Daniel's Work Thread: Toward a Hackerspace Scanner

[jck57]

you could always retrofit in situ to Jck57's method since the bearings are on the outside of each arm.

For my bearing pocket scheme, I would put the bearing on the inside of the joint. That way it would require a shorter spacer (a couple of flat washers). No bearing in the cradle carrier. The nut against the bearing would be recessed in the arm, but I don't see a problem with that.

A note about pressing in ball bearings: you don't want to hammer on the inner race. Find a socket that is slightly smaller than the outer race and hammer on that. Better yet, put the socket and bearing assembly in a vise and push it straight in. You can also use an extra bearing instead of the socket and it won't hurt the bearing because you aren't pressing exclusively on the inner race. A press fit is much better than a loose fit. You don't want the outer diameter of the bearing moving around in the bearing pocket.

[daniel_reetz]

Here's the design from the cad I was born with:

HA HA!!

[rob]

I think jck57's idea should work -- I'd like to use the same washers as we currently specify for the washers that hold the bearing in place. I'm also amenable to epoxy, but that can get very messy very quickly if too much is used. Imagine the wet epoxy getting all over the bearing.

[tresho]

I'm also amenable to epoxy, but that can get very messy very quickly if too much is used. Imagine the wet epoxy getting all over the bearing.

Epoxy putty is rather sticky, but is not at all prone to running. There is also epoxy paste, of a viscosity between fresh epoxy glue and fresh epoxy putty.

[daniel_reetz]

I'm not sold on the one-bearing idea... but I could be swayed if someone tried it and it was great.

Just bought the hall-effect sensors to make my CNC home to exactly the same place every time. Then I need to make a z-plate to zero the Z and I'm all done and ready to do... tons more testing on methods to reduce cutting time.

[daniel_reetz]

I've done quite a bit this weekend. Last night, I read Furniture Manufacturing In the New Millenium cover-to-cover. It's an interesting book, written by the president of Thermwood corp, that discusses all manner of issues that come up around trying to make furniture in hundreds of units instead of thousands. You can see an image of a Thermwood router in Fab52's post here.

I took home three major points from the book:
1. Setup time is the most expensive thing in production. Reducing setup time reduces all other costs.
2. Circles in CNC/CAM programming are bad. Eliminate circles.
3. Where optimization will have the greatest effect, optimize.

With that in mind, I spent the morning evaluating DIY Hall Effect homing switches. In particular, I copied the electronic design in this post. Unfortunately I burned up the single hall effect sensor I had laying around (an Allegro 02E). Instead of wasting more time on this, I simply ordered these nice, prebuilt sensors from DigiKey. Sadly, with DigiKey the shipping costs almost as much as the parts themselves. I expect them to arrive mid-week.

I will be able to install those sensors on X and Y axes, and use a touch plate for the Z axis. With that kind of setup, the machine will be able to be set up with fixtures drilled into the bed, so I can easily position plywood in a repeatable fashion.

Today's remaining plan of work:

2. Download Bob Warfield's G-Code Wizard and check the theoretical maximums of machine feed/speed.
3. Start optimizing parts for the scanner - in artwork - one by one, with the most effective optimizations first.

[jck57]

I'm not sold on the one-bearing idea... but I could be swayed if someone tried it and it was great.

If someone cared to try the one-bearing idea, and that person happened to already have a Hackerspace scanner, he or she could simply remove one of the bearings in each joint in each arm and then secure the remaining bearing with screws and washers. For testing, it wouldn't be necessary for the washers to have pockets.

[daniel_reetz]

True true. I don't want to totally disassemble my one-and-only test unit right now, but I do have some extra arm pieces that might be amenable to playing with the idea.

[rob]

Here is my one-bearing thing. It should work with any currently existing two-bearing versions if you remove one of the bearings. I used a 2" bolt and two nuts (add a washer or possibly two before the nuts if you're using the two-bearing version). Can you spot the ladybug in the picture?

IMG_0159.JPG (93.35 KiB) Viewed 9606 times

Then, I put on the arm with the bearings facing outwards. These are new arms that I cut which have pockets as deep as the bearing. I had to use a vice to press them in. These things are tight -- if I sell these, I'm definitely going to pre-install them.

IMG_0160.JPG (91.62 KiB) Viewed 9606 times

IMG_0161.JPG (86.65 KiB) Viewed 9606 times

A washer and nut complete the thing:

IMG_0162.JPG (88.75 KiB) Viewed 9606 times

And it works! The arm cannot come off away from the scanner, because that would make the bearing go deeper into the pocket, which is not possible. The arm cannot come off towards the scanner, because eventually the arm would hit the support.

The only issue is that if you make this out of soft wood, the bearing will not be as tight, and the arm can loosen -- but it will still not fall off, it will just rub against the support. So maybe epoxy for scanners made out of anything softer than baltic birch.

[daniel_reetz]

Today's remaining plan of work:

2. Download Bob Warfield's G-Code Wizard and check the theoretical maximums of machine feed/speed.

Well HA HA HA at this. Even the free trial has some kind of web-based activation system, and the software phones-home to check if you've clicked an activation link. It's also running inside Adobe Air. Not real pleased with that whole setup... and not able to use it after half an hour or so of messing around. I'll just move on to the next task:

3. Start optimizing parts for the scanner - in artwork - one by one, with the most effective optimizations first.