They specify what tolerance the bearing geometry is expected to achieve over various lengths, see the PMI manual I sent for instance. They also infer that the tolerance of the bed is up to the user as they do not know what tolerance machine the user is building. So you have to figure what tolerance bed you need to achieve to get the operational tolerance you want. The bearings will function with a relatively poor surface flatness.
The process is as follows:
1) Decide the vertical and horizontal tolerance you need in the machine, these are usually the same.
2) From this pick the bearing accuracy grade you need, also consider rail length
3) from this specify the bed tolerance needed to remain within the bearing tolerance and
4) this should then achieve the motion tolerance you need
Cheers Peter
Last edited by peteeng; 04-25-2019 at 05:52 PM.
Well that is not really what Jono5 was looking for that information is very basic and he already has that
The machined surface is what is important and how parallel they are to one another, the machined surface is all dependent on the bearing preload and the precision Grade of the Bearing, the mounting surface has to be as accurate as the Bearing Grade Preload
So the Bearing Grade and preload determines the mounting surface accuracy, so if Jono5 looks up his Bearing accuracy and Preload they will have a tolerance and the mounting surface has to be within that tolerance or he will have Bearings that will have a short life
Mactec54
They hand scrap the mounting machined surfaces on any good machining center they don't work to any tolerance but perfect the surface to suit the linear Bearing tolerance Here is a good article on Linear rails
Common errors in linear guide way selection and installation
Common mindsets
Two of the most common parameters of rails that engineers tend to over specify are size and tolerance class. Many engineers will call for a bearing that can handle a load larger than the application needs. Part of this oversizing is due to comfort levels developed from years of working with other linear motion systems. When it comes to handling a 3,000- lb load, for example, engineers used to specify a round rail bearing with at least a 2-in. diameter shaft, or larger, a pillow block, and support rail. Such a system would be 5-in. high. To go from this system to a profile-rail type design that measures 1½-in. high can be discomforting — even though both systems are rated to handle the same load.
It is not uncommon to see profile rail systems with the capacity to handle 10,000 lbs handling 500 to 1,000 lb. To some engineers, “it just looks right.”
A few engineers will admit that they over specify because “it’s cheap insurance.” The logic is that the system will last longer and be less likely to be the cause of a system shutdown. However, engineers should follow the selection formula, the life-load curves and so on, found in the manufacturer’s catalog to choose the correct size bearing. Over specifying the size can cost up to 50% more than selecting the right size bearing and it won’t necessarily provide “insurance” against downtime. Regardless of a linear bearing’s capacity, if a rail system is not properly installed it can still fail before its rated life
Tolerance
As with radial bearings, rails come in tolerance classes: “N”ormal, “H”igh, “P”recision, “S”uper “P”recision, and “U”ltra “P”recision. Normal grade assemblies, for example, will have height tolerances (H) of 60.004 in. and running parallelism of 0.0017 in. in 10 ft, The ultra precision grade will have height tolerances of 60.0002 in. and a running parallelism of 0.0003 in. in 10 ft
Engineers tend to over specify tolerance class too, but manufacturers’ debate whether this is a problem. For machine- tool applications, no one debates the need for the upper precision grades of rails. The debate begins for other applications. One side says it’s better to always specify a precision grade two or three times higher than what is needed by the application, because you can save cost and time for preparing the mounting. This side claims the mounting need not be as precisely ground in average applications that use one of the higher precision rails
The other side says you’ll save money by choosing the proper size rail (not oversize) and quality level for the right application and installing it properly. They say the mounting should be carefully prepared.
Several factors effect the debate:
The more accurate the product, the better the load sharing on all rolling elements. Therefore, theoretically, the actual life of the system will closely match the calculated life.
• A more precise system can minimize a stack up of tolerances.
• If conditions aren’t near perfect for linear-guide way rails, then the rails are subject to a decrease in life. The roller or ball goes through the hole created by the groove in the carriage and the groove in the rail, Anything that causes misalignment, may cause degradation because it forces the ball or roller into an elliptical rather than spherical shape. Round rails, by contrast, are more forgiving.
• Right out of the box, rails may appear to have a bow. (See the box, “Straight and narrow.”) This bow disappears after proper installation, yet engineers may specify a higher tolerance in order to avoid that initial appearance.
Above precision grade, if the application is not in a climate controlled environment, any benefit of high tolerance is lost. Summer temperatures in a shop easily reach 90 F. Winter temperatures may drop to 50 F. Any precision grade rail will flex in either condition because it is ground for temperatures that do not vary from 72 F. In the ultra precision grades, even body heat can cause rails to flex. So why buy precision that can’t be used?
Installation affects a rails features
Today’s rails require different methods of installation than previous linear motion systems. Round rail linear bearings used to be simply bolted into place. Their self-aligning features and ability to “roll” into co-planer alignment allowed some imperfections in installation.
As mentioned earlier, linear guide way rails are less forgiving of installation conditions. Engineers cannot similarly “bolt on” a precision rail and expect to receive the benefits of precision tolerances. If the machined quality of the rails, or tolerance, is important in an application, then engineers must pay attention to rail mounting fixtures.
A radial bearing will prematurely fail if put on a poorly machined shaft. Similarly, a linear profile rail will also fail if put on a poorly prepared surface. The reference edges that will hold the rails should be at least as precisely machined as the rails. Otherwise, both the life of the bearings and linear accuracies will suffer. For example, instead of carrying a rated load, the system will now be carrying an induced load, which may result in premature failure. Other problems may be brinneling or spalling on parts of the rail. It is not uncommon for customers to call the manufacturer complaining that the rails were not hardened sufficiently, when the problem was not poor heat treatment but improper installation.
After installation, rails and carriages will follow and reflect any deflection of the installed rail. Nothing evens out the bumps and twists of an improperly prepared surface or improperly installed system. Although, excessive variances on the mounting surface may be milled, scraped, or hand stoned into specification.
To properly install, first, follow the manufacturer’s recommendations for selecting the correct size and tolerance in a system, even if it doesn’t look right. Then, for moderate precision systems, use a reference edge. Push one rail up to that reference edge and let the other rail “float” into position. The procedure is not unlike locking and floating radial bearings. It will be the reference edge that determines the straightness accuracy. For moderate precision, a machinist straight edge and dial indicators to check and verify run-out are sufficient, as long as you stay within recommended deviations.
If the second rail is not floated into position, or set into a machined channel of its own, it should not exceed 0.0005 to 0.001 in. out of parallelism with the first rail. (These dimensions are a general rule of thumb. They may differ depending on the size of the bearing, whether the bearing was preloaded, and which manufacturer made the rail.) Care must also be taken in the other plane. Rails should be held coplanar, with no more than 0.004 in./ft of spread between them. Keep in mind that as the preload value increases, the permissible deviation decreases.
Mactec54
Hi Martec - Thanks for the info that's all very good.
Martec "The machined surface is what is important and how parallel they are to one another, the machined surface is all dependent on the bearing preload and the precision Grade of the Bearing, the mounting surface has to be as accurate as the Bearing Grade Preload" I said this in point 2) "pick the bearing accuracy grade you need, also consider rail length"
Martec "So the Bearing Grade and preload determines the mounting surface accuracy, so if Jono5 looks up his Bearing accuracy and Preload they will have a tolerance and the mounting surface has to be within that tolerance or he will have Bearings that will have a short life" I said that in point 3) "from this specify the bed tolerance needed to remain within the bearing tolerance"
Se we agree...
Cheers Peter S
That's what I posted the mounting surface is the most important part of any build that is using linear Bearings, or just bolting on component's that need a machined surface, so the better it can be Machined Scraped Ground Honed lapped the happier the linear Bearing will be
The surface tolerance needs to be over the whole length of the linear rails mounted surface not just what the linear bearing spec's are
Mactec54
Hi Martec -Yes agreed again, doesn't the running parallelism spec imply that the the tolerance is for the entire rail length? and in 2) "consider rail length"? Cheers Peter getting the rails to a spec is not a trivial task.. I was amazed one day at a toolmakers I use and one of their very large and expensive Mazaks was being adjusted by a japanese factory tech. He had a small bag with some dial gauges and a couple of scrapers. By end of day it was better than 0.0001mm...
Hi - This is an interesting discussion so I looked a couple of things up. Lets say we pick an N grade bearing and rail (Normal). So its spec is 34um for a 3m run. So if we place this on a perfectly flat surface it will run out 34um up and down and 34um sideways. You also have to think about matching the cars as they have a height spec of +/-0.1mm (and a H variation of 20um) so you have to grind match the cars or accommodate that in the build. A Grade AA 2.4m long granite surface plate has a flatness of 12.7um. Looking at machinery it's tough to get past 5um tolerance (IT1 grade), Mori spec 6um for their long mills. So say we set it up on a Mori and machine it. Do we set up on the averages? Then we scrape the highs down to the average but we still have the dips? Or do we set up and reference to the bottom and then scrape/stone down to the lowest dip? Anyone done this? Peter