Jon's Place

Monday, April 30, 2007

More Grippers

I'm constantly amazed at the capabilities of my Sherline CNC milling machine. Tonight I put a 1/16" end mill on it, and managed to machine a couple gears. The gripper needs some gears for the wrist rotator, and one of them ideally should be molded right in as part of the large round mounting base for the wrist. I didn't know how I was going to do it, but I do now.

The first picture, to the top right, shows a gear machined from 1/8" Delrin, with a D-hole cut in the middle for the motor shaft (Solarbotics GM-14a, which has a 3mm D-hole shaft).

The second picture, on the left, shows a gear milled into 1/4" Delrin, which is how I will build the gear with the attached mounting cylinder. both gears are the same size, which is a little over 5/8" outside diameter. Finally, the bottom right picture shows the wrist motor mounted in the gripper, with the drive gear attached.

*The BrainBot project is directed and funded by the Brain Engineering Lab and Neukom Institute


Wow, a lot happening over the past few days. I got a couple Firgelli linear actuators in the mail today, which fit quite nicely on the 3D printed actuator I got done. On the right, you can see a CAD model of the gripper.

On the left, is a picture of the printed gripper, with the actuator in its place. I can't wait to actually get it hooked up, and try it...

*The BrainBot project is directed and funded by the Brain Engineering Lab and Neukom Institute

Sunday, April 29, 2007

Surface Mount & Stencils

So, this morning I did my first IMU board with a stencil I had ordered from Pololu. All I can say is, wow! What a huge difference. It took about 2 minutes to get the stencil aligned, and about 30 seconds to spread the solder paste (I used a razor blade since I couldn't find a decent stainless steel spatula with a fine enough blade). The solder paste came out absolutely perfect when I peeled off the stencil, and I simply dropped the ATmega168 onto the pads, put all the other parts on, and stuck the board in the reflow oven.

When it came out, it looked perfect. I soldered on the bus connector and the other bottom components, connected the programmer, and powered it up. For the first time with an MLF-32 chip, everything worked perfectly (no solder bridges). I programmed it with the correct fuses and the IMU code, and it works great!

I've done a few of these boards by hand before, and let me tell you, trying to get solder paste onto pads that are 0.5mm apart is non-trivial. Using a stencil makes it trivial...

Big kudos to Jan at Pololu, and if you've got a lot of fine pitch surface mount stuff to do, I would strongly recommend looking into their laser cut stencils.

Friday, April 27, 2007

IMU Mounted

So, my friend Steve has the very first 6-axis IMU I have produced, and he is using it in his Bioloid humanoid, with the standard CM-5 controller and the stock software. It will be interesting to see how that works out, and of course I will update here as it happens.

Steve machined a neat little bracket from Sintra (a type of soft plastic) that allows the IMU to be mounted directly inside a Bioloid body, in between the shoulder servos, and in front of the CM-5. I'm going to be including one of these brackets with each IMU sold.

Steve's store, Bug'n'Bots, sells the Bioloid kits, and he'll be selling my IMU as well.

Thursday, April 26, 2007

BrainBot Gripper

So, my 3D printer friend John from Xardas printed a gripper for me, and I have to say I am blown away by it. The finish is great, and it has all these nice cavities that I modeled in place for running wires and placing sensors and motors. Amazing stuff, indeed.

I'm going to machine the second gripper from Delrin, and it will be interesting to compare them, both in terms of how they look and how they function, as well as how long they take to make.

*The BrainBot project is directed and funded by the Brain Engineering Lab and Neukom Institute

Tuesday, April 17, 2007

3D Printers

One of the technologies we may end up using eventually on this project is a 3D printer, which basically constructs objects using a 3D CAD model, out of melted/extruded ABS plastic. A 3D printer is basically a CNC milling machine with a print head rather than a cutting head, and some specialized software to control it.

A friend of mine (hi Eric!) met up with a guy in San Diego at a conference who has a commercial 3D printer. I checked out his website, and it turns out the printer guy is also in Second Life, which is a 3D world I hang out in, building and selling virtual flying vehicles. I hooked up with the guy, and sent him a CAD model of one of my vehicles, the Marlin XCV. The picture at the top right of this post is what the vehicle looks like inside Second Life.

John (the printer guy) printed the model, and mailed it back to me. I just got it today, and was very impressed... This is what the model looks like in real life (its about 6 cm long, and that's a Canadian quarter next to it).

Very cool stuff, indeed. If you want to get something printed for a reasonable price, his service is excellent - check it out at

Saturday, April 14, 2007


So, all this work I've been doing for the past few months is starting to show some results: BrainBot* is coming together. BrainBot is the Bioloid humanoid I am building for the Brain Engineering Lab. They plan to use it as an AI research tool for vision and object manipulation tasks.

The chest compartment is coming along, and I have the two wireless cameras mounted on their pan and tilt servos. We had a lot of problems getting wifi to work with the wifi module we had chosen, so we have decided to go with using one of the new gumstix boards onboard, which can do USB host. The gumstix provides us with the ability to use a USB interface chip to talk to the Bioloid bus directly, without needing a microcontroller. When they release the wifi module for it, we will be getting one of those as well - for now, we're going to use a USB wifi module. Thanks once again to some help from my brother (and the guys on the gumstix mailing list) I was able to get my USB wifi module working on the gumstix, and I've also got my PC talking to the Bioloid bus using the USB chip.

*The BrainBot project is directed and funded by the Brain Engineering Lab and Neukom Institute

Monday, April 9, 2007

CNC Milling

So, for the past week I've been working with my new CNC milling machine (its not really mine - it really belongs to the Brain Engineering Lab*, but its in my basement...)

Getting it set up was interesting, but after a few issues and a bunch of help from my brother Dave (who also has a CNC Sherline mill), I was able to get it running. I gave myself a crash course in G-code (the language the milling machine understands), and decided the free tools out there pretty much suck. For what I am going to be doing the most, which is drilling and cutting out flat panels, there is a very specific set of steps to follow:

  1. clamp down a blank panel
  2. "punch" the holes using a center-drill
  3. drill the holes
  4. mill out any interior pockets
  5. mill the outline
I found what I typically do is set up the panel I am cutting on top of a piece of 3/4" pine. If I take a #4-40 tapping drill and re-drill a couple of the holes in the middle, I can use a couple #4-40 machine screws to hold the "center" of the part down while I am cutting out the outline (between steps 4 and 5 above).

I found a macro on CNC Zone that runs in Rhino, and produces g-code from line segments (although it doesn't really do all that great a job). It produces the g-code instructions I need (plus a lot of extraneous instructions), so I wrote a little GUI in Squeak that imports a "drill file" and a "mill file" produced by the Rhino macro. This allows me to set a couple parameters, and produce a nice g-code file with comments and the pre-drill stage and so on.

Anyways, I'm having a blast with this CNC stuff, and have started work on building BrainBot's new chest compartment.

*The BrainBot project is directed and funded by the Brain Engineering Lab and Neukom Institute

Sunday, April 1, 2007

First IMU Board

So, I fired up the reflow oven, and did my first "production" IMU board. I like the way it came out, although I had to fix a few solder bridges - I'm definitely going to invest in a solder mask stencil so I can apply solder to ten boards at a time, with very little time required. I can machine a jig to hold the boards in place without too much trouble, since the boards came indivually. These boards have very few components, so placing the components is fairly straight forwards once the solder paste is applied.

I also need to machine a jig to hold the board once it gets to this state, so I can solder the two sensor headers in place and ensure they are exactly perpendicular to the board.