[The Blight]

"If the machine is so wonky that it takes multiple goes and operator hands on input to get it right.......that ain't CNC"

It really depends on what you are making. What about die and tool makers? Do you think they just scrap parts all over the place without trying to measure, and correct before it gets that far? Imagine working on a injection mould and just running through the program without actually checking what the machine is doing because "that ain't CNC". Nah! The operator should just lean back and cross his fingers!

I actually do this for a living, so I should know a thing or two about it. If I'm working on a big series of parts I can accept a few scrap parts along the way, but when working on single off parts or small series I can machine them to tight tolerances without ruining a single one. The material usually cost too much to scrap any parts at all. I have on several occasions machine parts from 316L (Stainless steel) that are Ø45mm (1,77") L300mm (11,81") to a H6 tolerance (+0 -0,016mm / +0" -0,00063") over the entire length. I haven't ruined a single one of these yet, and I have machined quite a few.

If you scrap multiple parts where the material costs $500 for a single one you wouldn't keep your job for long. After having machined 10.000s of parts in all sizes I haven't even filled a 10 gallon bucket with scrapped ones yet.

In one breath you say you are new to CNC, and the next one you are lecturing people about it like you know what you are talking about. Last I saw you don't even know how to program one of these things and you have 4000+ posts on a CNC forum!

[siam]

Aluminium is generally considered to be about the worse material to make machines out of. Low mass, reasonable stiffness, low damping is a bad combination. The G10 sheet could be good. But it's expensive. If the piece of plate you need is less than 600mm wide then you may be able to get it for nothing in synthetic stone. Small pieces aren't useful in kitchens and bathrooms, so they tend to just get thrown out. I bought a square meter at full price recently. It was only AU$150.

The plate size I need is only 432mm X 1150mm (17" X 45.25") and not truly a structural member as it will see only compressive forces from the linear rails.



Rather than grind the surface of the steel tubes flat, I will apply Epoxy glue (JB Weld or it's equivalent, shown here as red strips) to the length of each steel tube. The Blue plate has already been laid out flat on my tile floor, checked to insure it's flatness, and has been pre-marked to locate exact steel tube placement. The steel tubes are now inverted (epoxy side down) and placed onto the flat plate. So long as the plate can retain it's flatness during this assembly process, I will have a perfectly flat surface to mount the linear rails onto. The pic doesn't show them, but the linear rails are screwed onto the steel tubes, after being drilled. I wouldn't want to use a sheet of plastic for this plate, but many other materials should prove quite acceptable.

[siam]

It appears that the bokts are all on a flat steel tab.. Which would tell me that the whole rail table can flex in the z axis but will be less flexible in the X axis. What happens if you pop it with a hammer? Does it ring? Try it in both the z and x direction, if there is a different tone it will tell you which is the better secured.

My guess is the z direction will resonate more as it is depending on 5 flat tabs which will allow some more movement and a different frequency. The suggestion about adding a stone top is a good one, it would need to be epoxied to whatever sub structure you might connect it to too also so it will help quiet the frame down. You can use a diamond bit to bore holes through it and use a T type insert from the back to the top face and use epoxy on them as well and you have a nice flat mounting surface with threaded holes for your rails.

It is actually a great suggestion, you might consider working it into your build.

I'm going to need to delay my answer on this one,...I've already taken the green rail table out for a slight modification; I need to cut out some metal plate to provide clearance for the motor.

[siam]

Hi Siam,

How are you planning to do the headstock on your lathe.

Mark

The front part of the headstock consists of a 6mm thick flat plate welded in place and braced front and back with steel tubes.
A steel bearing housing will be mounted into the 5" diam hole and bolted in place; the flange on the bearing housing comes to within about a centimeter of the inside walls of the box formed by the steel tubes, and should be more than adequate at preventing movement in all directions.
The back part of the headstock is made from 3mm plate steel and will also host a bolted in place bearing housing; the back bearing will see no forces in the Z direction.

Attachment 294252

[handlewanker]

"If the machine is so wonky that it takes multiple goes and operator hands on input to get it right.......that ain't CNC"

It really depends on what you are making. What about die and tool makers? Do you think they just scrap parts all over the place without trying to measure, and correct before it gets that far? Imagine working on a injection mould and just running through the program without actually checking what the machine is doing because "that ain't CNC". Nah! The operator should just lean back and cross his fingers!

I actually do this for a living, so I should know a thing or two about it. If I'm working on a big series of parts I can accept a few scrap parts along the way, but when working on single off parts or small series I can machine them to tight tolerances without ruining a single one. The material usually cost too much to scrap any parts at all. I have on several occasions machine parts from 316L (Stainless steel) that are Ø45mm (1,77") L300mm (11,81") to a H6 tolerance (+0 -0,016mm / +0" -0,00063") over the entire length. I haven't ruined a single one of these yet, and I have machined quite a few.

If you scrap multiple parts where the material costs $500 for a single one you wouldn't keep your job for long. After having machined 10.000s of parts in all sizes I haven't even filled a 10 gallon bucket with scrapped ones yet.

In one breath you say you are new to CNC, and the next one you are lecturing people about it like you know what you are talking about. Last I saw you don't even know how to program one of these things and you have 4000+ posts on a CNC forum!

Hi, New to CNC as in not ever had hands on experience and certainly not lecturing....there is a difference......I have also not built a car or bus, but I do ride in the damm things when the occasion arises.

I also spent the last 17 years of my active working life designing and tooling up CNC and linear transfer machines, so I know what scrap rate is and how it can be handled.

I don't think Siam is contemplating doing any moulding or die work with the build he's making, and I don't think you built your machine either...whatever......to do any mould or die work of consequence......what you said to that effect is contrary to this thread which is wholly about DIY self designing and purpose building a CNC lathe......from a steel tubing material content.

BTW......the barrel load of parts that got scrapped occurred over a weeks production and probably amounted to a few cents each for a finished item, relegating the ones that failed to pass inspection to the scrap and reclamation bin is more cost effective than reworking them.....do the sums........you have to be in high volume mass production to understand the logic of scrap value percentages and reclamation values.

CNC is only a logical step up from the form tool where on one hand you have an operator controlled output and on the other a machine.......both methods are designed to have an acceptable scrap rate......ours was 2 1/2% max.
Ian.

[siam]

It appears that the bolts are all on a flat steel tab.. Which would tell me that the whole rail table can flex in the z axis but will be less flexible in the X axis.

If you add all five 6mm tabs together and include their 20mm span, you're looking at a fairly substantial chunk of steel (30mm thick X 20mm wide) retaining the rail bed.

What happens if you pop it with a hammer? Does it ring? Try it in both the z and x direction, if there is a different tone it will tell you which is the better secured.

OK, with the rail bed back in the frame, I tapped it with a hammer in multiple directions, and I can't make it ring at all. Interestingly enough, when I hold the rail bed in one hand and pop it with a hammer, it rings like a bloody bell,...and with rather pleasing tone I might add. But once back in the frame, I get nothing except a metal clanging sound with no duration at all. I'm both pleased and disappointed,...as I said, the ring was a rather nice one.

[RotarySMP]

Here is the headstock of the new mill spindle I am making. It is 12-15mm wall thickness (the original rectangular material had variable wall thickness), welded all over. Even with my crappy welds, I am pretty confident it will be rigid within the limits of a Deckel G1L engraver frame.

Your 6mm sheet metal skin is only attached to the frame with a 4inches of weld.... and 3mm is awfully floppy for a rear bearing support... [I don't think you need me to complete this thought ]

Have you designed your spindle cartridge yet? There is a guy on the German site selling some really nice spindle bearings. I just bought his FAG B7210/ET/P4S/UL 's. He still has some with SKF 7211 and 7009's in P4 which might be sizes your could design around.

Peters CNCECKE!

Regards,
Mark

[siam]

Here is the headstock of the new mill spindle I am making. It is 12-15mm wall thickness (the original rectangular material had variable wall thickness), welded all over. Even with my crappy welds, I am pretty confident it will be rigid within the limits of a Deckel G1L engraver frame.

Your mill spindle certainly appears to be a massive chunk of metal, and certainly should work well for you.

Your 6mm sheet metal skin is only attached to the frame with a 4inches of weld.... and 3mm is awfully floppy for a rear bearing support... [I don't think you need me to complete this thought

I understand your concerns, but there are just over 20 other welds on the 6mm plate that you can't see in the pic I posted. The photo is intended to show the box structure which surrounds, strengthens, & supports the front of the spindle cartridge. As for the rear 3mm steel plate,...if you think about the force vectors being applied to the rear bearing, they are all in a radial direction, like spokes on a wheel,...I believe the rear plate is more than adequate for the job it needs to do.

Have you designed your spindle cartridge yet? There is a guy on the German site selling some really nice spindle bearings. I just bought his FAG B7210/ET/P4S/UL 's. He still has some with SKF 7211 and 7009's in P4 which might be sizes you could design around.

Regards,
Mark

Thanks for the link, but I bought my bearings about 2 months ago when I was back in the US visiting family. I guess I'm on the right path though, as the bearings I bought for the front end are two FAG 7211s and a single Timken 210K on the back end.

[RotarySMP]

Hi Siam,

What software and drive system will you use? I used to use TurboCNC on the 7x12. The canned cycles were cool, and dos boots fast. I eventually dropped it for LinuxCNC as the mixed up tool and offset tables in TurboCNC never worked for me. Since converting to LinuxCNC I haven't used that lathe much, and haven't really got my head around NGCGUI yet.

Stepper of Servo system?

Regards,
Mark

[The Blight]

"CNC is machining with your fingers crossed, as the final dimension is totally up to the machine and the operator just takes a back seat.....there is no capability to correct a dimension once the stepper/servo motor....whatever.... does it's thing......your sizes are all pre ordained by the in built accuracy of the machine itself."

"If the machine is so wonky that it takes multiple goes and operator hands on input to get it right.......that ain't CNC."

"CNC is after all a production tool, as in multiple parts to drawing, scrap all those that don't measure up, just keep making until the order is complete"

Oh I'm sorry for interupting your lecture about what a CNC machine really is and how it differs from manual machining. I just pointed out a false statement you made, and then give an example as to why it's wrong before questoning the validity of it.

How many passes you have to take to achieve the desired tolerances plays no part in what defines what a CNC machine is. CNC stands for Computer Numeric Control. I gave you an example where CNC machines are used with 0% scrap rates which makes your initial statement wrong.

I also know perfectly well about cost/benefit analysis on manufacturing and at what point you start losing money. In some cases you lose money if you scrap a single part. High volume production is not the only thing you use CNC machines for. We do both high volume and small series production runs. One does not exclude the other. The machine I built was for hobby use and not for high volume runs, and I don't use this for work. For work I actually use heavy duty machining centers. Siem is building a machine for hobby use as well, and he can take as damn many passes as he wants to achieve the desired tolerances without anyony caring or lecturing him about it.

Welcome to the ignore list. I won't be replying to you again.

[handlewanker]

Oh I'm sorry for interupting your lecture about what a CNC machine really is and how it differs from manual machining. I just pointed out a false statement you made, and then give an example as to why it's wrong before questoning the validity of it.

How many passes you have to take to achieve the desired tolerances plays no part in what defines what a CNC machine is. CNC stands for Computer Numeric Control. I gave you an example where CNC machines are used with 0% scrap rates which makes your initial statement wrong.

I also know perfectly well about cost/benefit analysis on manufacturing and at what point you start losing money. In some cases you lose money if you scrap a single part. High volume production is not the only thing you use CNC machines for. We do both high volume and small series production runs. One does not exclude the other. The machine I built was for hobby use and not for high volume runs, and I don't use this for work. For work I actually use heavy duty machining centers. Siem is building a machine for hobby use as well, and he can take as damn many passes as he wants to achieve the desired tolerances without anyony caring or lecturing him about it.

Welcome to the ignore list. I won't be replying to you again.

Your prerogative......

[siam]

Hi Siam,

What software and drive system will you use? I used to use TurboCNC on the 7x12. The canned cycles were cool, and dos boots fast. I eventually dropped it for LinuxCNC as the mixed up tool and offset tables in TurboCNC never worked for me. Since converting to LinuxCNC I haven't used that lathe much, and haven't really got my head around NGCGUI yet.

Stepper of Servo system?

Regards,
Mark

I'm already using Mach3 on my mill, and I'm fairly familiar with it, so I will use it first. As an old retired guy who will use his lathe strictly for hobby purposes, I don't really have the need for the additional speed & power that servos have over steppers, so I will use steppers. Most of the projects I have in mind to make with the lathe will be one-of-a-kind parts, and I suspect that I will make a lot of cuts by holding down an "arrow key", or writing a single line of G-code. I do plan to cut threads and curved surfaces, which will no doubt take a bit of time for me to get right, but I believe that Mach3 has both of these functions.

[RotarySMP]

My experience with the 7x conversion was that I almost never generated G-code programs in advance and ran them to manufacture multiple parts. Normally I just used the MDI and canned cycles. I wrote a couple of Excel macros to generate the code for a few things like tapers. The few programs I did write in advance, I used a free DXF converter to make the tool path. Lathes being 2D, have pretty simple tool paths.

Using the jog keys to actually machine is not great. If you like spinning handwheels, adding a jogging encoder, or even better two, would be a better option, but really with the MDI, it is not necessary.

I took a look at Mach 3 when I ditched TurboCNC. it has excellent canned cycles and wizards. Can't imagine needing a CAM program with the strength of Mach 3's wizards. It is obviously a much more user friendly program than LinuxCNC.

I choose LinuxCNC because it is free, more powerful, and more open. The learning curve is pretty steep though, and miss the canned cylcles and wizards (should put in more effort to learn NGCGUI).

Mark

[samco]

There is also something called features that should make it into linuxcnc in the future

LinuxCNC :: Topic: LinuxCNC Features - a kind of NGCGUI (1/27)

I have not used it but it looks pretty powerful.

https://www.youtube.com/watch?v=fkOJhT69WEc

[siam]

Ordered the stepper motors & their power supply today from a Chinese company that I've purchased from before, http://www.wantmotor.com; their sales people listed in their eBay adds have great English writing skills, and are most helpful if you want to have all your items shipped in the same package, to save shipping costs.
Z axis: NEMA34, p/n 85BYGH450D-008, 2.13V rated at 5.6A (1,090 Oz-Inch). Cross Slide: NEMA 23, p/n WT57STH115-4204A, 3.78V at 4.2 A (428 Oz-Inch). Switching Power supply: 48V at 7.2A.
I bought the motor drivers a few months back from James Newton with MassMind.org ( Slowing 69.16.243.61&c=1&t=42287.29196875 ); the dirvers I bought are based on the Toshiba THB6064AH chip and will drive the motors at 50V at 4amp. I haven't had a chance to use these yet, but the mill I built a few years back uses another (older) driver he also sells, so I have no doubt these new drivers will perform as advertised.

[siam]

As I've mentioned in previous posts, living in a foreign country presents some challenges from time to time: Most (maybe all) of the steel alloys I was familiar with in the US aren't used here in Thailand; none of the machine shops I've visited so far have any knowledge of alloys such as A36, A2, O-1, S-7, 10XX, 11XX, 40XX. So I'm having to learn the Japanese & Chinese equivalents, which are used here,...Google has been a huge help. So, here's a CAD of the spindle design,...from what I've read, it's a very common configuration with two angular contact bearings mounted back-to-back inside the headstock housing with a light press fit and mounted onto the spindle with an interference fit. A single ball bearing is used on the tail end of the spindle with a light press into the housing and a close slip fit on the spindle.

Attachment 295064

I will be using Mach3 and I've including a "timing disk" with 30 slots around the outer circumference for good speed feedback at slow RPM, and a single index hole which can be used to reference the rotational location of the spindle for thread cutting. I may also place a standard resolver onto the motor shaft if I need it.

Attachment 295066

[handlewanker]

Hi, the spindle layout drawing would be more meaningful if it was drawn 2D as an assembly, instead of an exploded collection of parts......my opinion only as that's the way I see it.

By stating "back to back" for the angular contact bearings.........does this mean you are fitting them in the housing with the cups facing inwards?

In that configuration the bearing nearest the chuck will have it's inner race pressed out of the outer race (minutely) instead of hard into it when the spindle is loaded and so unable to carry the loading from the tool......the second bearing of the two is only a back up bearing to maintain the preload.

If the cups face inwards, (back to back) the inner races are held face to face by the nut on the spindle against a shoulder, which means you can only preload the bearings by the front cap spigot against the outer races or by adding a shim between the inner races,

In my opinion, if the cups face outwards (front to front) and against the housing shoulder, held there by the outer cover(s), the inner race of the bearing nearest the chuck will be forced into the cup of the outer race and the nut on the spindle then serves to hold the inner races together and into the cups......a 2D drawing makes this clearer.

The fit for the bearings would be a close fit in the bearing housing, as the end cover holds the outer races in place and a close sliding fit for the inner races on the spindle to allow the nut to press the inner races together.....no press fits at all for this configuration.

I would also have two deep groove radial bearings, not one, spaced apart with a spacer between the outer races and held captive in their housing with a close sliding fit on the spindle for the other end, to offset any wear that the belt pull will exert on them......again my opinion.
Ian.

[handlewanker]

Here is a pic of a mill spindle illustrating the angular contact bearing configuration I mentioned in the last post.....note that the bearings are face to face mounted.....that is with the cups facing outwards..........the top end is irrelevant for this discussion.
Ian.

[siam]

Hi, the spindle layout drawing would be more meaningful if it was drawn 2D as an assembly, instead of an exploded collection of parts......my opinion only as that's the way I see it.

By stating "back to back" for the angular contact bearings.........does this mean you are fitting them in the housing with the cups facing inwards?

In that configuration the bearing nearest the chuck will have it's inner race pressed out of the outer race (minutely) instead of hard into it when the spindle is loaded and so unable to carry the loading from the tool......the second bearing of the two is only a back up bearing to maintain the preload.

If the cups face inwards, (back to back) the inner races are held face to face by the nut on the spindle against a shoulder, which means you can only preload the bearings by the front cap spigot against the outer races or by adding a shim between the inner races,

In my opinion, if the cups face outwards (front to front) and against the housing shoulder, held there by the outer cover(s), the inner race of the bearing nearest the chuck will be forced into the cup of the outer race and the nut on the spindle then serves to hold the inner races together and into the cups......a 2D drawing makes this clearer.

Ian, I could be wrong, but it appears to me that the only difference in arranging the bearings back-to-back or front-to-front is in which bearing takes up the thrust load created by, say drilling. The center-to-center distance between my two thrust bearings is only one inch, so I don't see a compelling reason to use one arrangement over the other, as I see them as being essentially the same.

The fit for the bearings would be a close fit in the bearing housing, as the end cover holds the outer races in place and a close sliding fit for the inner races on the spindle to allow the nut to press the inner races together.....no press fits at all for this configuration.

If you do not press fit both thrust bearings onto the shaft, aren't you risking radial movement of the spindle inside the inner bearing race??

I would also have two deep groove radial bearings, not one, spaced apart with a spacer between the outer races and held captive in their housing with a close sliding fit on the spindle for the other end, to offset any wear that the belt pull will exert on them......again my opinion.
Ian.

I think both configurations are used in the industry. I chose to allow the shaft to slip inside the inner race on the back end of the spindle to allow for thermal expansion, and for ease of disassembly when I need to replace the drive belt. As this lathe is designed for hobby use, I'm not too worried about bearing wear,....I'm far more concerned about rust forming while the machine sits unused in my shop here in tropical Thailand.

[siam]

It seems that at least several more weeks (months??) may be needed to have a spindle made; I've visited 4 shops, and so far, not one wants to take on the job. A 5th shop now has my drawings and promised to get back to me with a quote within the next few days,...I'm hoping they come through.

In the meantime, I really want to move forward with the build, but need to have a fairly good means of determining where the spindle line runs down the length of the lathe, as this will help me locate the positioning of the rails. So I set up a very crude "spindle line" and aligned it 90 degrees to the headstock plate using my trusty T-Square and an aluminum square tube. This set up should provide me with +/- 0.5mm accuracy, (close to the headstock) as I can read those markings on a tape measure. What gives me a bit of concern is knowing that my T-Square alignment is going to deviate the further I move away from the chuck end. Still, I feel pretty good knowing that I will be able to make final adjustments when I finally instal a real spindle and chuck.

Attachment 295072

Note that the rails are merely clipped to the bed at one end and hung with a wire from the other. The Saddle will be cut from that 3/4" thick plate of 6061 T6 aluminum. So far, all is proceeded exactly as planned.