CNC Cookbook: Blog

After backlash, the second biggest source of frustration seems to be noise problems on homebrew CNC systems. I added a new Cookbook page aimed specifically at gathering up the wisdom I've found for dealing with noise problems.

He's got one with a chuck in it too. These just slide into a regular QCTP toolholder which is then clamped down on it. I've seen some chucks built into the full dovetail toolholder. One nice thing about this approach is you can set the cutting depth relative to the QCTP by sliding this holder and locking it down at the right spot. This would allow a series of tools to be aligned to the same depth and a carriage stop used to make repeatable production with a set of tools easy.

One thing about using the QCTP to hold these is you will have to dial them into the center of the workpiece. Widgit claims this is a more accurate way to drill holes on the lathe assume you do dial everything in. The fast way to get one dialed in is to place a precision pin in your lathe chuck and then clamp the hole or drill chuck from your holder onto it. You can adjust until it's centered just right.

I'm going to be making some drill and chuck holders for my gang tooling plate after I get it made. I probably won't make any QCTP style like these though. The gang tool plate uses much simpler holders that are square throughout and have no dovetails.

Somehow I forgot to mention I'd found these really handy parting off tools on eBay:

Read my mini-review of them on the parting off page. The short story is I love 'em!

Still needs a deburr and polish to finish off, but looks like a workable way to making one of these pulleys.

Using double angular contact bearings to hold at least one end of your ballscrews (the other is often "floating", either free or in a simpler bearing) is key to minimizing backlash. If the ballscrew itself can wiggle axially in the bearings, that translates to backlash. Some guys on CNCZone tracked down a bearing block from Misumi USA that looks pretty good for $70. These are called "BRW". These seem like a deal at 1/3 the cost of equivalent THK blocks. They use ABEC5 bearings, which is 1 step down from ABEC7.

That gives you some idea of what it costs to buy commercial bearing blocks if you don't want to have to make your own. At that price, these are not going to be ultimate high performance. That's less than the cost of a decent set of ABEC7 AC bearings like a VMC would use, but they might be just fine for most hobby machines.

You recall the EZClamps? I was waiting on some endmills to arrive to try them. I got bored this weekend and couldn't wait any longer, so I used a 1/16" endmill instead of a 1/8". This is the second thing I tackled with the mini-router. I didn't know if it would cut aluminum or not, but it seems to do just fine. I wanted some clamps to make it easy to hold stuff on the table, so I whipped up this design in Rhino 3D:

I didn't let it finish, because it would have taken several hours. Feedrate was 9 ipm, but had to be slowed to 5 ipm at the very beginning of each depth pass. Like the turtle, I used OneCNC to generate the G-codes. to hold the aluminum block I just superglued it to a piece of poly board which was then held to the router table with T-nuts. Very cool!

The finish is actually a bit nicer than it looks in the picture. I'm sure there's a lot that could be done through experimentation to improve finish and speed up the overall operation of it. Click here to download a short video clip of the router cutting the aluminum.

Model jet turbines have to be the ultimate machining project. They run at 100K rpms, so tolerances have to be just about perfect or they self-destruct instantly. Here is a gorgeous example on a cool thrust measurement sled:

Commercial manufacturers building these engines for hobbyists are getting $2500-$5000 apiece for them.

A lathe slotting attachment is on my ToDo Wish List, so I perked up when I saw this in the Wankel Engine thread below. To make the Wankel requires broaching a gear into the inside diameter of the rotor: no easy task! This fellow resolved that problem by creating a slotting attachment that moves the existing compound slide on its dovetails (clever!) and that uses a faceplate with holes around the periphery as a dviding head to index to each tooth position. Quite ingenious!

Slotter bolts a bracket in place of the bracket that normally holds the compound leadscrew...

Another view. The indexer for the "dividing head" is bolted where the follower rest would normally go...

All done on a 2D only (Z-axis not CNC'd) little mill. Looks similar in capacity to my CNC Mini-Router!

I came across this one randomly while searching for something else on the HSM boards. I don't know that I would ever need such a thing, but it is really nicely made (as are all of McGyver's projects!):

Just to be sure I keep more irons in the fire than I can handle, I ordered some metal from Speedy Metals today. I sent for a chunk of 11/4" thick cast iron plate to use in my gang tooling project for the CNC lathe. In addition, I ordered a couple of pieces of 1/4" thich aluminum 6061. I intend to use these as carriers to mount components to as I build the electronics enclosure for my IH CNC Mill conversion. Woohoo! Lots to do!

Over time, I have come to the conclusion that real machinists are used to the idea that they have to occassionally fix one of their machines, sometimes even when its new or when others might think it is unacceptible that there is a problem. The real experts seem to just take it in stride and get one with fixing the machine so it can be productive for them again.

At the hobby end of the spectrum, one often reads a lot of belly aching about one thing or another on hobby machine tools. Most of them are Asian, and there are going to be some quality control problems. Many times the machines and accessories are more kits than finished products. The hobbyist looks at this, throws up his hands, and starts to complain bitterly about quality and how they were sold a bill of goods.

I've heard no end of complaints that the "professional" range 12x36 and up Asian lathes can be almost unacceptible in quality unless you buy a Taiwan lathe. One guy goes to great pains to point out every little detail problem with his new Jet 14x40 lathe. I read his article, and he did a fine job polishing up the lathe, but I found absolutely nothing there that would have stopped a good machinist from making good parts with that lathe as it arrived out of the box. The venerable Widgitmaster whom you've heard me speak of many times owns a Birmingham YCL lathe and has nothing but good things to say about it, yet this is one of the more maligned (outside of Harbor Freight and Enco) lathes I've read about!

Let me tell you that after reading thousands and thousands of threads on various boards over the years, I have seen problems no matter which way you turn:

- Want a great piece of Ye Olde American Iron like a Bridgeport? Converting it to CNC? I'll send you a list of stories where this went badly wrong. Everything from worn out ways, bad new new expensive ballscrews, bad spindles, you name it.

- Bought a new VMC? Yeah, happens there too. Comes up on the PM boards all the time. Guys buy expensive machines that show up with problems or develop problems not long thereafter. I don't care what brand you pick, someone has a lemon story to tell you about it. One of the guys that was really vociferous about his IH problems to the point he got kicked off CNCZone still hasn't bought another mill as far as I can see. He wanted a Hurco but had a big fight with the dealer over there.

- Used VMC? Same stories. "It looked great at the dealer, I saw it run under power, I got it home and it is a total disaster." Yada, yada, yada.

- Tormach? There's a big blowup over on their Yahoo board as various owners are discovering that their column, which is pinned at the factory, is out of tram by various amounts. Many of them never checked it but started finding mysterious errors in their work.

Heck, if nothing else you're eventually likely to crash and break the machine anyway, or wear something out that will need a tune up.

Over time I have concluded that part of calling yourself a machinist is being able to diagnose, fix, and work on these machines. Having the ability to do the detective work needed to track down and fix an issue is what really separates the good ones from the also rans. The guys that are happily posting about how they diagnosed and took care of some problem or other always seem to be the same guys doing awesome work in their shops. The guys complaining that the paint doesn't look right on their new Asian lathes? Well, they may be great machinists, or they may just be guys who ought to be painting lathes.

This came up on eBay and looked like a better mousetrap. Works with either the pin type indicator mounting or the dovetails:

I never would have guessed it, but Geof on CNCZone says the following 4-axis mill set up was able to machine these aluminum bars to length, ensure the faces were square, and drill and tap a hole faster than he could do it in a lathe:

Hmmm. I'm guessing the reason for that is that he has to deal with both ends. A basic lathe will do one end great with a bar puller, but then you have to reload each piece by hand to get the other end. This setup gets 8 of them done per setup, which saves a lot of manual intervention. I believe their are lathes with more than one spindle that are designed to deal with the problem by transferring the part to the second spindle. Nevertheless, this is a very cool idea for saving time!

One of the problems a new machinist faces is understanding the variety of products that are similar, but slightly different. This raises the question of whether you need both varieties, when is one better than the other, and so forth. In some cases, it is kind of a brand issue. Interapid indicators currently have the best reputation if you want a super high quality test indicator, especially one that measures tenths. I keep a Tool Brands page with notes I jot down any time I hear something like this about a brand.

In other cases, there is some subtle difference in the two that isn't obvious at first glance, but that makes more sense later. Let's consider angle blocks, which is the subject of this current rant. I bought a couple of precision angle blocks so I'd have them on hand for setups. My two are similar to the two in the back row of this picture:

They're got that "precision" look that a ground finish and lots of drilled holes gives, which is instinctively attractive to the average machinist like myself. There are variations on these, of course:

I would have been tempted to think this might add useful variety to my angle block collection by giving me a larger block (it's about 6" wide) and adding those nice reinforcing webs. We've got a few holes, which look handy for clamping, and though we've lost the nice ground finish, maybe that's okay for some jobs. Probably the last angle blocks I would've considered are these ugly cast iron types that have a "handle":

Heck, there's no holes in these at all, and there's just one web in the middle. Why would I want one of these? Along comes the Widgitmaster with another intriguing setup. He's got to drill a longitudinal hole through a big tall plate of cast iron. How does he set that up? With my least favorite angle block of course:

The angle block is sitting in the Kurt vise, and the workpiece is clamped to it with a Kant-Twist clamp. BTW, I love Kant-Twist clamps and won't touch anything else if I have a Kant-Twist that is the right size. You can see that having the web/handle in the middle made room for the Kant-Twist. I'd have thought a second one on the left would make sense, but maybe Widgit didn't have one. The plate sits firmly on the mill table and is true against the angle block and table. This would be an awkward thing to set up very many other ways. I would probably have tried to stand it up in the vise or just clamp it to one of the other angle block styles sitting on the table. Here we get good support up where the action is, which has to be a good idea.

Apparently this type of angle block is more properly referred to as a "knee block". You learn something new every day.

Thanks Widgitmaster, something else to add to my acquisition list so it'll be there if I ever need it!

This morning I awoke to find a nice couple of posts in the CNCZone Polymer Concrete thread about a fellow who is using EG to stiffen the base and lower column of his existing RF45 (similar to my IH mill) mill. He's using West Marine 105 epoxy, West 206 slow hardener, and a mixture of river rock and sand as the aggregate. He fabricated a welded steel partition from 1/4" plate to manage where the epoxy is going:

Partitioning system is secured by bolts at the front and rear of the base...

Before pouring the epoxy, he also plumbed in a one shot oiling system. Here is the first pour he did to get started:

davo727 plans to pour his base, and the lower part of his column for this project. In addition, he has fabricated a much more solid assembly to bolt the column to the base:

The column support system is also being augmented by a central large diameter bolt that goes from the base all the way up to the top of the epoxy fill in the column. Steel plate above the epoxy and below in the base provide a solid sandwich for the bolt to latch on to. It occurs to me it may be pretty hard to tram in the column with that arrangement as the central bolt will be capable of considerable compression force. As in many things, a little care and time spent should get it done.

This will be quite interesting to watch and hear reports on how successful it has been. It's not really all that much work to do this kind of thing, so if there is a profound benefit it could well be worth it.

This design exercise was inspired by the Polymer Concrete thread on CNCZone. PC is a technique of using epoxy resin with an aggregate added to it to create a material that is not as strong as a traditional machine frame, but that has excellent dampening qualities. There is more than one kind of Polymer Concrete, but one using Epoxy Resin and Granite Aggregate for filler seems to be quite good for these applications.

In any event, I have sketched a table for a gantry-style mill and filled in my thoughts on how to go about building such a thing. As mentioned, this is just a design sketch, but it seems to me it has quite a lot of potential as a way to construct high precision machine tool components without a fancy shop. Here is an image of the finished table:

Whenever the Widgitmaster, a long time CNCZone contributor, has something to say, I'm all ears. I have learned so much from this guy because he takes the time to document his work step by step and he is a professional machinist with many years of experience. The work he produces is gorgeous, and I love the CNC Mini-Router I bought from him.

His latest project involves converting a turret he found on eBay to work on his Birmingham lathe. A turret is a device used in manual work that is similar to toolchangers for CNC. In fact, I suspect one would make an excellent starting point to building a CNC toolchanger. They look like this:

The turret is mocked up on his lathe with a slab of cast iron he is machining to hold it...

The turret is designed to rotate to bring each toolholder into position. The toolholders are held via the 8 T-slots you see machined into the turret. A QCTP is very nice, but imagine being able to bring 8 tools into position just by unlocking the turret, swiveling the right tool into place, and relocking? Widgitmaster is going to be turning out Widgits faster than ever before!

I'm not here to steal the Figiting Widgitmaster's thunder on this project, so you should go visit his thread if you want to learn more about turrets. However, I do want to point out some cool things I've learned reading the thread, and that's what follows with the entries below.

Let's start with that last picture of the turret. Check out the travel marks on the cross slide. Clever idea! I don't know how often I would use them, but it might be real handy to see that you're about to run out of travel soon before it actually happens! It is certainly easy enough to do:

I've mentioned this before, but I always feel it is worth repeating. One of the first things I learned from the Widgitmaster is the need to start out with all surfaces flat and square. This has been crucial for me on many projects such as my Kurt Vise Stop, and whenever I've missed this step it has resulted in problems later on. I've read that many shops keep their old manual mills busy just squaring blocks to feed the CNC machines.

For this project, lining up properly on the cross slide boss was hugely important. To satisfy that need. The Widgitmaster disassembled the cross slide, stuck it on the mill, and then used and indicator in the spindle together with the mill's DRO to take super accurate measurements:

Now he marks the offsets (measured earlier by DRO) as well as the diameter of the bore needed...

This is such a slick technique, I really liked it and would not have thought of it. I would probably have tried to use my surface plate and height gage to measure the boss. The trouble is, I wouldn't have indicated in and found the center of the boss. Instead, I would have been doing arithmetic on the boss diameter and what the height gage measured as the distance to the top edge of the boss measured in X and Y. It would have been close, but I'll bet the Widgitmaster's method is closer and probably even faster to get done properly.

One thing I notice: he is not using softjaws at all with this cast iron, although I usually see his vise set up with one set of jaws or another. I wonder why that is? I will ask!

Every owner of an Asian machine tool complains about lack of rigidity compared to Ye Olde American Iron. Every owner who buys a bigger tool says the smaller tool really lacked rigidity and the difference was immediately obvious with the new tool. But in how many cases is the limiting factor something to do with how the machine was being used? I notice the Widgitmaster goes out of his way to add rigidity in his setups. Most HSM's just throw stuff in the vise and tighten it down hard and assume that's good enough. Let's look at some of Widgitmaster's tricks on this turret project.

1-2-3 blocks add a lot of rigidity to this setup. I kind of wonder why he didn't drag out the 2-4-6's though! Note the jacks under the ends. This may be about rigidity and it may also be about not having the block of cast iron slip downward in the vise. Either way it looks like a good idea! He left 0.010" for finishing after running the 3" face mill. He'll use an end mill to do the finishing job as it will leave a nicer finish than the face mill...

Now he wants to set up 4 jacks under the 4 corners of the cast iron block. He makes sure they're exactly level using this surface gage and DTI along the edge and the mill table as a surface plate. Those 4 jacks are going to lend some beefy support!

I had to throw this one in gratuitously to illustrate the use of a planer gage. Widgit is using a planer gage to measure this distance. He can take a micrometer to the planer gage to see what the value was. I've also seen planer gages set to a particular distance and then used to set something up on that distance. I need to put one of these on my tooling "Want List"! Stuck in the collet is a piece of .7500" hardened drill blank pin. That's probably also something that would be handy to keep around...

Every now and then I read about someone being interested in a ballscrew setup where the ballnut rotates instead of the screw. I thought of these queries when I came across pictures of a Bridgeport BOSS CNC mill's X-axis arrangement:

In the case of this Bridgie, it allowed them to position the drive mechanism where it is well protected under the table and the servo motor and drive remains stationary. Doesn't it seem like it also requires fewer parts (i.e. would be cheaper to manufacture) and is less likely to suffer from backlash? I say this because one simply needs to mount the ends of the ball screw so they can't move. An AC bearing arrangement would hold the ballnut axially. It also seems like a more rigid arrangement to have the ballscrew rigidly held at the ends. Lubrication might be a little trickier as you'd need to lube the screw not the nut.

VFDs or "Variable Frequency Drives" are great gadgets that go between the power source and a motor and allow the motor's speed to be continuously varied. They do this by means of something called "Pulse Width Modulation" where they are essentially varying the frequency of the current going to the motor. Thing of the frequency as governing how many "pushes" the motor gets. 60 Hz or 60 cycles per second is normal. A typical Hitachi 3HP VFD can deliver 0.5 to 360 Hz, which is quite a speed range!

Enter a new and better gadget, which has the really sexy name "Vector Drive." A white paper by Reliance Electric gives the full details, but in a nutshell, these drives are called vector drives because they understand the vector relationship between magnetizing current and torque producing current and they do the right thing with that.

What does doing the "right thing" buy us? Well, several things are better for vector drives than plain old VFD's:

What's the downside? Largely cost. An equivalent vector drive probably costs 20 to 30% more than a VFD of the same horsepower capacity and it will require you to install a tachometer input to give it feedback on what the spindle is doing.

Most small mill and knee mill owners are used to R8 tooling which uses a thread drawbar to pull the holder up into the taper as the threads are tightened. Modern CNC mills use retention knobs and tension generated by Belleville washers to apply 1800 to 2500 lbs of force to pull the holder up into the taper. An air operated cylinder pushes down on the drawbar to compress the springs and allow removal. The whole assembly looks about like this on a Haas machine:

The retention ball or pull stud is held by a set of ball bearings. When the compression on the drawbar is released, the whole assembly moves up and a tapered bore causes the ball bearings to grip the stud tightly. Nifty, eh?

This is just a little sketch I did for the Widgitmaster who is worried about the impact of dust and contamination on his mini-routers:

No sooner had I written my notes below about Face Gear Rotary Tables than I found out about another fellow's rotary air indexer project over on CNCZone. ServoWizard is a 40 year veteran machinist and it looks like a bit of an inventor too. He built this indexer to make pulleys for high performance go karts:

They're really nice made, complex CNC parts. The number of teeth are engraved on the side of each pulley, a nice touch! He's using a CNC converted Bridgeport with a controller and software he wrote as his first programming project. Here is the air indexer itself:

The indexer uses two air cylinders and an electric solenoid valve for control. A larger air cylinder locks the pin into the indexing wheel. The smaller cylinder, visible below, actuates a sprag clutch to advance the wheel to the next positon. Sprag clutches are gizmos that look a bit like a roller bearing that have low friction in one direction of rotation and high friction in another. This gives the air cylinder a ratcheting ability to turn the indexer. An adjustable travel slide allows the stroke length to be setup for different index wheels so that the right index slot is positioned:

While Googling to try to determine how pneumatic indexing tables work, I came across a description of face gear rotary tables. These tables are extremely rigid, but they are limited to stopping positions at fixed locations on the circle, such as every 5 degrees. They are often pneumatic and operated simply by having the CNC spindle push down on a switch atop the table (with the spindle not rotating of course!). They are most useful for indexing work to fixed positions, for example to expose a new side or even create a horizontal tombstone arrange to increase the parts capacity of the machine:

The face gear mechanism works by meshing two face gears using axial movement. As the gears move closer to one another they are forced to engage precisely:

- Rotate to approximately the right position. "Approximately" means close enough that when the face gears are locked up again the teeth will force the final correction to make the position exactly right.

- Lock the face gears by moving them together. This will nudget the shaft into the precise desired position and once locked will keep it there with great rigidity.

The big advantage of these face gear mechanisms is rigidity. The disadvantage is they only move in fixed increments.

Here is a spin indexer converted for use as a CNC 4th axis using a step motor and right angle drive:

Indexers like these have very limited rigidity, especially if you are machining near the edge of the chuck which puts leverage to work against the rigidity of the indexer. The usual answer to this problem in commercial gear is to provide a pneumatic friction lock to help lock the axis, but this is still not as rigid as a face gear system.

I like that idea a lot, especially if you are using Asian machine tools of unknown trueness like I am. I wouldn't say it's something to do every time you tram the mill, but doing it at least once to see how things are performing seems like a good idea.

The second idea concerns making special tram goodies that may have other purposes:

Somewhere I came across Bottle Bob's site and saw this very nicely made copy of the SPI "Zero-It" indicator sweeper:

He says he leaves it set up in a 40 taper holder and uses it more than any other indicator holder he owns.

How do you close the diameter of the hole slightly to hold a ball detent in place? The operation is called "staking" and can either be done with a special tool as pictured or with a big ball bearing:

Cecil's 1/2 Scale Ma Deuce. Working guts are Ruger 10/22. He even made the scale ammo can!

Gun was made from CCS Prints plans. They also offer Browing 1917 and 1919 based on Ruger 10/22...

One thing I don't have yet for the mini-router is any kind of a clamping kit. Widgitmaster includes some T-Nuts with the router, but there are no clamps to go with them. I just used some bolts to do my first project, but something more convenient would be called for. Enter the EZ Clamp:

I've seen this style of clamp before and thought they would be quite handy as they do not require step blocks. I'm not sure what they're really called, so I just named mine "EZ-clamps". This is the sort of thing that is ideal for me to do with CNC. It would be hard for me to machine them manually, yet it seems straightforward to do with CNC.

I intend to make them out of 1/2" aluminum. It will be interesting to see whether the mini-router is powerful enough to make them. I'll be using a 1/8" end mill to do the cutting. I plan to clamp the aluminum atop some poly cutting board and have at it using relatively gentle feeds and depth of cut. The g-code program is straightforward to create using OneCNC. All I need now is for my end mills to arrive in the mail: eBay purchase.

PC motherboard form factors seem to get smaller and smaller. How about the VIA EPIA motherboards? Here is the current highest performance model, with a clock speed of 1.5 GHz:

The biggest issue is you've got LAN and USB, but no parallel port. That means running something other than the basic parallel breakout board with Mach 3, such as a GRex or ncPod. Still, this card is a mere 6.7" x 6.7". This form factor is known as "Mini-ITX".

This one has that essential parallel port for Mach 3, and it is still tiny at 9.6" x 8". It's very similar to the A8V-MX I used to create my rack mounted CNC lathe PC.

It really would be interesting some day to see just how small an enclosure could be built to house a PC, suitable breakout controller, 3 or 4 Gecko drives, and power supplies for the PC and step motors. My guess is it could be made very small and probably would cost about $1000 in components.

Speaking of small form factor PC's and my rack mounted CNC lathe PC, I had to go looking for a parallel printer ribbon cable. I need this to be able to run my Xylotex-based CNC mini-router off the same PC. The problem with the normal round parallel cables is they're too bulky for my tiny little case. I can't get the cable in and out without it running into the monitor support on the rack: Doh!

For the time being, I'm just running with the PC enclosure kicked forward out of the rack about 1". However, the ribbon cable style are much more compact and can fit into tighter spaces. It took me a fair bit of sniffing around, but I finally found some 4 foot cables at Jameco. These would be ideal for putting all of the components into one box I think. I used to buy a lot of electronics components from them years ago as they were one of the first surplus electronics outfits in Silicon Valley selling digital parts. It was fun to get reacquainted.

I would've made my own cables, but the parts added up to more than Jameco was charging and I have no idea when I'll need another one.

The venerable Widgitmaster has come through once again with a description of how he is reverse engineering a Hardinge Turret to fit his Asian lathe. He needs to make a custom mounting plate, and in order to do so, he wanted an accurate drawing of how the existing turret plate expects to be mounted. So, he clamped the base plate to an angle plate, set the result up on his granite surface plate, and started taking measurements with a height gage:

By clamping to an angle plate, you can flip the item 90 degrees and develop a true set of X, Y coordinates for various features simply by measuring their heights. You'd have to clamp a little differently than Widget shows here, but you get the idea. Now taking that list of coordinates, the Widgitmaster was able to produce a drawing on graph paper:

It would be super easy to take a list of coordinates and rapidly produce a Rhino 3D or other CAD drawing. I'll have to remember this the next time I am trying to reverse engineer something.

This evening after work I was able to do the last little bit I needed to run my first CNC part program to completion and cut some real chips. Here's a nice picture or two of what the result was:

As you can see I cut a couple of Hawaiian Turtle Glyphs into a block of wood scrap using my Widgitmaster CNC mini-router. Full details are on the mini-router page, but I can tell you it was great fun to finally get there and run a program!

So how do you etch glass with a CNC router? The trick is to use diamond burr bits ($6 a set at harbor freight) and cut in thousandths. Also it can't be tempered glass. Wow, that's something I'll have to try for myself sometime!

I woke up this morning and couldn't stand the thought that the only thing stopping me from running the mini-router was a set of couplers for the step motors. So, I headed downstairs to the shop, fired up my lathe, and whipped up 3 couplers. They're just cylinders with a 1/4" ID hole and two 1/4-20 set screws. They seem to be working out great.

It took me a little bit of effort to get Mach 3 all calibrated and running properly, but the tutorial on setting up Mach worked out just fine. The next thing I did was to fire up my copy of OneCNC Mill Advantage XR2 and gen up some g-codes. I decided on a simple logo, the word "CNC" in 1" high letters. The engraving toolpath in OneCNC will just run the cutter around the outline of the letters. I had to fool with it a bit, the Mach 3 post that came out of the box with OneCNC wasn't right somehow (it threw off big arcs for some reason), so I downloaded the latest post and gave that a try. Sure enough, all was well. It takes about 750 lines of g-code to do that simple logo, but it was great fun watching the router cut air.

Now I need to cut the logo for real. The main thing holding me back is I don't have a computer in the shop yet, and I don't want to make a big mess in the office with the chips flying. I'll have to think on that a bit.

There has been some progress on mounting the electronics in the toolbox. All I need now is a set of couplers and I can't mount the motors and the mini-router will be ready to start cutting! I also started a CNC Mini-Router page to chronicle the project in one place, and eventually to walk through some of the things I make with the mini-router.

This does not look at all hard to make, but if you'd rather buy one, Travers Tool has the clamp, $77.32 each, part no. 61-171-001.

Hooks to the column, swings out of the way when not needed, and has a quick acting cam action.

I love the magnetic base halogen lights I bought from eBay seller 800watt. I use them on my mill and lathe and they sure to help me to see what's going on with the work. I came across this unique light on the PM boards, and was impressed. I don't know when I'll get around to building one, but it seems to be well worth a look. Here are some photos:

This fellow cannibalized a set of holiday lights he got for $12 to harvest the LED's. The PC Board wires 2 LED's in siries and then hooks each pair in parallel 22 times, for a total of 44 LEDs. The unit is powered by a 4.7v phone charger. Someone on the thread suggested that each pair of series LEDs ought to have a 1/8 watt 270 ohm surface mount resistor in series because in the current design, if an LED shorts out it could take a bunch of others with it.

Because the light surrounds the spindle at close range, you get a lot of illumination with minimal shadows, which seems ideal for mill work. Lights similar to this one are available on eBay for use with microscopes in the $60 to $90 range.

One of the many (too many!) projects I've got running along is to make my Widgitmaster CNC mini-router operational. It's an incredibly cute little machine that just needs to have some step motors bolted up and connected through a parallel port to be operational. Towards that end, I decided to use a small Craftsman toolbox to hold the electronics:

I figure its about the same size as the mini-router, its cute, and I can throw a few accessories in the tray if I want to take the mini-router somewhere to show it off. I had also ordered the complete 3-axis kit from Xylotex, which has all the electronics and step motors needed to get going:

Having decided to put the electronics into the toolbox, my next thought was to create some sort of a mounting plate on which to attach the power supply and the Xylotex board. My original thought was a piece of aluminum, but I happened to spy a half forgotten package sitting in a corner of my office that needed to be dragged down to the shop. Based on a write up on the Industrial Hobbies site called "Saving Your Table", I had ordered some cheap poly cutting boards off the Internet. I got like 6 boards for $24 or something. This material looked like a good thing to make my mounting plate out of, and it would be fun to work with a new material. I had never messed with plastics much save for some experiments with plexiglass associated with my old PC modding activities--pre-machineshop to be sure!

The polyboard worked out well. It mills beautifully, without melting onto the endmills. It drills and countersinks as well. I set the board up with holes drilled for power supply, Xylotex board, and 4 mounting standoffs. The latter I turned from some "mystery metal" that I thought was 12L14, but it became obvious it was something tougher once I got to working with it. Here is what the board wound up looking like before I mounted anything on it or cleaned it up. You can see the standoffs underneath:

Top of board. Stand-offs are mounted using countersunk 1/4-20 flathead cap screws. The little cutout at lower left is a handle cutout--this piece of poly was originally a kitchen cutting board!

The standoffs were just turned quickly on my lathe. I got them to approximately the same length, faced them, drilled them for a 1/4-20 tap, and the ran the parting off tool. That last step was my first hint they weren't 12L14 as they chattered something fierce!

Stand-off. The gunk is some tape adhesive. When I finish I will clean everything up with brake cleaner so you won't see that stuff...

This was my first experience with countersunk flat head cap screws, and I have to say, I like them better than the socket head cap screws I had been using for everything. I ordered several boxes of them with my last Enco order, so you'll see them cropping up more often in my projects. FWIW, when you countersink, you just need to countersink to a depth of 1/2 the diameter of the flat head. You can take my tip and use a spot drill to do the countersinking and combine the spotting and countersinking chores in one operation too.

Next it was time to tap the standoffs. I decided to try power tapping, which I had read about in various places. Basically, I stuck the tap in an Albrecht chuck on the mill, dialed in the slowest speed, made sure the quill could feed freely, and loosley feed it in. The first one worked pretty well, albeit taking several tries and backing off. The tap seemed to be having some difficulty cutting, but I assumed it was my inexperience at power tapping. I did not put together the chattering cutoffs with the difficult tapping to realize I had something other than 12L14 quite yet.

The second standoff was a write-off--the tap broke off before I was a quarter of the way into the hole. Doh! ?!@#$%!!! It was getting late, so I shut down the shop and went to bed. The next morning, as I was turning and parting a new stand off to try again with, it dawned on me that what I was cutting was something a lot tougher than 12L14. I decided to leave the power tapping aside and hand tap in the interest of moving through the project and because I wasn't too sure what I was dealing with.

I also took the opportunity to try some upgraded taps I had bought from Enco along with the flat head cap screws:

Left to right: 3 flute 45 degree spiral flute plug tap, thread forming plug tap, thread forming bottoming tap...

Having broken my Craftsman tap, it seemed as good a time as any to try out these new "professional" taps. Like me, I am sure you've read in many places that hardware store taps are junk and shouldn't be used. Perhaps you are also like me in thinking, "Yeah, but hardware store taps are what I've got and they've worked for me so far." In fairness, I did buy the new taps with a thought I would try them and see how much better they are. I just hand't gotten around to it until the mystery steel and broken tap forced my hand.

Since I was dealing with brand new (and somewhat expensive) taps, I decided not to try power tapping. Too much experimentation can be a bad thing for progress. So I had 3 spacers to thread, and 3 new taps to try. I stuck each spacer into my Kurt vise and locked it down against a V-block and then went at it.

I first tried the bottoming thread forming tap. These are usually recommended for aluminum and mild steel, but I thought it worth a try here too from a learning standpoint if nothing else. I found it pretty tough going, but manageable. I used Tap Magic with all 3 of these taps.

Next I tried the thread forming plug tap. It was easier, but still took a fair amount of effort. Evidently the tapered end gives things an easier start compared to the bottoming tap.

Lastly, I fired up the spiral point tap. It's a Cleveland 45 degree spiral flute tap. This was the hands down winner by far! Turning effort was the lowest I've ever felt while tapping. Control was excellent. The 3 flutes sent up some wicked steel splinters out the top of hole, so you'd want to be careful around those. All in all, I was amazed at how much better this tap worked than anything I have used in the past. Cost from Enco was $8.86. You can bet I'll be on the lookout for more of these and will want to eventually round up a full set. I wanted to try power tapping these because it was so easy to do by hand, but I had finished all the spacers, so that must wait for another day.

Kinda makes the .50 cal rifles look lightweight, doesn't it? It's a Lahti 20mm semi-automatic anti-tank rifle from WWII. I would imagine they are thoroughly illegal in the US (and most everywhere else) unless you have some ridiculously impossible to obtain permits. In my opinion, they ought to make them legal everywhere. The darned thing is so impossible to hide and so incredibly expensive to shoot I'm sure it would accomplish all the ends the gun control world would like to see.

I'm always fascinated to read the stories of folks who start from nothing, often in their garages, and manage to build up an interesting manufacturing business. Recently, I came across this fellow, who frequents the boards as "2kjettaguy" and has a business called "42DraftDesigns" that makes aftermarket accessories for Volkswagens and Audis. I own an Audi TT that many of their products would work with, cool!

At this stage, his 2500 sq ft shop includes welding, 2 Sharp 2412 VMC's, a Jet 14-40 lathe, an injection molder, and it looks like a knee mill of some kind. There is a staff of 6 running the machines. Here is a view of it:

One of the two Sharps is running 2 vises, and the other runs a fixturing plate made of 1" 6061. He has a number of fixtures that drop down on the plate to facilitate manufacturing his products. He uses a CAM program his father (who is a programmer) developed. Look how clean everything is, just like the other pro shops you see. In some cases the VMC's are used to manufacture, in others, they are used to prototype. For example, he makes high performance exhaust parts, and will prototype the flanges, weld up his test pieces, and if he likes the result, the flanges are then farmed out to a laser cutter. He still welds the assemblies in house, however.

He started first with a little MaxNC mill and then moved to a Bridgeport Series 1 CNC with retrofit controls he installed:

And here is a piece he did that I thought was cool. It is for a friend, not his company:

I came across this while researching the write up on 42DraftDesigns and found it interesting. He needed some custom extrusions for a speed part he wants to manufacture that look like this:

In any event, a firm called Minalex quoted $1100 for a custom die and then $4.63/lb for the finished extrusions up to 100 feet. That gives a ballpart for what custom aluminum extrusions cost.

This makes sense, as changing tools is a mindless task that shop with few employees ought not to be engaging in. In real world terms, a VMC ought to be at least 4x faster than a CNC'd knee mill. That's why there's so many used knee mills out there for $10,000 or less even though they still have life left in them.

It seems to me that this toolchange overhead problem is much worse on lathe work. I find I am changing tools on the lathe much more often than on the mill. I can get a lot done with one endmill until I need to start drilling or something. This has gotten me interested in gang tooling for lathes, which looks to me like the easy way to solve that problem. Omni-turn's catalog provides a lot of great pictures and ideas around gang tooling lathes.

Coming up with a cheap and easy toolchanger for the mill promises to be more challenging!

Another good suggestion from that same thread is to do a different setup on each vise on the table. Why? Because it will uncover any problem sooner before you've run all the parts through the bad setup or program.

Ballscrew mounting bearing blocks commonly employ preloaded angular contact bearings to reduce backlash. They securely hold the leadscrew (a ballscrew in this case) so that it can't move in and out along its axis. Recently I read an account where someone greatly reduced the backlash associated with the ACME leadscrews in their mill by forcing a little preload in the mounting bearings. The collars that hold the leadscrew were simply pinned to the leadscrew shaft. He found that by removing the pins and placing a bolt in the end of the shaft with a pressure washer, one could dial in preload to reduce the backlash. In his case, he reduced it from 0.012" to 0.0025", which is actually pretty darned good for an ACME leadscrew on an Asian mill.

Now the downside is that the bearings these mills use are not angular contact bearings and aren't really designed to take a lot of preload. Too much force can ruin them completely, so go easy on them. Even modest forces will likely shorten their life, so you may want to use the newfound precision to machine a custom bearing block capable of holding some real angular contact bearings.

I am always inspired when I read that someone went into the guts of their machine and improved it with a simple fix like this. My reaction is always that here is a person who really knows how to think about machinery and understand what's really going on. Most of us would take the darned thing apart, fail to find anything that looked amiss, bolt it back together and give up on improving it. A real machinist or engineer is methodical and has an intuition about these things.

Cutting tapers is seemingly a daunting process on a manual lathe and a non-event for the CNC crowd. For this reason, there are a number of purpose-builtattachments for manual lathes to make the job easier. Typically they involve a device mounted on the back of the lathe that involves setting a bar at the proper taper angle and then using that bar as a rail to move the cross slide at the same angle. Such devices are available commercially (for $700 or more) or they can be readily constructed in the shop.

An alternate approach involves attachments for the tailstock that place the center, ummm, off center. Recall that the pain of setting the tailstock over and then getting it back into proper alignment is one reason people want the attachments. It is therefore ironic that one can use an attachment to accomplish the same setover. Traditionally I have seen these done by adapting a boring head to fit the tailstock. This is a pretty easy approach as the boring head will already have a nice dovetail and adjustment screw.

I thought this was a very nicely made little gadget that might make a good model if one wanted to build such a thing. There are several interesting features to note. First, take note that the live center actually has a little ball on the end, not a point. There is a discussion of why the ball is necessary in the PM article I linked above, so take a look if you are curious. Second, there is an integrated level, because of course it matters how the offset is achieved. Lastly is a micrometer dial to perform the fine adjustment together with a scale on top that shows the grosser degree of the adjustment. Again, this is very nicely done.

Unfortunately, like so much else from the manual machining world, these gadgets are no longer being made as there is no longer enough demand for them.

This hadn't occurred to me until I was looking at a gang tooled lathe and noticed they offer this option. With CNC, you can arrange your programming relative to where you will part off, and this allows the parting off tool to be in a fixed location. Given the desireability of mounting the tool upside down and behind the workpiece for smaller lathes, it makes even more sense. This particular commercial machine has it as a $3000 option, which seems a bit much. All that is required is an attachment to the lathe bed in a fixed location with a stepper motor and leadscrew that will feed the cutoff tool into the work. Ideally you want to approach from an angle 90 degrees off the plane of the gang tooling so as not to get in the way of that tooling. From that perspective, it might make the most sense to have the parting tool descend from above.

I don't know what I will ever do practically with this idea, but it was interesting to come across. I think I would mount a tool setting sensor to this attachment if I were to create one. In practice, one would write the g-codes so they're referenced entirely against the parting off position, which would be very near the chuck jaws. One could write the g-code program to position a stop mounted on the gang plate to allow the workpiece in the chuck to be properly positioned before the chuck is tightened down. Hit the button to proceed and your workpiece would be properly indexed relative to its length and chuck position, the tool would then touch off the sensor on the parting off attachment, and off you would go. If one insisted on mounting the attachment in the same plane as the gang tooling, perhaps it can be behind the chuck so as to be to the left of the gang tooling and in a place where that tooling likely wouldn't go too often. Or perhaps the retract travel would be enough to get it out of the way.

The attachment in the link I have provided is mounted to the headstock, for whatever that is worth.

Gang tooling is just a way of using an extended lathe slide to hold a lot of tools. The gang tooling assumes no tailstock, and so moves to the right before positioning a new tool so that all of the tools clear the workpiece. Once the new tool is positioned, the gang tooling moves back in to bring the tool to bear. It's a very simple way to get the benefits of a toolchanger for a few tools without needing a lot of expensive mechanical gadgetry. Omni-turn's catalog provides a lot of great pictures and ideas around gang tooling lathes. Here are just a few things I learned:

If you use your lathe tailstock with a drill chuck very often you will appreciate this sensitive feed attachement:

It's very nicely made, isn't it? One thing I might look to add is a digital caliper that shows depth. There shuold be a way to rig that up. In addition, this attachment only has 1/2" of travel, and more would be nice.

Shaft is in rotary table and tail stock. The little digital controller is nifty!

I've had my eye on building some machinery dollies for some time. Since I want to purchase some machinery, the time has come to move ahead on these. I purchased a set of casters, and will shortly go pick up a load of square tubing to build this:

Its sort of a cross between a two wheel dolly and a toe jack. That blue cylinder in the middle represents a bottle jack. One would sidle one of these dollies up to either side of a machine or crate, connect the two together with load straps and perhaps some more tubing, give a couple of pumps on the jacks, and voila, the machine or crate has sprouted wheels. You gingerly roll it off to its new position and then release the jacks, loosen the load straps, pull the dollies out on either side, and you are done with the move.

Some folks have to get by using their lathe as a mill. It is amazing what can actually be done in this fashion. How about this nifty way to face the end of a big block with a fly cutter in the 4-jaw:

Actually, just some Rhino 3D doodles. A friend wants to get some Mauser actions and make some rifles. Naturally, I'd like to do the stock using CNC and a router. Here is what I came up with:

It's not quite right, but it isn't bad for a first attempt. I think it could probably be tweaked into decent shape with some more work.

I just found this handy reference on the web. It's Niagara Cutter's recommendations for milling cutters and their use. Note that these are no end mills, but rather cutters such as slitting saws. I've actually never used this style, but I have some (another cheap eBay purchase) in the toolbox patiently awaiting the need. When I get around to it, I'll pull this reference out and figure out how they're supposed to work.

Sorry for the long hiatus on the blog, but I've been travelling on business. I came back to hear that the Fidgeting Widgitmaster, whom I've written about often in the past, is selling his very first CNC router that he built. It's a beauty! He published a link to his build log photos, and I happened across a really neat idea for integrating limit switches and protecting them from debris:

The design is not unlike the optical limits available from places like Industrial Hobbies, but is based on a mechanical microswitch. It's really elegant. Looks like a little bearing on the end of the arm to keep things rolling smoothly, a couple return springs, and I can't quite make out what trips the limit, but presumably it will trip in either direction, so there must be a couple of protrusions on either side of the microswitch. Inside this cavity everything is protected from dust and other contamination. The ballscrew even has some sort of wiper arrangement. Read through the rest of the build log, this machine is a real work of art. Shows you how a pro approaches machine design!

I came across this photo essay of a tour of the Sieg machine tool factory in China by the Littlemachineshop.com. This sort of thing is always interesting. One interesting thing was how much actual inspection was going on. I think we sometimes feel the Asian machines are assembled and shipped without any quality control, but these pictures would indicate otherwise. The second thing to note is that its nearly all manual. They own a single CNC vertical machining center--no mention of what it is used for. My guess would be making molds for the castings. The rest is done on manual machines, including machines we consider antiquated such as scrapers. Here is a surface grinder being used to grind the dovetails for a mill bed:

I'm adding scissor knurlers to my list of todos. I'm still using the old style, but they're not the greatest. There is a thread about these over on HSM as we speak with a couple of designs. One is very elegant, and the other looks dead easy to build:

I liked this description from the HSM board of how to install a key on a the base of a milling machine vise so that it will always be in tram on the table:

Another thought, if you don't want to modify the vise, is to make a tramming fixture that you clamp in the vise jaws. Said fixture would mate with the T-slot precisely enough to ensure tram as the vise is being bolted down.

It's a manual bed millthat looks to be very sturdy. I've never really liked knee mills of the Bridgeport persuasion as well as bed mills, though they are far more plentiful in the used and Asian clone markets. I thought this machine was an interesting alternative.

I've been making steady progress, still not done, but there is more to look at. The rack is fabricated, save for drilling mounting holes for the driver electronics rack mount enclosure. You can see the swing arm on top, with keyboard tray and LCD touch panel. The lathe's control panel is just resting there, but it will ultimately have a holster there. You'll be able to grab it and drag it over close to the lathe as a pendant or operate it right where it sits.

A chap named Archie recently dropped me a line and turned me on to what looks a very useful piece of gear for moving heavy machines around the shop: the Rol-A-Lift:

I understand they're available for rental from various places. Seems like a better deal than toe jacks, although I could imagine machines with overhangs that you couldn't get this thing under properly. A pair of these with a lift strap between them, a couple of pumps on the jacks, and you are ready to move your machine. Pretty cool, eh?

Most of the cost associated with little things is shipping and handling. Even if you go to the store, the cost of your effort to take time away from your shop and go to the hardware store to buy a bolt or a nut dwarfs the cost of the nut or bolt in most cases. For that reason, I always buy more than I need, and I organize my storage of these little bits so that maybe the next time there will be one on the shelf. Rushing to the store to buy a tap? Buy 2. Need a 1/4-20 socket head cap screw of a particular length? Buy 4 in that length and 4 more in the longer and shorter lengths. I've been doing this for quite some time, and it pays off handsomely when you find yourself able to do more in you shop without constantly running off to the hardware store or ordering on the Internet.

I'm trying to finish off the electronics enclosures for my CNC Lathe conversion, and needed strain reliefs for the power cord. I got 600 (!) of them on eBay for $1.99 + 10.99 shipping and handling. When they get here, I will never need to buy a strain relief again. Now I could've run over to Radio Shack, dinked around, possibly been routed from the closer store to the bigger store as happens, and had the darned thing tonight for an investment of about 1 hour's time. I'd rather just order it, it will get here, and I won't have to deal with acquiring them again.

I organize all these little goodies in multi-compartment plastic boxes which are then stored in a big storage rack I built:

Label the plastic parts organizers on the top and side. I've taken to keeping the taps and dies with the appropriate hardware box so its all right there when I start messing with 1/4-20 hardware.

When I order from Enco and other suppliers, I try to avoid onesy twoseys. If I need one 1/2" endmill, I probably could use two or three, so I order two or three. This is especially helpful with Enco, where I always try to reach the $50 minimum to use their free shipping offer anyway.

Sometimes, I have gone on the hunt to jump start my collection. For example, large hardware assortments are available on eBay relatively cheaply. You can buy assortments of things like O-rings, key stock, and bronze bushings which are handy to have around. At Christmas and my birthday, I always put a DIY hardware assortment on my list. I give people a general description, like "1/4-20 hardware", and then tell them to go buy 2 packages of every size bolt they have, a set of nylock nuts, some regular nuts, some lock washers, and some regular washers. Throw in a plastic storage box for them, and you have a cheap gift that I am delighted to recieve.

Once you have a well-stocked small parts larder, you will find it is incredibly useful and increases your shop productivity!

I had an inspiration during my commute today. I was pondering conversion of a rotary table for use as a 4th axis on a CNC mill. The conversion is relatively straightforward and has been documented in a number of places. One thing that commercial units often have is a lock that clamps the axis when the rotation has completed before machining for greater rigidity. What occured to me is that one could adapt an automotive disc brake to this purpose. Why not machine the rotor to replace the table of the rotab and mount the caliper to allow the axis to be locked for machining?

Being able to lock the axis might be particular important in dealing with the inevitable backlash of a standard rotary table conversion.

Reading about this task on CNCZone I learned of a handy shortcut. One fellow says, "Forget the countersink, just do it with your spot drill up front, then drill the hole, and you are done." "Brilliant!", says I. Rather than spot drill, drill, and then countersink, let's kill 2 birds with one stone. The one fly in the ointment is that the countersink spec for US flat head cap screws calls for an 82 degree countersink, whereas spot drills are 90 degrees. This immediately led me to produce the following little chart that lays it all out:

First we have the size of the bolt, followed by the expected diameter of the head according to a manufacturer on the web. We can apply a little bit of geometery and realize that for a 90 degree angle, the cone of the cap screw is the hypoteneuse of a right isosceles triangle and 1/2 the head diameter is therefore the depth. Now a little trig and we can figure the error between the 82 degree countersink and our 90 degree spot drill's countersinking. As you can see, it isn't miniscule, but I doubt it is the end of the world either.

Note that 82 degree spotting drills are also available, though in fewer sizes. They will be spot on with no error. As far as I can see on Enco, 82 degree and 90 degree spotting drills cost the same. You will have to buy a pretty good sized diameter spotting drill given that you need one at least equal to the head diameter of the flat head cap screw you end to spot and countersink with it. A 3/4" diameter spotting drill is only good for 5/16" and smaller screws. These spotting drills are not cheap, and cost about 2x what a countersink does, so you'd better be able to justify that with the time savings. I think I would learn to sharpen drill bits in short order if I was planning on doing very much of it.

I have ordered up some flat head cap screws from Enco (having needed them for another project anyway), but I can see I now need to order some spotting drills so I can experiment with this!

I am focused on building the enclosures for the electronics at the moment. I'm creating a self-contained rack that will have one enclosure for the PC, and another for the GRex, DC supply, and Gecko drivers. The rack will incorporate a swing arm with keyboard tray, monitor mount, and a place to hang my lathe's control panel. It is fabricated of rectangular tubing. I'm further along than this, but here is a teaser pic:

Swing-arm goes on the pivot shaft at the rear. That's the PC enclosure sitting there. More to come!

As part of the process of building my enclosures, I have to take apart the working step motor driver enclosure. I want to drill the rear panel for every concievable connection I made need in the future so that once I have it bolted back together, adding a new feature will not require any chassis modification, it simply requires running the wiring to the GRex. Towards that end, I bought some cable marking products from CableOrganizer.com to make it easier to keep up with the wiring:

Label machine and the two cable marker dispensers. Middle one has labels you can write on, lefthand one has digits...

Here are the step motor cables. I labeled the axes on each cable. The numbers tell me which terminal number on the Gecko drive the lead goes to.

The first time I heard ships could be made from concrete I have to admit I was surprised. This was done in World War 2, though I am not sure how often since. There is one partially sunk on the beach near where I live. So if you can build ships from concrete, why not CNC machines? Before we go further, here is a example from a company called CNCBridges (bridges are made of concrete too, eh?) of an expansion kit for the popular Sieg X2 machines that is built of something called polymer concrete:

One of the first things brought up is that we are looking at composite materials, and that for this application, a composite of epoxy and granite is one of the best. Polyester (fiberglass) resin is considered obsolete for this application because it shrinks while curing. Epoxy supposedly has zero shrinkage, which is ideal. There are a lot of approaches to using the stuff. The pictures above have it case into the desired shapes for the machine. Another possibility has it filling a welded frame to damp resonances and increase rigidity.

There are a some higher end companies in this business, ITW Polymer Castings and Accures Casting. Accures has a "starter kit" that does 3 cubic feet (a weight of over 400 lbs) for $340. That would be an interesting place to start, giving they have probably worked out a lot of bugs in the process. Check some of the properties of their material: it damps 10 times better than cast iron and 30 times better than aluminum, for example. Precision Epoxy makes flooring products for the motorsports world. Check out their epoxy surface plat product. Very interesting. Claimed to be 0.005" flat just by pouring! They're using it for precision chassis set up on racing cars. Penske, AJ Foyt, and others use them. I guess that would require a big surface plate!

An article in Machine Design shows machines being cast from this stuff with accuracies claimed to be 0.001"! Light Machines has been building small mills out of this stuff for about 20 years. They use the ITW "Zanite" product.

For those who are wondering why ordinary concrete cannot be used instead of epoxy, the issue seems to boil down to shrinkage, extremely long curing times (months!), and water which may lead to corrosion of the metal components. Interestingly, Bamber, the fellow whose thesis Principles of Rapid Machine Design, I learned a lot from, uses ordinary concrete and rebar to damp the frame on the machine he built, and it produced excellent results. I wouldn't rule it out, but the epoxy looks set to produce a superior result at a higher cost.

Getting the exact composition right is interesting. You don't want too much epoxy, or you are building a plastic machine, not a composite machine. Apparently it takes a variety of sizes of media ranging from very small 20-200 micron size up to fish tank gravel (BB sized) pieces of granite or sand. I find myself wondering about bead blast media as a component when I read this?

Steel inserts are cast into place to provide threaded hole locations. Hexagonal cross section inserts are made that have a recess in the center to provide strength against pull out. They should not be placed closer than one diameter from the edge. One challenge for the DIY crowd is that achieving high accuracies means using very accurate molds.

One outfit trying to build a commercial machine weighed in with some tips. Here is a shot of the rail mounting on their small mill:

Note the rails have a precision guide edge modled into the composite base. The base itself is L-shaped to support the table and the Z-Axis. So the Z and Y rails are bolted to the polymer base. Here are some of their tips:

Raw epoxy is solvent free and ships non hazmat.
Cheap reactive dilutents are used to thin the epoxy and thin the price.They also thin the properties.Ask your supplier if their epoxy contains Nonyl Phenol.
If your supplyer cannot ship non hazmat the epoxy has dangerous additaves.
Raw, neet epoxy has low vapour pressure{smell is not bad]
Reactive dilutents have very strong odour,might i add unbearable.Again if it ships non hazmat it is probably OK epoxy.To confirm request MSDS from the supplier.
We need thin epoxy to maximize filler loading.This would be called Laminating epoxy.Thinnest ? Probably 600cps which would be the consistancy of #one Canadian maple syrup.Suddenly I feel like bacon& pancakes.
Epoxy glue is 10,000cps or better.No use to us granite guys.Similar to honey.
We want a long pot life to maximize mixing time and reduce heat build up.Many times we have mixed or de gassed for too long and the epoxy is really hot.Now we are sweating granite bricks trying to get it in the mold fast before gell.

Aggregates from <0.1mm (Sand dust) up to 16mm (on the larger castings from 80mm section thickness). Whereby you want to aim for 1/5 to 1/8 your smallest feature.

Resin ratios by weight of 7% to 10%.

Shaking frequencies up to 70Hz and accelerations to 25 m/3 (2.5g) This is basically your industrial 2 pole motor with a unbalance. If you pick up a cheap motor and VFD off ebay, you can vary the speed until you hit the sweet spot for each casting. The largest castings use a combination of a shaking table and shaking motors bolted to the outside of the form.

For optimal bond, any steel inserts you bond into the part should be sand blasted.

The practical method for prototypes and us homebuilders is wooden forms with steel sections moulded in which are subsequently milled or ground to tolerance. The big guys are going more and more to cast to tolerance, but also do grinding of the mineral to tolerance. If you have a surface plate and an eye for detail, grinding and scraping alignment rails could be a garage job.

Tolerances to +- 0,5mm should be posible with wooden molds. Edge distance for inserts should be >3D. If you intend to reuse the form, it should taper about 5°. Round all internal edges.

A rule of thumb would be a polymer concrete machine base is made the same size as a cast iron part, but solid where the C.I part would be cored, should have nearly the same weight and approx 3.5x the stiffness.

As the shrinkage of epoxy occurs around the time it gels, much if the shrinkage is compensated by further filling of the still liquid mix. The final shrinkage is given as 0.02 - 0.03% linear.

One fellow has a clever idea, which is to partially sink 80/20 extrusions in the polymer concrete. This would deliver their precision together with the superior rigidity and damping the polymer provides.

One of the best finds in the thread is this German site, where some hobbyists are making machines using composites. You'll need a translator though!

As far as suppliers go, there is the Accures starter kit, or you could try to buy the materials from an online supplier like US Composites or Epoxy Systems.

The level of sophistication increases as the thread goes on. Degassing is a problem with epoxy. Mixing it embeds a lot of air. You'd like to get rid of the air. There are two methods. One, use a pain stripping heat gun. Keep it moving constantly! Two, try vacuum degassing. Someone pointed out there are vaccum systems available from auto suppliers that are designed to suck the oil out of a car. They generate 15 or 20 inches of vacuum, which is plenty to degass. Now you just need a suitable vessel to do it in. Vacuum systems are also used for veneering by woodworkers, and would also be usable for this purpose.

A Plate Machining Fixture and some Homemade T-Nuts for my Phase II Rotary Table...

Check out a coil for an auto air conditioning compressor clutch. They're easy to come by, relatively inexpensive, easy to mount, and pack a lot of holding power for their size.

In the thread on a portable CNC plasma cutter, someone mentioned jib-style versions. I could not come up with an example from the Internet, but drew this quick sketch of one:

The sketch shows a 48" arm of 2 x 2 material, perhaps an 80/20 extrusion. The base is an 8" diameter cyliner, and the jib is supported at the end with a downlink. It probably makes sense to run that center shaft up high enough to hold the plasma cable above the fray. There is some engineering needed to make the rotating axis run smooth and true, but it isn't that bad. The Z can ride out the extrusion on a skate bearing arrangement of some sort.

It's an interesting thought, and I suppose given the needs of a plasma system, it could be quite light and doable. One could even imagine a magnetically attachable base.

I'll have to ponder this one further. Need to consider the programming implications on Mach 3 as well. Ideally, you want this to look just like a regular x, Y, Z gantry when programming. I'm pretty sure Mach can do that, though. One problem is the torch speed is going to vary wildly at different radii from the base. That alone is probably a reason not to use this design for plasma, which is pretty sensitive to that speed.