Gregor would be proud

Machining vertex pieces for a v2 ‘Mendel’ RepRap.

I’ve been rather fascinated lately with the RepRap project.  In a nutshell, this is a project to build a low-cost rapid prototyping machine. Specifically, RepRap aims to design and build rapid prototypers that can make as many of their own parts as possible, which is a noble goal, but not a design facet that particularly interests me right now. There are other rapid prototyping projects such as Fab@Home and Makerbot’s CupCake CNC (which is based on the earlier RepRap design known as ‘Darwin’), but the latest RepRap design known as ‘Mendel’ is really elegant.  Mendel is a far simpler design than its predecessor, has a larger working envelope, and should prove far more scalable – stretching any of the axes for more travel should be pretty trivial if the design ever needs expansion for making larger parts.

Ideally, building a Mendel for myself would start with getting a set of rapid prototyped Mendel parts from someone else, but there seems to be hardly anyone making Mendel parts yet.  Besides, I really have no interest in having a ‘pure’ machine that was replicated as much as possible – in fact, I’d prefer to actually machine my own parts for it.  I’ve been eagerly following Shane Wighton’s blog, as he also wanted to build a Mendel and has access to a machine shop.  We both came across the same issue of trying to view the 3D assembly of Mendel in the downloadable solid model files – the files were created in an academic licensed version of Solid Edge, which the freely downloadable Solid Edge Viewer refuses to display.  I thought perhaps it was a video driver issue until I tried running the program on another machine.   A student friend who has access to an academic version at school was kind enough to create an eDrawings version of the assembly for me. I’ve used eDrawings created from SolidWorks many times, and they work beautifully.   However, the .easm created from a trial copy of eDrawings for Solid Edge was abyssmal.  Missing parts galore, including one of the steppers (why it singled out a single stepper is beyond me – they’re all from the same part file).  Geometric Ltd. seems to think that the software is worth $395.   I suggest they revise that figure downwards by a few orders of magnitude.

I finally realized I’d just have to re-create the parts in SolidWorks.  Not a huge deal – SolidWorks barfed a bit on importing the entire assembly of Mendel (no issues with being from an academic version of Solid Edge – take note Siemens, as even competing products are doing better with your own files), but opens the individual .par files quite happily.  Like Shane, I decided to start with the vertex pieces – while he opted to use a sine bar in the vise to provide the right angle on the vertex pieces, his use of a tooling ball (which I had never heard of before, and it took me a little while to figure out how they are used) gave me an idea of how I could do all the machining without worrying about angles. The trick is simply in relocating the hole that Shane was using for the tooling ball so that it runs in line with the two outside holes (and using the same diameter for all 3). Then, by using short 5/16″ rods through 2 holes at a time, I can accurately cut any of the faces.

I started with 3 pieces of scrap 3/4″ aluminum plate and drilled six 5/16″ holes though each at the coordinates that I had determined by a CAD sketch.  I then bandsawed the pieces in a chevron shape:

vertex drilled and sawed

Then I ran two pins through the holes of what would be a ‘leg’ on the part and clamped it in the mill vise:

pins and vise

In order to set the tool height, I placed a parallel across the pins and brought the endmill down onto it, then locked the quill, after which I lowered the knee by the appropriate amount:

height setting on pins

This allowed me to machine the ‘convex’ side of each part.  The other piece clamped in the left of the vise is just there to even out the force on the vise, not to have any machining done on it (I actually had something always clamped on the left, but removed it for most of these photos for clarity).  To machine the concave side as well as the end of each ‘leg’, I needed to use an end stop:

inside of vertex milling

Note that I started using 5/16″ drill bits rather than the 5/16″ pins of the previous photographs.  The stainless steel pins were a very tight fit (I didn’t have a 5/16″ reamer on hand, so the holes are slightly undersized) and I had resorted to using a hammer and punch to drive them in and out of the holes.  This got old really fast, so I just used a pair of drill bits instead – they had drilled the holes anyhow and were a loose enough fit that I could pull them out by hand, which really sped things up.  After machining the ends, all milling was complete, and I just had to drill the two cross holes:

drilling cross holes

A touch with a countersink tool on all holes and a pass of a file on any rough edges  completed the work.  I’ll toss them into the vibratory deburrer later to give them a nice even finish.

completed vertices

One last thing – I’d like to give a shout out for Milwaukee Makerspace, which is a group of hackers/makers/tinkerers hoping to start a local ‘makerspace’ (think ‘a place for geeks to play with machine tools’).  Come join us – the more people we can get, the cheaper it will be for everyone.

5C Collet Chuck Mounting

Having been enamored with the 5C collet chuck on the 9×20 lathe, I certainly wanted the same for the big Keiyo Seiki.  Not having a whole lot of use for the 9×20 anymore, I decided to simply move the chuck to the much bigger lathe.  I looked around online for a suitable adapter plate, but I wasn’t sure of precisely what I needed, so I went back to New England Brass & Tool and Bob had just the adapter plate I needed in stock, and at a price lower than I had figured.

Frond and rear of the adapter plate
Front and rear of the adapter plate

This was a semi-finished backplate, which means that it still needs final machining to fit an attached chuck (more on this later).  However, it also had the recess for the spindle’s indicating lug off-center.  I’m not sure if this was actually intentional, as the lathe’s indicating lug is right on center with the rest of the bolt pattern:

Note the indicating lug at the top left of the spindle - it follows the same spacing as the threaded holes around the perminter, unlike the backplate.
Note the indicating lug at the top left of the spindle - it follows the same spacing as the threaded holes around the perimeter, unlike the backplate.

Well, there were two possible solutions – either drill a new indicating lug recess on the backplate, or drill and countersink new mounting holes through the backplate and use the existing recess.  I decided to just drill a new recess, opting for simplicity, as I’m sure I wouldn’t be able to drill 3 new mounting holes with the same accuracy as the existing ones.  The indicating lug hole doesn’t have to be super precise anyhow – I think it’s simply there to make sure that the same holes on the backplate match up with the same holes on the spindle with each mounting, ensuring better accuracy.

I use a Blake Co-Ax indicator to determine the center of each of the mounting holes.
I use a Blake Co-Ax indicator to determine the center of each of the mounting holes.

I clamped the plate to the mill’s table with a hold-down clamp that was almost the perfect size (I filed the tail end of the clamp a little to get it to fit the inside of the plate).  I then found the center of one of the mounting holes and set its location as the origin on the DRO.

The DRO basically feels like cheating after having used just handwheel dials.
The DRO basically feels like cheating after having used just handwheel dials.

I then moved around to the other holes and the recess for the indicating lug, noting the coordinates for each one.  I then whipped up a quick CAD drawing with each of the 4 points to see how far off the lug recess from the bolt circle was (if anything).  It looked to be off of the bolt circle by only 0.002″, which I’d simply consider measurement error on my part.  I then determined the X,Y coordinates for a lug recess centered between two of the mounting holes.  Back at the mill, I shuttled the table to this location, locked the ways (on my old Tree, locking really doesn’t put a lot of clamping on the gibs, but it helps keep things steady), and proceeded to center drill the spot, then drill down about 0.4″ with a 1/2″ drill.  The recess needed to be just a little over 3/4″, but I didn’t have a 3/4″ drill, so I used a 3/4″ endmill to bore the depth.

Using an endmill to hog out most of the recess
Using an endmill to hog out most of the recess
Finishing up with a boring bar
Finishing up with a boring head

After bringing the recess to size with a boring head, I removed it from the table, cleaned it off and tried attaching it to the spindle.  The screws went in rather tight, and it had difficultly squeezing flat against the spindle.  I guessed that my hole for the indicating lug was off by just a bit, and I was squishing the lug.

Technically it fits, but took more torque than should be needed
Technically it fits, but took more torque than should be needed

After removing the plate, I had a look at the lug recess and saw the telltale signs of metal interference:

Evidence of interference
Seeing linear marks here indicates that the lug was not centered in the recess and was binding on this edge

I clamped the plate back onto the mill table and bumped the 3/4″ endmill up against the marred edge of the recess.  I then zeroed the DRO, retracted the quill, moved over about 0.005″, then milled down about 0.3″ to relieve the area that was binding.  I put the plate back on the spindle, and the screws tightened up a bit easier this time, so I considered the rear of the plate to be complete.

The front of the plate has a raised boss that slips inside the rear edge of the chuck to keep it centered, and this boss must be cut to size once the plate is mounted to the spindle.  This ensures that the boss is cut perfectly concentric with the lathe’s spindle (something impossible for the manufacturer of the plate to do, as every spindle will run just a hair different).

Machining the boss on the front of the plate to final diameter
Machining the boss on the front of the plate to final diameter

I had to take the diameter of the boss down about 0.060″ or so – I used a carbide bit and took pretty light passes so I could ‘sneak up’ on the final dimension without cutting any further than necessary.  Once I got close, I’d stop the lathe, clean off the chips (dust, really – the plate is cast iron, which creates more of a coarse powder, like fine sand, rather than chips like you’d get from aluminum or steel), and try fitting the chuck to the plate.  After the final thousandth of an inch, the chuck slipped on with no side play, and I fastened it in place with the mounting screws.

Finally - all mounted!  The chuck looks downright puny on such a large machine.
Finally - all mounted! The chuck looks downright puny on such a large machine.

Now I was curious to see just how accurate the chuck was – with the backplate cut so perfectly, I should ideally see zero runout on the chuck.  I attached a dial test indicator to a magnetic base and had a look.

Testing runout on the 5C collet chuck
Testing runout on the 5C collet chuck

Before even checking the runout, I decided to see how rigid the chuck and spindle are on the lathe – on the 9×20, I could get the indicator to deflect a thou or two just by pushing firmly on the chuck perpendicular to the spindle axis.  I pushed with about the same amount of force with the same chuck mounted on the Keiyo Seiki, and watched the indicator needle anxiously.  Not.  Even. A. Single. Twitch.  This beast is solid.  A side effect of such rigidity is that a runout measurement should be a lot more accurate, so let’s see what we have…

Total Indicated Runout (TIR) is under 0.0015", the difference between the two extremes shown.
Total Indicated Runout (TIR) is under 0.0015", the difference between the two extremes shown.

Appears to be just under 1.5 thou – not perfect, but good enough for the moment.  On the 9×20 I had cut the boss on the backplate a bit undersize accidentally, but the extra slop actually allowed me to adjust away the runout by carefully snugging up the mounting screws, checking TIR, gently tapping the chuck in the appropriate direction with a mallet, checking again, tightening the screws further, and so on to make the chuck run true.  Of course, the collets themselves have runout as well, but I don’t worry much about that if I can get the chuck adjusted well.  But enough of that for now – time to cut some metal!

First cut with the 5C collet chuck on the Keiyo Seiki.
First cut with the 5C collet chuck on the Keiyo Seiki. You can see little metal shards sticking to the surface of the part due to a less-than-sharp cutter being used. After swapping in a better cutter, the resulting surface finish was nice and clean.

Toolpost for the Keiyo Seiki

My big Keiyo Seiki lathe came with a simple lantern style toolpost, which is considered pretty ancient as far as toolholding technology goes.  While they do have their uses (the narrow profile is quite nice for certain workpieces or setups), a quick change toolpost will be much more useful to me, especially when doing small scale production.

I wanted to get something really good for toolholding on this lathe, and the Aloris wedge-type quick change toolposts we had on the lathes in class were the Cadillacs of the field. Unfortunately, they have a pricetag to match – a local machinery dealer has a used Aloris CA set but wants $700 for the kit. Much as I hate to contribute to a trade imbalance, getting a Chinese made clone would be the only way to afford such a setup. I looked around at some of the offerings, and word on some of the machining forums was that the units from Quality Machine Tools weren’t half bad (but that the set screws and other hardware they come with are poor and should be replaced).

The big decision was whether to get a CXA or CA style toolpost – the difference is size. Measuring between the top of the T-slot on the lathe’s compound and the tip of a live center in the tailstock gave me a distance of 1-15/16″. Using the dimensions found in the Shars Tool Co catalog (I assumed that it’s probably the exact same stuff as what Quality Machine Tools sells), I determined that either CXA or CA would work fine for the range of sizes that their toolholders were designed for. I decided to go for the biggie and ordered the larger CA unit, as it would allow me to use 1″ square tooling.  Plus, I remember chatting with a retired machinist many years ago, and he noted a rule of thumb when buying machine tools – if you have a choice between a machine that weighs 5 tons and one that weighs 7 tons, and both otherwise have identical specifications, always buy the heavier one. The extra mass means extra rigidity and better vibration dampening, which translates to being able to hold tighter tolerances and better finishes on machined parts. Never pass up the chance to listen to an old machinist – those guys are treasure chests of practical wisdom.

The massive CA clone toolpost next to a rather feeble looking lantern toolpost

When the toolpost set arrived, I was rather taken aback at the weight – this was a beast.

The stock T-nut, larger than the compound's T-slot.

The T-nut that the toolpost bolts to the compound with is oversize so that the end user can cut it to fit (there being a very wide array of T-slot sizes).  I took a few measurements and tossed the T-nut on the mill.

I started by cutting the width to size. The 'smoke' you see is vaporizing WD-40, which I sprayed on as a lubricant/coolant.
Next I took the corners to size.
After a few passes to take some material off the top, the slimmed T-nut now fits beautifully.
Fully mounted and ready for action.

I still need to replace the set screws and other hardware on the toolholders, but thus far I’m quite impressed with the pieces for the price.  The dovetails fit up very closely, and honestly, I can’t imagine that the extra $500+ to get a ‘real’ Aloris would actually net me much more.  Now to scour Ebay for some beefy 1″ Kennametal insert holders…

Confessions of a machine tool junkie

It’s a pretty sure sign that you’re a machine tool junkie when you browse through the craigslist ads, see a machine that you have no room for, no acceptable source of AC power for, and no conceivable need for, but you still check the balance in your checking account.  Such was the case the other week when I saw a Sunnen LB hone for only $250. I have no immediate need for one, though there are a few projects in mind where it would be useful. There’s hardly any info on them available (entering ‘sunnen model lb hone’ into google yielded as the first result… …that very craigslist ad), and this one was in need of new belts at a minimum (I had stopped by to have a look at the unit, being only 5 minutes away from work). Still, after pondering it overnight, I left a voicemail for the seller the next day telling him I’d take it. I got a voicemail back informing me that it had already been sold. Phew, what a relief!

I have a number of machines currently in my possession, though in this post I’ll just talk about the lathes, as the newest one is the source of most monetary expenditures as of late.  A puny little Grizzly 7×10 lathe was my first machine tool purchase many years ago, and it was so small that it’s not even offered anymore, having been replaced by the 7×12.  For those unfamiliar with machine tool lingo, calling a lathe a “7×10” or a “12×36” is roughly analogous to calling an engine a “305 cubic inch”, as the ‘displacement’ is essentially what is being described – the first number refers to the size of the largest diameter part that the lathe can hold, and the second number refers to the longest length part that can be turned ‘between centers’, which is a common method of workholding.  In the case of the 7×10, it could work with a short stubby piece of 7″ in diameter, or a long thin piece of 10″ in length, hence the 7×10 designation.  It wasn’t a great lathe, but it had all the features I wanted, especially the ability to cut threads.  I knew I’d probably wind up getting a larger unit someday, but this one sufficed just fine as a ‘trainer’ model.

The poor little 7x10 now sits unused under piles of debris in the garage
The poor little 7x10 now sits unused under piles of debris in the garage

Sure enough, many years later I really needed an upgrade, as the little 7×10 just didn’t have the torque, rigidity, or capacity that I needed.  Plus, having to swap gears (plastic ones, no less) to change thread pitches was a pain.  I eventually worked out a trade with Doc Nickel, who was also on the upgrade path – I had a bunch of solenoid valves, and he had a Grizzly 9×20 that he had outgrown. It was just the thing I had been looking for, and it certainly had a bit of history behind it, having modified countless paintball guns into custom works of art. Of course, I then wanted to modify the machine after a time, and I decided to replace the 3-jaw chuck with a 5C collet chuck for improved workholding on round pieces (getting a good grip on parts with a 3-jaw chuck generally leaves unsightly dents in the part where the jaws have dug in). Not being exactly sure of what mounting system I needed for the chuck, I called New England Brass & Tool, and Bob Cumings made certain to get me just what I needed, even throwing in a depth stop as a freebie, as he knew I’d find it useful. The 5C was indeed a great addition, and in fact I never took it off.

The well-used 9x20
The 9x20, with 5C collet chuck and quick change toolpost

While that was all well and good for workholding, the toolholding still needed attention – like the 7×10, the 9×20 had only a turret style toolpost. Not that a turret toolpost isn’t bad, but the stock toolposts on these little import lathes simply aren’t very good – they’re poor imitations of ‘real’ turret toolposts, and are used simply because they’re inexpensive to manufacture. I bought a Phase II piston-style quick change tool post, which, as the name implies, allows for quick changes of tools, each held securely in its own dovetailed toolholder. Additionally, I removed the compound from the lathe entirely, as they are not particularly rigid on these lathes, and I couldn’t forsee turning many tapers. Instead, I mounted the toolpost to a thick aluminum plate which itself got bolted to the cross-slide (which thankfully has T-slots perfect for this). This served me well for quite some time, but of course it couldn’t last. Once I moved up to a full size vertical mill I naturally needed a lathe to match, especially after taking a lathe class at MATC – after using a Summit 18×40 for a semester, I could barely even look at my feeble 9×20.  (Not to mention that the class lathes were outfitted with DROs, Aloris (real Aloris, mind you) wedge style quick change tool posts and a full complement of Kennametal tooling)

I first had my eye on a used Jet 14×40 at a local machine tool dealer – it needed some parts, but it had a big 3″ bore through the headstock.  The dealer wanted $3000 for it, and I figured I could save up the cash for it.  Naturally, though, someone swooped in and bought it.  I then had my eye on a nice big Andrychow TUG-40 from the same dealer – certainly a nicer, bigger lathe, but with a pricetag to match – something like $5700.  Ouch.  I figured I’d have to make acquiring a big lathe a much longer term project.  I browsed craigslist for months on end, hoping to find just the right one at a price that wouldn’t kill me.  Then, in January, I found the one I had been waiting for.

Keiyo Seiki KM-1800C
Keiyo Seiki KM-1800C

What a beauty – a 17×48 monster, even larger than the lathe I had used in class.  With a 7.5HP motor, this beast should easily be able to chew through anything I dare to feed it.  And a full quick change gearbox – no more messing with change gears or even belts for speed changes!  The price was a steal at only $2100, and I knew that I better nab it now or live a life of regret.  Of course, buying it and actually taking delivery are two very different things.  I called Big Red Movers for a quote on actually moving it home, but they tossed out a price of $1000 for the service. Yeah, like I’m going to drop half the cost of the machine on just transporting it. A bit of chatting with coworkers finally hooked me up with someone with a truck and dump trailer who was able to help me haul it over to work. It was forklifted onto the trailer at the seller, and forklifted off the trailer at work, where I could temporarily store it while figuring out a way to transport it home.

I decided on building roller bases for the lathe in order to get it home and into the garage, as these had worked very well for my vertical mill. As long as the unit could be loaded onto a rollback tow truck, it could be rolled right off the truck bed (using the truck’s winch to gently lower it down the bed) into the garage. My dad managed to dig up most of the metal needed for building the roller bases, and I welded them up at work. I used the same Fairbanks swivel casters as I had on the mill bases, hoping that they’d be able to withstand the heavier load of the lathe. Once I had the bases built, I forklifted the lathe onto them, and wonder of wonders, they held. After that, it was a simple matter of hiring a rollback tow truck, rolling the lathe onto the bed (via the loading dock at work), and then bringing it home. I asked dad if he could help me with the unloading, as it makes me feel a lot better having someone around who has far more experience in such matters. Fortunately, the lathe came down the bed extremely smoothly, and we parked it neatly in the garage with a minimum of fuss. I’d deal with unloading it from the roller bases at a later time, but for now I could simply smile with satisfaction at finally having a ‘real’ lathe.

Simple CNC Engraving Head

My very first summer job was working at the local trophy shop, where we had several computerized engravers.  As it turns out, they were extremely similar to the Taig CNC mill that I would many years later acquire – driven by steppers in the X and Y axes and controlled via a desktop PC, it’s really not a stretch to consider them specialized CNC tools.  Our most commonly used engraver was a ‘diamond drag’ type, which essentially used a diamond tipped tool to scratch through the lacquer covering a sheet of brass or aluminum (our other engraver used a rotary cutter for engraving on plastics).

Once I had my Taig, I realized I could certainly do a little engraving myself, I just needed the right tool.  On the engraver at the trophy shop, the ‘head’ was actually pneumatic – rather than a Z axis, there was simply a 3-way solenoid valve that would apply air to a spring retracted engraver.  More air pressure meant deeper engraving.  There had been plans in Home Shop Machinist (might have been the sister publication, Machinist’s Workshop) for a spring-loaded engraver for use on CNC mills a while back, but I wasn’t thrilled with the design – not only did it reduce the working travel in Z, but had to go in the spindle.  I wanted an engraver that could be attached to the side of the headstock (Taig headstocks have T-slots for such purposes) which would not only be a bit more rigid, but would allow me to machine a shallow flat on a part and immediately engrave in that area without any tool changes.

Construction is simple – the only purchased parts were the diamond drag engraver (I used a DG-250 from Antares), a pair of clamp collars, and a spring. The body was a hunk of scrap extrusion from work – it had a 5/16″ hole already down the length of it. I machined a pair of bushings from some Oilite bronze rod stock and pressed them into each end of the body (I had bored each end out a little bit). Though I had reamed the bushings to have a 1/4″ hole through them, there was enough axial misalignment between the two to cause the engraver to bind when I tried to run it through. So I tried something a little crazy – I chucked a long piece of 1/4″ drill rod in the lathe, lightly scuffed the last two inches or so of the rod with a diamond hone, slathered it in oil, and ran it through the bushings. This trued things up just enough to allow the engraver to slide through the bushings freely, but with zero side play. A bit of strip steel, some threaded rod and wingnuts, and it’s ready to attach.

Assembled engraving head
Assembled engraving head
Engraving head parts
Engraving head parts

The mother of all CNC programs

I’ve been working on cocker threaded Phantom bodies for quite some time, and the latest incarnation is a ‘full round’ version.  I had a job shop do the internal machine work on them as I didn’t yet have the big lathe running (which I’m now having some regrets on – the cost was substantial, it took over 5 months, and despite what I was told at first, they want the invoice paid in full NOW).  I wanted to do the final machining in a single setup, as they’re round and alignment would be a concern were I to try multiple setups.  Furtheremore, a single setup would require 4th axis operation, as there are features on the top and bottom of the part.  In short, there were a lot of ‘firsts’ for me in this project, and it’s the most ambitious CNC program I’ve yet done (experienced CNC machinists would laugh at its simplicity, but hey, I’m learning).

Roundhead body on the mill
Detent slot milling operation in progress

I broke the program down into pieces, some of which I had already done.  There are 5 subroutines in use:

  • sear slot
  • threaded insert counterbore
  • feed port
  • threadmilling for insert counterbore
  • detent slotting

The sear slot is first, then the underside counterbore, rotate 180 degrees, then the topside counterbore and feed port.  Tool is changed to the threadmill, top counterbore is threaded, rotate 180 degrees and thread the bottom counterbore.  Finally switch to the ball end mill and run the detent slots.  Total run time is still over a half hour, which I hope to cut down a bit by removing some unneeded ‘air’ cuts, increasing feeds, etc.  Still, the satisfaction is immense – I’ve had a lot of tools (the 4th axis, tooling plate, tailstock, bullnose live center, etc.) sitting idle for a very long time (they were all purchased with this project in mind), and it’s a great feeling to finally have them put to use.

Mach3 while running the CNC program
The three tools used in the program, held in A2Z endmill holders
The three tools used in the program, held in A2Z endmill holders