I've done some hacking of parts tonight.
It seems to me that fisher's 2x3x3 cannot be made with a Giant Cube-regular combinations or a Regular cube-mini cube combination.


This pic is a modification of the one shown on twisty.
The red circle, is the space that the inner cube must be able to rotate in. In the diagram, this is all fine. But, in a real cube, neither combination allows clearance of the inner cube's rotation on the edge pieces. It works, but there would be a gap between 2 edge pieces. anyone understand this?
It would have to be something more like this without casting pieces


Has anyone built this that can prove me wrong?

There are a couple of points:

First, you can always cut away the offending corners of the kernel similar to what is seen in Figure 5.

Second, either the drawing may not be to scale, or your pieces may not exactly match the drawing. I think it's clear to see that if you extend the "posts" (attachments between kernel and outer pieces), you can easily fit an unmodified cube inside. The circle in Figure 1 shows this clearly. If the circle overlapped the outer pieces, you would simply extend the "posts" until it doesn't.

Check your math, I'm sure you can get it in there.

Sandy

That is not my argument
It can ALWAYS be made to fit,
but if you make it fit so that the inner cube can rotate (which you obviously need)
then there will be spacing between the upper + lower "edge" (2 color) pieces

Yes, I understand exactly what you mean. My attempt at building this was quickly halted a few weeks ago for the very same reason. I thought maybe you guys were using bigger cubes than what is available to me here in Canada.
I don't have a solution for you just yet, but if you find one before I do, please let us know.
One idea is to fill the inside cubies of the mini-cube with resin and then cut down its overall size and corners to make it work. That's a lot of work and I'm not sure it's a sound idea.

Anyway, that's my 2 cents

Jake

NOTE: I'm making the following assumptions of cube measurements based on Tyler's message and the original article:

Key Chain (as kernel): 31mm
Standard Cube (as additions): 57mm

The measurements of a completed standard cubie-sized 2x3x3 should be 38x57x57mm. By Pythagorus, the turning diameter required to turn a slice of the 31mm kernel is:

SQR(2 * 31^2) = 43.8mm

Which I think we'd all agree is larger than 38mm. But don't worry, there are at least three solutions. Here are your options as I see them:

A) Shave down the eight corner and four edge pieces of the kernel that will hit the inner sides of your outer cubies. A straight diagonal shave of about 5mm per corner will do nicely (3 looks like it'll work initially, but you'd have to "round" the shaving to match the turning arc, which may proove more of a pain in the butt). 5mm is almost exactly 5/7th of a perfect diagonal, which is pretty much the ratio we see in Figure 5.

BTW: Thanks for making me figure this out. I was wondering why the shaving didn't go all the way in that picture. The simple answer is... it doesn't have to, so why weaken the kernel stability any more than neccessary?

B) Another option is to use different cubes that will fit without modification. Here's the turning diameter of the 20mm necklace kernel:

SQR(2 * 20^2) = 28.3mm

Lots of room to spare, then. If you use a necklace Rubik and a standard Rubik, they'll fit fine with no kernel shaving required. I wonder about the suitability of the necklace, however. Although it's cute and all, it's horrible to play with in my opinion.

C) Your final, and most distressing, option (as I see it) is to modify your outer pieces away from the standard cubie sizes (while maintaining their sqaure-faced ratios, of course) using some kind of construction material. This would require a whole bunch more work than reading the article (or the math behind it as revealed above) until you can safely visualize that shaving down the kernel is the simplest and (if we can assume Tony Fisher knew his stuff) the best solution.

Sandy

I have discovered a flaw:
This takes into account that the 3x3x3 sides of a 2x3x3 can turn 90 degress, where the others sides, only turn 180 degrees.

If it needs to turn 90 degrees, then grooves must be set up on both sides of corner parts. maybe this was necessary anyway?
i dont know yet, as i have only made 1 piece

perhaps i could try using a 20mm inside cube, instead of 30 (31?)mm cube.
where can i get a micro cube?

i think i will try your choice A sandy
hope it works...

Thanks sandy!!!
it works!!!!!!!!
I cut down the corners (almost in half), and the 4 edge pieces (in half).
seems to fit!
i have made 1 edge piece so far... and it turns properly and does not have any interference.
I will be making all the edge pieces first.
then corners.
then centers.
Since this is my first attempt, the pieces probably wont be all of the same size (edges differernt than corners + center, etc)
but if i make a second it will.

I am using a giant and a Studio cube (so that the kernel is great in the turn) and it has been lubricated like crazy.

Ill post some pics later.

Current Progress:
Kernel completed (i destroyed that 57mm)
1 edge piece fitted + tested + glued on.
I hope to have almost the whole thing done by tomorrow night.

That kernel took me 2.5 hours to design.
1 hour of playing around in my basement.
30 minutes on the computer
1 hour during a movie i was supposed to be watching. I went to the movies with some friends, saw the worst movie ever, so i just designed the mechanism in my head.

The actual building is very hard, and destroys the inner cube. Mine will not be very stable, becuase i will not use milliput to reinforce it. I should, but i wont for this one. If i ever make another, maybe i'll reinforce it.

Pictures of the kernel tomorrow. Right now, it is in the sink getting cleaned up.