Building Machines

The Easy Way to Square a Gantry!

Posted in Building Machines on May 7th, 2012 by Judah – Be the first to comment

UntitledI figured it out!

As is often the case, after stepping away from the problem for a bit I came up with a simple solution:

First, secure one side of the gantry to the table (or have a friend hold it).  Next, apply power to the machine, but engage an e-stop.  Now take a long square (I used a drywall T-square) and place the short side flush against the side of the gantry that is secured so the long side is pointing across the table to the unsecured side.  Make sure that one edge of the long side is lined up with some landmark that appears on both sides of the gantry (I used the vertical edge shown in the above pic).  Finally, go over to the unsecured side of the gantry, manually deflect it until the same landmark is lined up with the appropriate edge of the square, and then release the e-stop switch to apply power, thus locking it in.

Granted, this is not a super-precise method, but given that you’ll probably get your deflection down to less than an eighth of an inch over the entire width of your table, it should be close enough for most work.

Ok, back to making Kickstarter rewards!

That’s right! I’ve got a blog, don’t I?

Posted in Adventures, Building Machines, CNC Machines on January 23rd, 2012 by Judah – 4 Comments

IMG_0092My apologies for the long silence; the last month has been a bit of a rollercoaster.

Right after I finished milling out the parts for the next Kikori gantry up at MakeIt Labs, it got shut down.  This meant that instead of getting to work on the kickstarter rewards I had to dismantle my entire operation up there, move everything down to my new space at the Artisan’s Asylum (living up to their name, as always), and build a new gantry first.

If you’ve been following my photostream on Flickr, you’ve watched as I assembled a new Kikori gantry.  While I did find a few minor things that needed tweaking, for the most part it came together beautifully.  This was especially encouraging since the Kikori up at MakeIt was having all sorts of mysterious issues that prevented it from maintaining positional accuracy (given that the new machine has none of these issues, I’m told that it could’ve been caused by a “dirty” power supply in the MakeIt building, which would’ve caused the machine to miss steps).  This showed that even under a worse-case scenario, the Kikori is still capable of self-replication.

In rebuilding the Kikori I also tried out a couple different methods of driving the X axis.  The problem I was having with the original setup of using a single NEMA 23 motor to drive sprockets on both sides of the gantry was that I’d get significant twisting along the drive shafts; enough that the gantry would ‘chatter’ as is moved in that direction.  My first solution was to upgrade to a NEMA 34 motor which uses a 1/2″ drive shaft instead of the 23’s 1/4″.  While this did eliminate the chatter, it also revealed that without the twisting  rod acting like a spring, the motor wasn’t powerful enough.  Finally, I decided to try using two seperate NEMA 23 motors to directly drive the sprockets on either side.  This proved to be by far the best solution: it gave me the best strength, the smoothest motion, and also eliminated the need for drive shafts, bearings, and couplers.

The Kikori is now performing beautifully, maintaining accuracy to within 1/32 of an inch even at feed rates of 100 IPM!  This means that production of the Kickstarter rewards has finally begun.  I’ll be milling out the sets of gantry parts first, then working my way down the list of rewards.  I hope to get them all out within a month, but we’ll see how things work out.

This also means that I’m about to start selling Kikori kits as well!  After all the changes and additions, it looks like I’ll just be able to keep it under the $5,000 goal, but I’m going to try to do a special introductory sale to get things moving.  I’ll send out an announcement soon!

The Kikori is Fully Functional!

Posted in Building Machines on November 3rd, 2011 by Judah – 9 Comments

It’s making things!

About damn time too.  Turns out that the problem all along was that the computer was missing steps (not moving as far as it thought it did), so after I swapped it out for an old p4, it’s running like a dream.

I’ve been told by people on here and elsewhere that the fact that it’s a dual-core machine that can’t do core isolation might have been the culprit (since newer dual-core machines don’t have this issue).  Whatever the case, it’s fixed now!

The picture is of a new router plate I made for the bottom of the Z gantry.  This new plate serves 4 functions: diverting the air blowing out of the router (so it doesn’t blow dust everywhere), allowing access to the collet (see here and here), creating an attachment point for a vacuum hose, and providing a surface onto which a future removable brush shroud can be mounted (this is why the vac hose is where it is, even though it’s largely ineffective without the shroud).

I’m now in the final stages of testing and refinement before I make the next set of gantry parts.  Once that happens I can move into the Asylum, build the next version, and FINALLY GET DOWN TO BUSINESS.

Trying a new computer

Posted in Building Machines, CNC Machines on October 26th, 2011 by Judah – 10 Comments

The Kikori continues to have issues.

I replaced the couplers, I added set-screws to them, I even applied loctite 680 retaining compound to prevent slipping, and still it would not return to where it started after even a four minute air-cut.

At this point, I’m pretty sure I’ve eliminated all mechanical sources of error, which leaves only the computer that’s driving it.

I’ve been using an old PC given to me by a local friend of mine, and I did notice that when I ran the latency test for EMC, the numbers were on the high side, getting up to 50k ns (it should be under 20k, preferably under 10k).  This could explain my accuracy problems.

I took the PC to another friend who helped me optimize it and even set things up so I could use another computer to control the first one (which did reduce the latency quite a bit).  However, air-cut testing still showed that it was getting off.

I’ve decided to try another computer, this time with a single-core 2.8GHz P4.  I’ll let you know how that goes!

So. Close.

Posted in Building Machines on October 3rd, 2011 by Judah – 3 Comments

The new couplers are doing great!

Mostly.  I think.

Since I’m done traveling for the next few months, I finally got to spend some time working on the Kikori last week. I wanted to mill out one of my tablet stands, but to make sure everything was staying nice and accurate, I did a test first.  I decided to drill a hole at a known location, do an air-cut of the hole pattern, and then have the machine return to the hole location to see if it had gotten off at all.

It had.  Only about 1/16th of an inch in X, sometimes in Y (I tested it multiple times), but it had.

The two most likely culprits are set screws working themselves loose and couplers slipping.  I applied loctite to the set screws before I left, so today they should be set.  I’ll make registration marks on the couplers to check on slippage there.  Wish me luck!

Coupler Upgrade!

Posted in Building Machines on September 22nd, 2011 by Judah – 6 Comments

Remember the coupler issues I was having?  Fixed ’em!

While I was in Oakland I did a fair bit of research on what my options were, and finally decided on these “light duty” couplers from McMaster-Carr (Part no. 6115K18).  As you can see, not only are they aluminum (instead of plastic (nylon?) like the last ones), they are HUGE.  You can see the size difference here.  I actually had to dremel out the x-axis motor mount to get it to fit, but thankfully it didn’t need much.

After slapping those puppies in and tightening up some things (I really need to remember to put loctite on those set screws), I milled out a couple new test pieces.

Beautiful.  Not only did they have wonderfully smooth edges, free of steps and ripples, but they also fit together snugly!  This means that they are likely within 0.01″ of their intended dimensions.  I’ll have a better sense of it’s accuracy once I mill a harder material like masonite.

Of course, now that my machine is finally working well I’ve got to go to Chicago for my nephew’s bris, but once I get back on Monday, I’LL FINALLY GET TO MAKE STUFF!!!

I’m very excited.

The Kikori Has a New Home!

Posted in Building Machines, CNC Machines on July 16th, 2011 by Judah – Be the first to comment

The Kikori is now at MakeIt Labs!

Granted, right now it’s little more than a pile of MDF and various bits of gantry, but it shouldn’t take long to put back together.

My biggest concern right now is the torsion box.  Unfortunately, very few of the ribs survived the move intact; most of them are in two if not three pieces.  Thankfully, I have all the pieces and it’s clear which pieces go together. This means that while I can reassemble the torsion box, but I’m not sure how strong it will be.

My current plan is to put it back together as-is and immediately use it to make a new set of ribs out of 3/4″ MDF instead of 1/2″.  This means I would have to use the reverse side of the skins since these new ribs would no longer fit into the slots the original ribs fit into, but given that the slots were too tight to begin with, this might actually be a good thing.  Another option is to make them out of 1/2″ MDF again, and simply make them taller this time, increasing their strength that way.  I could sand out the slots to make them looser.

Of course, if the reassembled torsion box seems fairly strong even with the broken ribs I might just use it as-is. After all, having a perfectly flat milling plane is only super-important if you’re doing 3D carving; it doesn’t matter as much for simple profiles.  Besides, I’m itching to start prototyping the launcher!


Posted in Building Machines, CNC Machines on May 21st, 2011 by Judah – 2 Comments

Ok, the kikori has officially surpassed my expectations.

A couple days ago it made it’s first thing: a balisong screwdriver I designed which you can see here.  I wasn’t that impressed with this since a couple pieces didn’t survive the milling process and the ones that did didn’t fit together that well.  While reviewing the video I took of the milling it occurred to me that one reason for this might have been very small movement of the spoil board the material had been taped to (spoil board is what goes under your workpiece so that when you cut all the way through it you’re not cutting into your table).  Sure enough, when I compared the first and last frames of the video you could clearly see a small shift.

The next day, with this in mind, I taped down the spoil board and ran a set of parts for another design of mine: a folding tablet stand (pictured above).  Right after the job was done I could tell there was a lot of improvement, but I didn’t see just how much until I cleaned off the parts and started putting them together.

This picture shows the main arm of the stand that has a slot cut in it.  Sitting perfectly in the slot is another piece.  When I designed this stand I made the smaller piece to be just 0.02″ narrower than the slot so it would sit in there loosely.  When I measured these milled parts the actual size difference was 0.022″.  This means that my machine was only 0.002″ off.


That is a ridiculously small error margin, far smaller than I thought I’d ever achieve with this machine.  I’m amazed.

Unfortunately, I was unable to use this first set of parts because I discovered some small flaws with the actual design, but an hour later I had applied the necessary changes and I milled out the version pictured above.  I accidentally milled those parts out at twice the speed of the first set, so they don’t have the same tight tolerances, but they still fit together beautifully.

Today I’ll finish up some other designs and perhaps get to some more milling tomorrow.  I’ve started a new Flickr set called Kikori Work where I’ll be posting pictures of stuff as I mill it out.  Now that the kikori’s up and running, I’ll probably be adding to it quite frequently, so be sure to check it out for the latest stuff.

I’m also getting ready to launch my Kickstarter project sometime next week, so stay tuned for updates on that.  I’ve got a potential list of rewards I’ll post here soon, and I’d love everyone’s feedback.

Assembly is Complete!

Posted in Building Machines, CNC Machines on May 18th, 2011 by Judah – Be the first to comment

Last night I finally finished assembly of the kikori!  After taking a break for a good maniacal laugh, I got started on testing by putting a sharpie in the router and having it draw ‘EMC2 AXIS’.  Not surprisingly, I ran into issues right off the bat.  Thankfully, tackling them was fairly straight-forward.

The issue was that there was some slop in the X and Y axes.  The source of this slop was the set-screws in the drive sprockets; if the screws aren’t tightened down all the way onto the flats I machined into the drive shafts, the sprocket will rock back and forth when the gantry changes direction.  On the Y axis this was fairly easy to fix; since the gantry is comparatively light, I was able to rock it back and forth while tightening the screw until it was solid.  However, on the X axis the weight of the gantry meant that this approach would cause me to strip the little screw before it would tighten all the way.  Therefor, this afternoon I implemented the fix I used on the blackfoot I built in Boston.

I replaced the small set screw with a socket-head screw.  This allowed me to use a larger allen wrench which in turn allowed me to apply more torque in tightening it.  The only downside was that I then had to grind a flat on the head of the screw so it wouldn’t rub against the roller-chain.

After all that I ran another test and found that while things were much better, there was still some play in the X axis.  This turned out to be from the Lovejoy couplers I was using to connect the motor to the two drive shafts.  I’ve used these couplers before, and they had no perceptible backlash, but these had enough to translate to about a eighth of an inch of travel.  Luckily for me, I know a local guy who bought a blackfoot kit that he has yet to assemble, and he lent me the ridged couplings from it.  Once those were slapped in, it ran like a dream.  There was still a tiny bit of wobble, but I’m fairly certain that that was the sharpie not being solidly mounted in the router.

There’s Movement!

Posted in Building Machines, CNC Machines on May 16th, 2011 by Judah – Be the first to comment

I’ve got two of the three axes moving under their own power!  I’ve mounted the stepper motors to the gantry, added the roller chain on the Y axis, and connected everything to a computer.  Check it out: