Looking for CNC Mill Part I

[crankorgan]

I put the chocolate on the raisins. I took replacement Roller Blade Bearings and mounted them on aluminum angle. They are mounted staggered. I used this in both a 90 degree and 45 degree linear system. Nobody used this idea anywhere because it seems like it can't work. But it does!

Over on the Majosoft website under selfmade parts he has the first design anybody used for a homemade system using bearings. There were also people using bearings for plasma cutters. When I realized you could buy really good bearing as Roller Blade Replacement Bearing it sent a ripple through the hobby. Instead of $7 each the same quality bearing could be bought for $2 each. People laughed at the idea! While they laughed we built!
Roller Blade Bearing designed are now the most popular. Bearing riding on gas pipe no less!

[ger21]

If anyone has Fine Woodworking issue #101, August '93, take a look at the table saw sliding table article. This is where I first saw the rollerblade bearings on pipe. I used this idea in a panel saw I built 8 or 9 years ago. But it was John's website that showed it could be used on a CNC.

Gerry

[crankorgan]

Originally posted by ger21
If anyone has Fine Woodworking issue #101, August '93, take a look at the table saw sliding table article. This is where I first saw the rollerblade bearings on pipe. I used this idea in a panel saw I built 8 or 9 years ago. But it was John's website that showed it could be used on a CNC.

Gerry

Several people have told me it was an old idea. I think now it's an old idea with a new twist. Did they mount the bearings staggered on aluminum angle?


John

[ger21]

Yep, You'd be amazed how much it looks like yours.

Gerry

[crankorgan]

Originally posted by ger21
Yep, You'd be amazed how much it looks like yours.

Gerry

I was originally going with Roller Skate Wheels as sort of a joke. Then I saw I could buy tubes of bearings for $16 The gas pipe was left over from the Z-Beam of an older machine. The piece of aluminum was on the workbenck from something else. Mounting the gas pipe with the flanges was like stepping back in time to when I used to read my dad's old Popular Science and Mechanics books. The aluminum window channel was a big breakthrough also. The Pipe Dream design was made just for fun. The 7th Sojourn was designed because of a request from an RC guy who asked for a smaller machine than the Pipe Dream but bigger than Brute. The 7th is one of my favorites. Phoenix is slowly catching on. It was designed for longer RC parts. For
some reason nobody sees it on my website. Strange!

John

[clipclop]

Hi all ,
I just started reading this thread , for any aussies looking for controllers chech out silicone chip here : http://www.siliconchip.com.au/cms/A_30593/article.html
They have kits and info .
Also someone was asking about 5 wire steppers , from what i can find out they should be able to be wired to a 6 wire driver , as the centre conection to the coils is just internal instead of at the controler .
now my question ! are the steppers from old 5 1/2" floppy drives any use ? how strong are they ?
The ones I have are Sankyo msje200a53 12 volt
Stewart

[crankorgan]

Originally posted by clipclop
Hi all ,
I just started reading this thread , for any aussies looking for controllers chech out silicone chip here : http://www.siliconchip.com.au/cms/A_30593/article.html
They have kits and info .
Also someone was asking about 5 wire steppers , from what i can find out they should be able to be wired to a 6 wire driver , as the centre conection to the coils is just internal instead of at the controler .
now my question ! are the steppers from old 5 1/2" floppy drives any use ? how strong are they ?
The ones I have are Sankyo msje200a53 12 volt
Stewart

The UK has the cheapest controller company. 5 Wire motors can be 5 phase. The will say .72 degrees on them. Floppy motors are 12 volts 5 or 6 wire unipolar and only 20 oz. THe work great in plotters. My first mill used floppy motors. I only got 3" per minute of usable torq. Very slow but I was able to run my first GCodes on it.

[clipclop]

Cranky,
Do you know the max RPM of the floppy motors?
as I was thinking maybe 3to 1 reduction to increase torque and definition ?
Stewart

[crankorgan]

A floppy motor running with 12 volts will give you 2 turns per second or 120 RPMs with 20oz of torq. Using 24 volts with resistors the motor will spin twice as fast. But torq goes down. On Ebay you can buy three NEMA style 80 oz 6 wire steppers for $45 there are several people selling the motors seen here on PanzyProofs machine.

[clipclop]

Cranky,
have you heard how the latest version of Kcam is ?Its just had a patch to version 4 in the last couple weeks .
Stewart

[Jerry Dotson]

I have used Kcam for 2 years. The patch was for v 4.09 and concerned some computers not being able to get the system timing right. I like it fine. I like simplicity.

[clipclop]

Thanks Jerry , Im new to this just trying to get all the bits together to build a mill , figured the way the free software is going to 30 day trials I better d/l while I can .
Stewart

[xairflyer]

Just finished setting up the X axis table on my machine, ready to install the stepper and drive screw.

Something I thought about, when sliding the table back and forth on the rails, How does dust/chips etc effect the running of the bearings on the rails ? Can't imagine the rail staying clean and "bump" free !

My bearings/angle.

[balsaman]

Vacuum it off now and then.

Eric

[xairflyer]

I am talking about when it is in the middle of cutting, you can't stand there with a vacumn cleaner !
I could devise a form of wiper, but I wanted to know, is it a problem or does it occur.

[Doug Blackmon]

Originally posted by xairflyer
I am talking about when it is in the middle of cutting, you can't stand there with a vacumn cleaner !
I could devise a form of wiper, but I wanted to know, is it a problem or does it occur.


One idea may be to take another piece of angle and sandwich it to the piece with the bearnings so that it covers the upper bearnings. This should control most pieces from getting on the rail. You could also mount a small brush to the lead and tail end of the cutting platform as well to take care of materal that the angle missed.



Doughttp://home.earthlink.net/~daflea/images/Idea.JPG

[gjahnke]

A while back somebody was posting baout looking at a belt driven spindle (maybe cranky??). Did anybody ever look any further into this?

A guy I work with has a minimill he bought from Homier, I was thinking I might go over and talk him into letting me tear apart his spindle and see how it is done. As I understand it, it is a gear driven unit.

Has anybody ever pulled a minimill apart? I am guessing they are using either aluminum or bronze bearings as opposed to ball bearings. It seems like at the relativly low speeds at which most of the minimills operate, you could get away with aluminum bearings just fine, as long as you have a decent mass to them (to keep them from heating up too bad). Alternatively, I was thinking about using either bronze bushings, or graphite plugged bronze bushings and a gear driven spindle (I thing a belt drive would put too much lateral strain on the shaft).

The machine is for my brother, who lives too far away to be swinging by to use mine. He wants to be able to machine foam, wood (hardwoods and soft) and limited steel. The machine I have in mind is a fixed gantry which incorporates the Y and Z axis movement. I plan for the X axis endplates to be 2" cast aluminum. The gantry I planned to cast from steel. I also thought I would cast the working table out of aluminum (I recently helped a friend cast an aluminum table for a bridgeport machine that has worked out fine, even when machining steel).

I am still trying to decide whether to do steppers or servos. I have a cheap source for servos, but he is not technically inclined and steppers seem to be an easier setup, so he might just have to live with a slower machine.

I know you can machine steel w/ a rotozip, assuming a good endmill, a rigid frame and a SLOW cutting speed and multiple pass cutting, but I would prefer to do something with, say, a 3/8 or 1/2 hp induction motor not only to speed things up a bit, but also to cut down on the noise a bit.

[PanzyPoof]

xairflyer looking good ! As to your worries about the rails and dust ! Just brush them between runs, no biggie :)
If your Married? your life may depend on how much dust you drag around the house.:(

[uscra112]

Short of pressurized-oil (i.e. hydrostatic) lube for those spindle bearings, there ain't a plain bearing ion the world that'll stand the RPM you want for routing and still have any capacity to bear the side loads. Use ball bearings. Keep them fairly small, too.

[shim]

Hi,
Wondering if there is anything different between a normal bearing and a rollerblade bearing.
Any body have a picture of a rollerblade bearing next to a normal metal bearing?

Thanks.
-shim

[oz9ny]

Originally posted by shim
Hi,
Wondering if there is anything different between a normal bearing and a rollerblade bearing.
Any body have a picture of a rollerblade bearing next to a normal metal bearing?

Thanks.
-shim

There is no difference at all! made by the same companies SKF,NEC you name them

Quality codes is the same ABEC xx or similar.

If you buy the same quantity of bearings in another size you can get the same discount.

/Niels

[balsaman]

They are the same. Rollerblade bearings are cheaper because they make so many of them.

Bearings can be open, shielded, or sealed. Rollerblade bearings are generally the shielded variety. Sealed have a black plastic seal covering the balls, shielded use a thin metal cover, and open you can see the balls between the races.

[clipclop]

Brgs come in mixed format too , you can get a brg with a sheild one side and a rubber seal on the other . or a sheild/seal on one side with the other open sealed brgs have a slightly higher turning resistance to sheilded types due to friction/drag of the rubber .
Twin rubber sealed brgs have the designation 2rs , cant remember the other designations .
From my experiance rubber sealed brgs last longer as they keep the grit out better than sheilded types .
Stewart

[uscra112]

But the seals create friction. There's a parameter we use in machine tool design called "stiction". What it refers to is the ratio of static to dynamic friction in a system. (Static friction is always higher in sliding contacts.) A high ratio makes it hard for the servo system to make small, precision moves. Worth thinking about . . . .

[NEWMERIC]

This is my first post in this most important quest of building a CNC mill. I'm a journeyman wood pattern maker who now works from home using CAD/CAM to model and toolpath patterns for the foundry industry.Any questions you have concerning CAD/CAM software are welcome. I use Worknc (France) and Cimatron (Israel) CAD/CAM software.

Who would have ever thought that in 1976 when I decided to pursue the pattern making trade at Milwaukee tech and trade high school and learning how to build highly accurate patterns by hand using traditional wood working machinery and techniques, that I would someday be doing the same thing from a home office with one hand at the keyboard and feeding my infant son his bottle with the other. We live in a fast changing world and I intend on getting an answer in these discussions so we can all build a highly accurate +-.0003" cnc mill. I'm greatly dissapointed that the only forums I find are for open loop stepper motor designs. I need to build a benchtop closed loop ac or dc?? servo design.

It was disappointing to see after I attended the International Machine Tool Show in Chicogo in 2002 that with all the vendors selling individual components that make up these accurate and expensive machines that there isn't a single book or single source of information that you or I could go to and learn how to combine and assemble these individual components like the big machine makers do and build (or retro fit) one of these highly accurate machines ourselves.

To give you an idea of what I would like to accomplish would be to build a machine I saw that is used to mill small injection molds at high speeds. It cost thirty thousand dollars. I want to do the same but avoid
any and all markups on the final price I can. Providing the labor and avoiding the profit the machine manufacturer would get. I don't think I'm asking for much. Just a book or a website that deals with building high end CNC benchtop mills that you can buy for 10 to 30 thousand dollars from other machine builders but do it yourself. Can anyone help please???

With the highest regards
Byron

[gjahnke]

Check out http://groups.yahoo.com/group/CAD_CAM_EDM_DRO/ for info on homebuilt servo systems. For the money you are talking, though, you can pick up a good used system. A really good homebuilt system usually has a resolution (realistic) of around .001, which is usually "good enough" for the homebuilder since the tolerance on cutters is more than that.

Just out of curiosity, why do you need that much precision for investment casting? Standard tolerances for investment casting are +-.010 for up to one inch, and +-.005 for each inch thereafter. What kind of process are you using for the casting that gives a +-.0003 resolution?

You are going to have some trouble finding any such machine from a manufacturer. While several machines have a theoretical resolution in that range, the only one I have ever seen that had realistic tolerances in that level was at Los Alamos (nuke cores are milled to +-.0005). We have a machine at work with a theoretical resolution of .0007, and you have to start by milling your cutter, since they don't generally come from the manufaturere close enough to spec to make that kind of resolution.

[NEWMERIC]

.Thanks for your help. I will do my best to keep you posted on any progress in developing a easy way to chose and assemble high end components without being a wanabee rocket scientist.
Plug and play seems to be creeping its way into this field and I would think its only a matter of time before we'll see an easy reference table puplished by the manufacturers of these components that will allow us to build a high end machine from scratch.

Pattern makers are the least common denominator of the most common multiple. Therfor we strive for the highest initial accuracy. Those tolearances you refer to are the end result because the pattern makers tolerances are amplified by the post processing of the parts. That is why pattern makers must work within .002".
You are correct about the tolerances one can realisticly expect. And our machine and cutter tolerances at the shop fall into that catagory.
I am going to produce plastic injection molds for plastic kits. this will require
very small cutters and very high rpm's. I will look up the machine that I mentioned and post it as soon as I can
dig up the liturature. The machine can engrave the surface detail found on a dime and this is what I require.
The old pantagraph way of building a 1/4 scale epoxy tracing models and using the pantagraph to scale down and engrave or machine the steel mold cavities to produce plastic model kits is no longer the best way to produce these molds. I will be machining highly accurate and detailed carbon plugs for Electro displacment machining and burning the mold cavities.Take one look at a plastic kit from Tamiya
and invision the steel molds that are used to produce those exquisite little parts and you will know why I seek that kind of accuracy. I plan on running .002" diameter cutters. maybe not in steel but in machinable wax and then pouring a epoxy takeoff from it and then electroforming the nickle tool insert from the epoxy takeoff. very cool stuff.

[gjahnke]

Extremely cool. I have built two desktop CNC units. The first was more a proof of concept machine than anything else and was built for less than $100 from parts that I scavanged. The second I actually put some time and money into. It has a theorestical x/y tolerance of +-.0005, so a theoretical location tolerance of +-.001. The problem I ran into with actual tolerances was with the mass of the unit and vibration. I cast the machine from steel and aluminum (am getting ready to start another one for my brother). Actually sitting on a desktop, the real resolution was closer to .004 when cutting steel and .002 on aluminum. I ended up building a heavy stand for the machine from 21 lb steel I beams and 135 lb railroad track. The machine is bolted to the stand securely. By adding the additional mass (all together the stand weighst about 450 lbs) it dampened the vibration significantly and brought the realistic tolerances a LOT closer to the theoretical tolerances.

I mounted my feed motors on gel isolating pads so the amount of vibartion they transfer to the machine is limited. Doing that with the spindle, though, would have just encouraged more spindle vibration, so I decided it was best to solid mount the spindle and use the mass of the machine to dampen the vibration. It just wasn't heavy enough (the finished machine weighed a couple hundred pounds). I think this will be the bane of "benchtop" CNC for a while. For precision, the best you can really do IMHO is a "small stand" CNC.

The precision mills I saw at Los alamos were supposed to be accurate within a couple of microns. They were mounted on great big concrete piers that went several hundred feet into the ground and were seisemically isolated. The theory was that any vibration would be contained within the machine and the pier, so it would all vibrate together. I am assuming that the theory holds true (I have seen some precision gimballed flight simulators built on the same kind of platform).


My first machine (the proof of concept one), I used standard 1/4-20 leadscrew (from the hardware store). I was actually very impressed with the accuracy that these gave, and realistically they would have served my purpose just fine (mostly I use the little machine for cutting circuit boards and some gun parts). I did not want just fine. I wanted perfection. I ended up buying a used precision lathe to cut my own lead screws. I gave $1200 for it, including a full set of indexing gears for it. I thought that was a pretty good deal. The leadscrews I wanted were $300 each, and I was going to need 4 of them. I decided that I would rather spend the $1200 AND have a precision lathe than spend the $1200 and just end up with 4 lead screws. I ended up spending another prbably $200 on bar stock while I learned to use the lathe and the indexing gears, but to my way of thinking, that meant that I spent $1200 on the leadscrews, and $200 on the lathe. Pretty good investment from my viewpoint.

When I cast toerance critical parts (usually in either steel or aluminum) I usually design the part, then oversize it by betwen 3 and 5 percent (depending on finished size). I cut my wax impression, cast the part, then go back to my original drawing (that has not been oversized) and re-mill the finished casting. Usually in steel I end up with a finished tolerance of around .0014, and in aluminum it is around .0012.

Not perfect, but on a lot of parts I can have the thing milled in the time it would take just to get my big machine set up.

Steppers seem to work fine if you keep to finer threads. Of course, this makes for a slower machine. A servo setup with optical linear index would be the way to go if you are looking to spend some money and get some real speed. Personally, went with steppers and then cut 1/2 fine thread lead screws. My thinking on this was that I could still get a reasonable cutting speed and not be tempted to over run the thing and wear out my spindle (I don't know if you have priced precision motorized spindels yet, but they are not cheap). Of course, in plastic this would not be so much of a problem.

If I was rebuilding a machine for myself from scratch, I might be tempted to go with servos and 10 tpi leadscrews. I have seen on the web where some folks go with steppers and 10 tpi precision lead screws. I don't really see the logic in this, since by going to the coarser thread, but sticking with the same steppers, they are effectively halving their resolution anyway when compared to using 20 tpi screws. It seems like it would almost be better to go with stock 1/4-20 allthread from the hardware store, run a precision die down it, then tap your own nuts with a precision 1/4-20 tap. I have not tested this theory, but I have run a precision die down a piece of 1/4-20 allthread before, and you end up cutting a lot of material off. It seems like it would be cheaper (~$50) to do it that way than to buy the precision screws (~$500 for what most guys seem to be buying), and when you consider the added resolution gain, the finished result should be more a machine with more precision than using the good leadscrews. Just an observation.

Resolution should also be improved by "half stepping" the motors. The problem I have had with this is that speed really goes to pot. If I try to speed up the process in software, I end up with missed steps. In theory, whether you are full stepping or half stepping the motors, you should have the same theoretical full RPM. After some experimentation, I believe the problem is with the speed of the parralel connection btwn the computer I use to operate my little mill and the mill controller itself. I was thinking about designing a USB connection to see if that would solve the problem. Anybody have any thought on this?

[ger21]

You might want to take alook here:

http://www.5bears.com/cnc.htm

Gerry

[xairflyer]

Seen somewhere diodes added from each side of the coils to positive, so I decided to add them in to my driver circuit and see what happens. K179 kit from www.kitsrus.com

I connected each one first by mistake to the 12v (bypassing the dropping resistors) and the motor ran at the same speed but sounded very rough.

Connected them then to the other side, matching the centre taps, as I am using two resistors (just in case it would make a difference) and the motor ran very very slow.

Realised then afterwards that fitting diodes like this was really for protection of transistors and as I am using fets, it just made things worst.

Any views ?

[gjahnke]

I like the kit. The diodes that are used on some controllers are used to control power bleed back into your controller.

Diodes come in different values (just like resistors and capacitors). If you are just hell bent on having them, make sure that you get diodes that match the current flow you need. If the diodes you got DO match the current flow to the motor AFTER the resistors, and you hooked them up BEFORE the resistors, you might have fried them (tough call w/o more info).

Unless you really need to protect something in your circuit, I would toss the diodes in a drawer and not worry about them. My first control circuit was transistor driven, and I didn't even bother to put diodes on that one, and never had any problems with it.

[Yukonho]

I think the best solution by far is to spend $125 on the xylotex 3 axis chopper board. 1/8 steps, true chopper (no resistors) smoooooth running motors, no hassle at all. Plug and play. With MAch 2, it is incredible. I am so happy I didn't bother trying to screw around with anything else.
co

[gjahnke]

I still advocate building your own controller. Not only is it cheaper, but it will be easier to expand later, like when you decide you want to add a fourth axis (which is invariably followed by a fifth, sometimes even as sixth). Personally, I have worked on too many propreitary systems to trust them much. I have found that when you buy a propreitary system, you are often SOL two years later when you need to replace a part and the manufaturer has gone out of buisness. If you build your own board, you KNOW what is there and how it works. When you burn up a chip, you KNOW what the chip is and can usually find a spare or a drop in replacement, as long as you make good decisions when it comes to chip selection.

Anybody interested in designing their own board might want to take a look at the allegro 5804 chip. It is what I based my controller on. It has some pretty nifty features built in, like externally selectable wave drive, two phase drive and half step. You can do a lot worse for $1.20 each.

Of course, I tend to be kind of a spend thrift (some might even say cheap), and couldn't bear the thought of spending $125 for a 3 axis controller, when I could throw a 5 axis unit together for just under $25 in about an hour.

I like the kit that xair ordered. I didn't see the price (which might make me like it less). He could mount the parralel interface on a seperate board, connect the actual controller boards via molex connectors, and that way if he burned up a board, he could plug a spare in and be back up and running in about 2 minutes, if he was so inclined.

That particular design would also be incredibly easy to add a pendant. All you would have to do is run control wires from each board from the internal/external control switch and the direction switch. Pretty nifty. I considered re-designing my board to include a pendant (I was thinking about using a simple 555 for the timing pulse) and will probably do it sometime in the future (you can't runsh these things. opf course).

[xairflyer]

gjahnke

I don't think the price will upset you, I got mine from the UK dealer for £9.95 !
There seems to be lots of distributors of KitsRus kits in the US and price's vary http://www.mpja.com/ are selling this kit re numbered 8479kt for $15.95 .

You mentioned the 5804 they also have a kit using this as well and a pulse generator on board (555), I have one of these also, been playing with the K179 kit because the K109 is limited to a 1.25amp motor in std guise.
The K109 kit sold by mpja is $16.73. (8409kt)

All and all I think as I have said before on this group, damn good value for money.

[xairflyer]

Forgot to add in this range there is is also a PIC controlled 5804 driver kit as well K94.

And a two stepper driver board designed to connect streight to the LPT port and only uses transistors K113 got one of these as well, to play with, hav'nt tried it yet. Very basic.

[gjahnke]

You are 100% correct on the power problem with the 5804. This is easy enough to rectify. I actually chose the 5804 because one of our IC salespersons wandered into my office one day while I was working on a design for my controller board. She had 25 samples of the 5804 shipped to me.

I still think that a USB controller is the way to go. I am working on getting one designed and prototyped. When I get a functional (and cheap) design, I will post the schematic somewhere.

I am thinking that not only would the USB module give you faster communication between the PC and the machine, you would also have more I/O ports available (think "8 axis desktop mill). This would also allow you to control multiple machines from the same computer (multiple machines at once, of you machine is good enough).

A buddy of mine bought all the Gingery books and built the lathe, the drill press, the shaper and the horizontal mill, then he converted them all to CNC. If you are not familiar with the gingery books, do a google search for gingery. HE writes a bunch of books about how to build the above tools by casting the parts from aluminum. What my buddy did is bought the books, re-created the drawings in autocad, then brought them over and we cut them out of foam on my mill. We glued on sprues for each piece, dipped them in a plaster slurry and let them dry (repeated that 3 times), and then cast them all. This was a LOT quicker than making wood forms and then casting them in greensand like they do in the book.

I think it took about 5 hours to cut all the foam for the castings (we were drinking beer, not actually doing anything). Another 2 3 hours dipping and drying (more beer drinking), then he cast all the parts the next day (I think it took about 3 hours to cast the parts).

I don't know how long exactly it took from start to finish, but I know that he was off for 4 days, and in that time built the shaper, the lathe, and had the mill just about done. Pretty good results for playing around on vacation.

Anyway, the point is, he could run all this stuff from one machine (if he was so inclined) if there was a good, cheap, reliable USB controller available.

[xairflyer]

Been looking at modifying this 5804 kit to allow for bigger motors and adding a chopper circuitry if I can locate a diagram for it.

[MrChips]

gjahnke

How well do thoes all alum machines hold up? Seems they would not be beefy enough for cutting steel.

Thanks
Hager

[gjahnke]

Xair, I don't bother messing w/ a chopper. I use fets and build my power supply to a decent spec. Haven't had any problems so far. I ran it past some of the engineers at work who specialise in such things, and they said that going to a chopper MIGHT give a limited increase in performance depending on circumstances, but if reliability is key (and it always is for me) and parts availability is important (again, a must have in my book) that "other options should be seriosly examined before going to a chopper type circuit" .

MrChips....As far as I have seen they hold up really well, or not worth a damn, depending on how you treat your machine. I made a lot of my desktop CNC machine out of aluminum (the X axis ends, the table, parts of the Z axis and gantry). If I chuck up a dull 1/2 bit, try to run it at 18,000 RPM and at maxim speed, the machine will not last long. Using sharp bits and reasonable feed rates, I expect it to last a long time.

My friend who built the shaper, lathe, horizontal mill and drill press thinks they perform really well. I have only had occasion to use the lathe, and can honestly say I was impressed by what it was capable of (especially considering the $20 price tag!).

I would expect that as long as you take care of your machine, they should last a long time. A lot of folks seem to have built them and I haven't seen any "wears quicky" complaints. Of course, if you notice things starting to get sloppy or parts starting to wear, you can always cut a new pattern (or re-use your wood core if you go with the greensand casting method) and cast a new piece. With the exception of the lathe beam, which has to be hand scraped according to the book (we just milled it in my mill) I would imagine you could completely replace any part of any one of those machines in about 2 hours.

I have probably 70 hrs on my mill machining steel, and have not seen any wear on the aluminum parts. I was a little concerned about this also, so I cast 2 of everything I needed. It looks like my worries were unfounded.

[MrChips]

Thanks gjahnke
Getting use results/practices from a person that has first hand information is always better than speculation.

I have never done casting of any type, and the foam method seems appealing to me over sand casting. Could you recommend a web site where they discuss the pros and cons of both methods? As well as casting sites that you feel are worthwhile.

In sand casting when bedding the wooden master in the sand, how do you know you have it exactly half way in the sand so that the cast piece won’t be thicker or thinner than the original?
Thanks
Hager

[uscra112]

Wow - here we are discussing CASTINGS, now!

I know that the "lost foam" method is widely used in the auto industry where I now work, (as a supplier). I also saw parts made for machine tools when I was in that business 20 years ago. The foam pattern is placed in a box and rammed-up with green sand, then baked to melt out the foam, and dehydrate the sand. The plaster method I know is done in the firearms industry for small and intricate cast steel parts. The pattern is dipped in plaster until it has a suitably thick coating, then placed in a sand-box to support it while the foam is melted out and the metal is poured in. In plaster molds you have to provide runners or at least holes for the air to escape. A old-style green-sand mold is porous, so the air can escape thru the sand, but plaster is not.

Web sites:
http://www.backyardmetalcasting.com/
http://www.visi.com/~darus/foundry/

Bedding a pattern goes something like this:
You have a two-part box. You pack the pattern in the lower part until it's bedded halfway. Then you set the upper part on. The upper part has a removable top. You lay a piece of paper (plastic nowadays, more likely) that is cut out around the pattern on the sand in the lower box. You then fill and ram up the upper box. Put the top on the upper box. Now, if the sand is well rammed-up, you can separate the two boxes and have half the mold in each box. The thickness of the part cannot be other than the thickness of the pattern. The mold is them baked to evaporate out the water, and harden it up. Now the metal can be poured in.

The lost foam method is a variant of the ancient lost-wax method. In either case you simply melt the pattern out, rather than splitting the mold to recover a wooden pattern.

MUCH easier, IF you can make the pattern(s) easily. That's why lost-foam is the bananas for small-run castings, so long as the shape wanted is easily made up in foam.

[MrChips]

Now I understand.
I was thinking the bottom half of the mold would be made.

Remove the original and make the top half, then put the two halves of the mold together.

DUHHHH.
Thanks

[gjahnke]

also check out http://www.buildyouridea.com/ he casts some CNC mill parts using the lost foam pattern.

If you are going to cast machine parts, stick with the slurry dip (investment) method. If you try to ram sand over styrofoam, the styrofoam always seems to compress or otherwise distort.

When you use the paster method, you just take the dried mold, with the foam still inside, and bury it in sand with the sprue sticking out. Regular playground sand works just fine. Pour in your hot metal (the foam vaporises on contact) and wait for a bit for the metal to cool.

Tolerances are similar to what you would expect from traditional invesment (lost wax) casting. Nothing could be simpler.

For my plaster slurry, I have quite buying plaster of paris. Now I get a box of sheetrock mud at the hardweare store and mix water with that until it is about like pancake better.

Little piddling around projects seem to take on a whole new deminsion once you have the ablity to cast your own metal parts (practice w/ aluminum for a while, then E-mail me and I will let you in on some steel casting secrets).

I thnik I may hve to build the gingery shaper. I know shapers are usually considered to be kinf of an outdated technology, but the sure do come in hands when you are looking to cut a dovetail or a bevel.

[uscra112]

Logan made a benchtop shaper many years ago. I have one that was made after South Bend bought the line. It is handy at times. You might still find one for sale on eBay.

[gjahnke]

I am a little hesitant about buying stuff like that frob ebay. I do buy a lot on ebay, but something like an old shaper that have waearing parts that are not readily availabe kinf of makes me think twice.

Besides, once you get the vertical mill done (CNC mill), toolmaking seems to be a hobby all its own. Suddenly, I have tons of ideas for ways to make new and exciting tools, or to improve my current inventory.

[uscra112]

Yah. I got into rebuilding machine tools for a living, many years ago, and became completely infatuated with whole field. In the 19th century, the sector of the population that now does recombinant DNA and nanotechnology was busy developing steelmaking and metalworking technology that we now take for granted. Reliving that is still fun.

[gjahnke]

I posted a question a page back about building a spindle and bearing types. What do you guys think about using precision Timken (tapered) bearings in the spindle. The ones I have on hand are for a 1/2" shaft and are rated for a 710 lb radial load, and a 530lb thrust load. Tehy are supposede to be within a +.0005 -0 tolerance ID (shaft diamer) and OD (bearing diameter) for an overall tolerance of .001. In theory, using a press fit for the outer ring w/ a bronze collar, the OD diameter could be margenalised, so only the +.0005 tolerance for ID would need to be accounted for. I think you could program this out w/ backlash adjustment. Any thoughts?

[clipclop]

If you turn up the spindle shaft with a larger dia on the bottom nere the tool you can radius the reduction to the smaller dia so that the brg will self centre buy trying your 2 brgs on the shaft use the small id 1 at the top and machine your shaft to fit the brg's and you will have no free play .
Its also a good idear to reduce the dia between the brgs as this alows easy fittment , and reduces galling of the shaft when fitting the lower brg .
Tapered brg's are the best for spindle use as they are adjustable to give zero play dont forget to alow for seals above and below the brg's as they will need there own oil supply .
high speed use of tapered brgs needs carefull houseing design to alow proper lubrication , they need a constant oil supply but not full up with oil as they will overheat , I have found that the brgs need to be 1/3 flooded and space for the oil to circulate through the brg. so there will need to be a groove behind the brg from top to bottom to alow this .- the taper acts as a pump to move the oil it will travel from the small dia to laege dia of the rollers .
If these few things are done the spindle and brgs will probly outlive you .
Stewart

[gjahnke]

What would you suggest using for bearing seals?

[clipclop]

if you increse spindle dia to 3/4" below the brg get a 3/4" id seal , you may have to look through a catalogue at a brg supplier to see what your opptions are , the tapered brgs will make a supperior spindle but require some fancy machineing to get it right .
you may need to make the seal holder as a bolt on flange so that you can get a seal to fit the shaft and still get the outer race of the brg in place - this is how I would do it the seal carrier then can have a ridge to hold the outer race inplace
the top needs a cover plate as well, this needs to be dome a bit to allow room for the lock nuts used for brg preload , or you could use a spacer tube and have the seal run on this then clamp the pulley with the lock nuts when you set the preload on the brgs , dont foget to allow for seal drag when setting the preload on the brgs .
Stewart
OH seal type- I would use a full rubber covered seal with twin lips they last well with minimal leakage dont forget to lubricate between the lips when installing or they will burn out , lubricate with wbrg grease buy half filling the groove between the lips they will last long time then

[uscra112]

Actually Timkens aren't all that good for high-speed spindles. The thrust adjustment gets extremely critical - too tight and they'll overheat, too loose and they vibrate, and the faster you go, the narrower the band in which they'll be adjusted right.

I can explain why, but it gets into terms like elasto-hydrodymanic lubrication and Hertzian Zones. All the high-speed (15,000 and up) machine tool spindles I ever saw used multiple radial-thrust ball bearings. There are "4-point-contact" ball bearings out there now that do the same thing with fewer parts.

[clipclop]

I agree adjustment is criticle and a light free moveing oil with good antifoam properties is needed for high speeds , but the endresult is a very prescise spindle with no free play

[uscra112]

Take it from an old, crusty machine tool engineer - ball bearings are best. You cannot preload a Timken bearing that will be run at router-spindle speeds unless you plan to run it without any oil at all. Ball bearings can be preloaded with oil present. Radial-thrust ball bearings can be preloaded both radially and axially.

The only successful machine tool spindles I know of that used Tapered-roller bearings were horizontal boring-mill spindles, which rarely ever go over 3000 RPM. Even then, the later DeVliegs had to have an automatic (hydraulic) preload adjustment to keep the preload correct over the range of speeds they were supposed to run at. They also had to water-cool the bearing housings.

Most if not all high speed milling spindles have air-oil mist lube to the bearings. VERY little oil, and very thin oil. Anti-foam properties are irrelevant, since the bearing is not bathed in oil.

[gjahnke]

That sucks. I have been holding onto those timkin bearing for years, just waiting for a chance to use them. I guess I can put out the $30 for a set of sealed ball bearings.

I don't really want to have to mess with an oiling system if I don't have to.

[uscra112]

Hey, something else will turn up. They make good bearings for ballscrews, for example.

Sealed, grease-lubed ball bearings will work at the power levels you're going to use. And not too expensive, either. You won't need high-priced super-tolerance ones for this spindle. Send me a PM and I'll tell you how to set them up.

[Yukonho]

Sorry to interupt this particular train of thought.
I am currently using heavy duty drawer slides for my Z axis, and they are fine for light materials, but not solid enough for the materials I want to cut. I have searched high and low and cant find any quick and inexpensive alternatives.
Can anyone offer a suggestion?
Thanks
co

[Gary Blaylock]

Hi everyone,
Just found this thread a few days ago. I have been planning on making a CNC router for a while and this thread makes it look promising. I've seen some really good looking routers here. I have checked some web sites, recommended on here, for my electronics and I am thinking about the package from Hobby CNC. It is the 31V 3 Axis Package. The recommened transformer (power supply) is to be purchased seperately from an electronis supplier. Is anyone here using this setup? This will be my first attempt at this so I would like to make sure my electrics will be right.

Thanks in advance for any info on this package or any other suggestions.

Plan to be in the 1 percent club!

Thanks
Gary

[gjahnke]

Yukonho.....Once you move past the point whre drawer slides are acceptable, you can kiss "quick" and "inexpensive" goodbye. If you want quick, get some linear bearings and rail. Good stuff will set you back better than $900. For inexpensive, go back through the thread and see what guys have done with black pipe (plumbing pipe) and roller blade bearings. There is also a guy at www.buildyouridea.com that has some pretty nifty bearings set up. He cast his carriers out of aluminum.

If you are mostly cutting light stuff, and just want to be able to cut something heavier occasionally, you can always just slow your machine way down.

[Yukonho]

Hi gjahnke
I have seen some short linear ways that had maybe 3-5 inches of travel for $30 and even less on e-bay. There was one seller called The Manufacturist that offered something like that pretty regularly, but I cant find them now. There is one auction that I found that might work for me, but nowhere near as many as a few months ago.
I have seen the buildyouridea site. I am looking for something low profile, rollerblade bearings will take up too much space and leave my spindle hanging way out away from the y axis.
I have gone through much of this thread again and have not found anything that will help me.
I am hoping someone can provide me a link to one of the less expensive linear setups I am talking about.
co

[gjahnke]

When you say "heavier materials" how heavy are you talking? Also, how big a cutting area are you after?

I assumed you were talking steel. If that is the case, I would trust a $30 e-bay linear rail about as far as I could toss a '57 caddy.

[bgriggs]

Yukonho,

Igus makes affordable linear slides. You can check out their stuff at http://www.igus.com.

Their products use Drylin self lubricating surfaces instead of ball bearings. I have used the N80 slides and like them. I plan to post a thread on how I used them as soon as I can take the pictures.


Bill

[uscra112]

For Z-axis only, the THK or NSK preloaded ball bearing linear slides are worth the money. Not that I have anything against the IGUS - I have used them, (at work), but the Z-axis needs the extra stiffness of the preloaded types.

[Yukonho]

Steel? heck no, I use a dremel as a spindle and my router is made from wood.
I cut a couple of brackets that hold the dremel a lot more solid to the z now and that helped a lot. At least this will work until I can find a deal on some low profile ways for the Z.
co

[uscra112]

Ggjanke - sent you a PM - in three pieces.

[gjahnke]

Thanks

[PanzyPoof]

Whow See the thread is rock solid for over a year now:)
Off to see how many 1%ers are on Crankys page later:D

[Gary Blaylock]

I'm into the process of building my router. I downloaded KCam so that I could see what is involved in setting up the G-codes. I imported a simple drawing, saved as a .dxf file, from Acad which is a rectangle with two holes into KCam. It appears that KCam has plotted my tool path. How do you know if the tool path is offset to the outside of the rectangle and offset to the inside of the holes.
I'm just starting out in CNC and need some pointers to get me started.
Thanks for any help and will post some pics soon if anyone is interested.

Thanks
Gary

[ger21]

In order to offset the toolpath you need to use G41 and/or G42 codes. I don't think KCAM shows the offsets in the toolpath displays, but I'll try to check and see later. Most people using hobby software packages offset the lines 1/2 the tool diamater in AutoCAD before saving the .dxf, so G41 and G42 are not needed.

Gerry

[Gary Blaylock]

Hi ger21,

Is this correct: using a .25 bit offset the outside of my drawing by .125 and offset to the inside of the holes by .125. I assume that i need to make this offset in a different layer and import only the offset layer into KCam?

Should I look at another program such as Deskam?

Are there any books which will help in this learning process?

Thanks for the help,
Gary

[Gary Blaylock]

One more question.

Can I set the holes to be cut out first before before cutting the outside?

Thanks,
Gary

[Jojje]

Hi Gary,

Get uself a free cam-program like FreeMill and u can forget all about offset and G41/42 codes ;-)

/// Jojje

[DICKEYBIRD]

Originally posted by Gary Blaylock
Can I set the holes to be cut out first before before cutting the outside? Gary, AceConverter lets you assign cutting order by layers if you want but the easiest way is to put little gaps in your toolpath so the parts will be firmly held to the sheet. This way you can allow your G-code conversion program to optimize the toolpath for the shortest cut time without worrying about when to cut the holes. It's simple to cut the little tabs off with your exacto when you're done.

Also, don't assume that a 1/8" router bit is exactly 1/8"! Always run a test cut on a known width piece of stock, ie: cut it in half with your new bit and measure the result. My high price solid carbide 2 flute 1/16" and 1/8" bits actually cut .055" and .116"! :rolleyes:

Hang in there, after you get over the initial confusion, it's just common sense stuff. :)

[Gary Blaylock]

Thanks for all the help. Are there any good books for a beginner to help in reading the G-code?

I would like to see some pics of everones work!

Thanks,
Gary

[Jojje]

For my pics:

http://www.bergan.se/prod01.htm
http://www.bergan.se/prod02.htm

/// Jojje

[ger21]

Gary, you might want to try my AutoCAD G-code exporter. you can get it here:
http://www.cnczone.com/showthread.php?s=&threadid=2443
It only works with polylines right now, and cuts in the order the lines are drawn. Draw your circles first (Draw rectangles and use fillet to make sure they are polylines) and they will be cut first. It's pretty basic right now, but a little easier to use than ACE (imho).



Gerry

[Blackhawk3D]

I have just finished reading this great thread (Whew!), and now am very excited about building my own CNC Router. I want something the size of the Pheonix or Balsaman's, but don't know which design I sould go with. I know that rolling gantry designs are said to be harder to setup for a beginner, and are less stiff. Should I build the Pheonix, or make the 7th sojourn longer? THanks. I am leaning towards the PHeonix, but don't know if it is as stiff.

[Blackhawk3D]

Balsaman, would your machine be easily modified to use aluminum window channel and drill rod for the slides? Thanks.

[balsaman]

I guess so, but I am not sure it is the best design if using that hardware. Table is probably too wide for it's depth.

E

[Gary Blaylock]

I am assembling a control board from CNC Hobbies and have a capacitor missing. I have e-mailed Dave twice but haven't received a reply. Maybe he is away this week.

I would like to have the controller ready by this weekend and I am wondering if the capacitors I found at Radio Shack may work.

The capacitor is a .1uF. I found two different capacitors:

part # 272-109A
0.1uF
Ceramic Capacitor
50WVDC maximum
Operating temperture -40C to +85C
Tolerance +80/-20%

part #272-135
0.1uF
Ceramic Disc Capacitor
50WVDC maximum
High K dielectric
Operating temperture -25C to +85C
Tolerance +/-20%
Lead spacing 5mm
Temperture Coefficient Y5U

Does any one know if one of these will work as a replacement?

[gjahnke]

Depending on which capacitor, it would probably work fine. Go with the second one (because of the closer tolerance).

Is .1 uf the value of the original? IF so, a .1 uf ceramic capacitor is a .1 uf ceramic capacitor wherever it comes from.

[Gary Blaylock]

Thanks gjahnke. I will install it tonight.

Gary

[gjahnke]

Gary, how about an update? Did it work? Are your motors spinning?

uscra...check your PM

[Tiramisu]

Hello,

I just completed my router and have a question on KCam. I am having a problem cutting thru the material. I have the Z axis table setup as follows:

Travel: 0
Normal Cut: .50
Deep Cut .50

The Z axis home is .50 above the table but for some reason the Z always stops at .125 above the table no matter what I do. Since this is my first experience at CNC (besides my foam cutter) I am sure its just a simple thing as I do not know what I am doing yet.

Attached is a picture of my machine that I just got running today. Its not quite finished as I need to finish painting and attach the vac table. I looked at Balsamans pictures and followed his ideas (thanks dude!). The tracks are rather cheap Igus Drylin N27 rails. Leadscrews, leadnuts, and controller are from cnclinx.com. So far I am happy with it as it cut a near perfect circle which I guess is a good test.

Anway..any help with setting up the Z on KCam would be greatly appreciated.

Thanks,

MM

[crankorgan]

Zero is home with Z axis all the way up.



Travel is a value above the work but not touching it.

Like -1.25

Normal cut is the depth you want to cut

Like -1.50

Deep Cut is the next cutting depth.

Like -1.75

Also read about G90 absolute coordinates

and G91 relative coordinates

[Tiramisu]

Hello Cranky,

Ok..I dunno why I am having so much issues with this Z stuff..probably too long of a day for me. If is was not for these boards I would be totally lost.

Ok..so home is the top of the Z..I have 4" of total travel but I told KCam I just had .50" for now. So 0.0 should be .50" from the table top to the bottom of the router bit correct?

Ok..now the Travel is some value above the material. So I am guessing this is the distance between the router bit and the top of the material? if so..I am cutting 1/16 balsa sheet so the travel number would be -.4375 and the Normal cut would be -.50?

Thanks for your help on this,

MM

[crankorgan]

Setting both cutting depths to the same value could be causing a programming error. All programs have them. They show up on beginners who put the wrong numbers in certain places. KCam is not a smooth as TurboCNC but you can learn about DXF to GCode converting with it. Offsets and changes can be done in the editor. TurboCNC can only run motors.

[Tiramisu]

Hello,

Well I am definetly a beginner putting in the wrong numbers in certain places. As far as the DXF portion of it I have gotten it too work ok. I drew up the circle in TurboCad and put in the offsets there as per a previous discussion on this thread. I deleted the original drawing lines and just saved the offsets and imported into KCam. I would like to use TurboCNC but I hate DOS and my controller only works with Kcam according to the instructions. So are my numbers that I came up with above correct? I guess I will find out soon enough as I am going to fire it up in awhile...

[crankorgan]

I would keep KCam for plotting and converting. You can reboot Windows 95 and 98 into DOS in order to run TurboCNC. Which controller will only work in Windows?

[Tiramisu]

I have cnclinx.com controller. I noticed Tal does not sell it or have info on his website anymore so I am not sure what the deal is with it. All I remember is that it said it was designed for KCam. I will try it with TurboCNC and see if it responds.

[crankorgan]

There might be on trick! Most controllers need three wires per axis. Ground-Step-Direction. Some use a forth wire to turn the motors on and off. All controllers can do this. You just have to set the correct pin in the software. TURBOCNC can be run in a DOS WINDOW while in WINDOWS. The motors will run rough but you will be able to get it running before you boot up in MSDOS mode.

I build lots of machines. Are you using draw slides of some kind? I have never see they type you are using.
They look expensive.

John

[Tiramisu]

John,

I will try and use TurboCNC and let you know how it works..you are probably right and its just a pin setup issue. I use Vexta motors and they have 6 wires but I noticed KCam only used 3 like you said.

The slides are Igus Drylin N27 type. I paid about $125 for all 3 axis direct from Igus. They seem to work just fine but I had issues with the shuttle mounts and will probably not use them on future projects.

[balsaman]

the cad drawing will likely be at z=0, so you should set Kcam up so a deep cut is 0. Are you importing a dxf or gcode?

make travel at .5" for now.

The Z should travel down from .5 to 0. Make sure you don't have that backwards

[Tiramisu]

Hi Balsaman,

I am importing DXF into KCam. .50 to 0..ok I will try that right now and let you know.

Thanks

[Tiramisu]

Hey,

I got it all too work ok now so thanks for everybodys help. I used .50 travel and 0.0 deep cut. I just have to make sure the Z is at .50 and program is at .50 before I start it.

Now my issue is router cutting speed and bits..I get quite a bit of burning on the bit and tearout on the balsa. This lame trim router I have does not seem to keep a constant speed..might have to switch to a Dremel.

[RJG]

Hello all. I have read this thread from start to end and have been inspired to the point that I am thinking about making a CNC machine too. Very exciting stuff. I have been learning a lot!

Slightly off topic. Some where along the line here someone mentioned making screw threads in wood and pouring thinned down superglue into the hole to foritfy things. I have gone back over the posts and can't find the information. Can anyone explain the technique to me? How do you thin down supper glue? Do you let it dry be fore you put the screw back in the hole? Will this work for hardwood? Will this work better than using those threaded brass inserts?confused:

thanks,

RJG :

[Tiramisu]

Hi,

Just buy Super Thin CA at your hobby shop..it will flow into anything.

[RJG]

So, how does the proceedure work?
1. Thread hole
2. Pour CA into hole
3. Let dry before inserting screw? or insert screw before CA is dry?
4. Can the screw be removed?

[crankorgan]

Tiramisu,
The pins in the software have nothing to do with the number of wires on the motor. If your controller works with KCam it should work with TurboCNC.

John

[crankorgan]

If you draw a DXF in Turbocad and import it into KCam the depths are set by the settings in KCam.

[RJG]

Found the answer on Thread #75, post #1119:

"Pre drill the MDF of course. And a good trick is pour some thin CA into the pilot hole let dry then screw. MDF tends to crack/seperate around a screw".