Bob:

It looks like you can set membergro up members to require moderatio n, so by adjusting the newbie threshold you could require that, say, the first 5 posts are moderated .  While it would be more work for you it could catch most of the spammers.  See the section on Post moderatio n at

http://wiki.simplemachines.org/smf/Permissions

Only about 1/3 of your total members have ever posted, so it shouldn't be a big impositio n, and it seems pretty common to restrict new signups.  I didn't see a way to change individua l member permissio ns in my quick doc scan, just for membergro ups.

Kirk

Msimpson, Meshlab will read .stl files, and exports a number of different types so it might have one that works with Autocad.  The list is at

http://meshlab.sourceforge.net/wiki/index.php/Supported_Formats

I installed it under W7; it imported a GM .stl file without any problems.  The operation s I tried all ran without errors (mostly simplific ation filters), and only took a few seconds on an old AMD dual core notebook.  I don't have a way to verify the output here, but at least it doesn't crash and burn.

Art: Sounds like you've read "The Mythical Man Month" if you can resist make-it-better-itis.

Interrupt drivers are a whole lot of fun, the hardware never quite matches the documenta tion (and PC hardware must be much worse), they screw up at the worst possible time, and the ICE memory is never big enough.  Must have been some rocket scientist .

Looks like Heekscad fixed stl import, it loads and generates gcode with the windows version.  Been too many other things the last few months to fire up the CNC router, but it's time to chase the spiders away and try bevels again.

Kirk

Geometros talks about coming step, vrml, and 3ds file formats in their faq, but it sure looks like they died as a company in early 2007 - no mention in the press, no contact info anywhere except for placing orders, and no updates since then.  If you haven't already you should snag their source code just in case.

My cad program will write but not read stl (which is my problem, not yours, of course).  There doesn't seem to be any usable open source cam programs that will work with stl - pycam claims to, but it's incredibl y slow (think days on a quad core) - nothing else I found seemed to work for producing 3D gcode for bevel gears.  While I'm  pretty ignorant about 3d formats it seems like stl is more of an approxima tion than a real model descripti on.  step has an iso standard (fwiw), seems to be pretty popular, and could be a good choice.  Should you ever have any free time and a reasonabl e way to implement it, that is.

Thanks for all the hard work.  I've had gotten more enjoyment from GM than any other recent piece of software, except maybe the microcont roller in my coffeemak er.

Kirk

Check out sprockete er, http://www.idleamusements.com/Software/sprocketeer2.htm .

Kirk

Tolerance (gear point distribut ion, I assume) didn't make any differenc e.  Values from .009 to .000001 all received the same error.  With 1E-6 the shaft hole size didn't make a differenc e, either, in half inch steps from 4 to 0.5, and then 0.25.

However, when changing the number of spokes from 12 to 6 a 4" shaft worked fine with shaft sizes up to 4", and all types except damper (which looked legit; the spokes had a kink in the middle).  I didn't shrink gpd, it was left at 1e-5.

It looks related to the number of spokes, not the center hole size.  I didn't try values between 6 and 12 to see where the limit was.

Carrie will just have to learn to interpola te between the even numbers 

Kirk

It fails on a 60t create wheel; I didn't expect the 15t pinion to work   Perhaps when mobius gears are added?

With 60t wheel and 60t pinion create wheel fails the same way when trying to add the wheel with circle1.  When trying all the spoke types straight and circle2 will add to a project, circle1, triangle (nice looking on the gear design screen) and damper (looks pretty strange) will get the solid creation failed message.  Of the five types only damper looks like an obvious failure on the design screen, not enough space for the kink in the spokes.

Haven't tried gcode on the working gears, busy night at work.

Kirk

Thanks, Art!  I second that appreciat ion.  Looking forwards to more laser and 3D printer pics, too.

Sorry to be a complaine r, but there seems to be another bug, although this is undoubted ly stretchin g GM's purpose.

My intent was to create an 6" external 60t spur gear with circle1 spokes, edit out the shaft cutting gcode, cut an internal 60t gear inside it (with planetary gears to hide the movement), and eventuall y turn it into a clock for one of my daughters (a big fan; she has a GM planetary gearset hanging in her room at U of M).  Create wheel receives "Solid Creation Failed."

DP 10.5, 60 teeth, 4.0" shaft diameter.  Spokes are Circle1, 12 spokes, rim, leg, and boss ratios all at 5%.

The 15t pinion looks pretty cool, with the spokes on the outside.

Kirk

Looks like I didn't do that right and only attached the second picture; here's the first...

Art, so the correct approach is to convince the wife I need a bigger router?  I'll tell her you said so 

We must both have been typing at the same time. Thanks for the fix!

Kirk

Here's a couple cell phone pix of a 4 sided elliptica l with circle1 spokes.  The first is a bit out of focus.

It looks like the 5 sided gear failed after generatin g all the spokes, but before the first tooth.  The error message reference d a spoke, however.

Kirk

Gcode generatio n still fails with elliptica ls.  I adjusted the boss and leg ratios to make the little holes around the bosses go away (and to make them less fragile in cheap plywood), but code generatio n still fails with the 5 sided gear.  Hopefully this time the project file is actually attached.  This version reports itself as 3.31, and was just downloade d and installed .  Machine is W7Pro/AMD quad core.

The 4 and 5 sided version's spokes look different - the 5 has overlappi ng circles with an opening at the intersect ions, the 4 sided touches but does not overlap.  Not sure why or how to make them the same.

I'll reboot into linux and try to cut the 4 sided gear that did generate.

Kirk

Nice picture, John!  The internal/external overlaid gear is pretty cool, too.

Changing a gear once it's placed is a real pain.  Art, would it be possible someday to do something like exploding a gear train - remove all the gears and place them across the top of the screen with no connectio n, say?  Then allow individua l disconnec ted gears to be deleted and use drag and drop to put the survivors back, adding new gears along the way.  "Will it mesh" might make this trickier, but it already allows incompati ble elliptica l gears (added a 5 tooth to a 21 tooth OK, but the pinion won't rotate onscreen) so it could just be "designer beware".  Way down on the priority list, of course.  Or is this too big a project? 

John, do you have any photos of finished projects you could post in show and tell?

Kirk

That's my preferred approach, and seems like a good way to handle decorativ e cuts that won't fit.  Lots better than silent errors   Great progress on GM!

Kirk

GCode generated for my spoke test.  Thanks!

Hopefully I can get to the machine tomorrow and try some sample cuts.

Kirk

I don't know if this is a different problem, but I finally made time to do some work on the router.  When trying to generate gcode for the attached project I get errors "Gcoding failed on Spoke 5. Check tool size.", and a similar error on the second gear.  Tool size is .06, and I've adjusted the boss and leg widths to remove some very small holes where the spokes intersect ed the bosses.  I'm unable to see anything that would cause the problem.  Circle1 spokes (a hit with the black walnut provider, btw).

I tried to do a sell job on a laser to get a smaller cut width, but that didn't fly with the managemen t.  Not willing to push to far, since the planned summer lathe and milling machine acquisiti on will cause enough discussio n...

Kirk

What are the rules for setting the tolerance?  I'd assumed that it was like stepover, around 1/4 of the bitsize is a good number.  Since I mostly use a 1/16" endmill it seems like 0.01 is a nice number (yes, I know it's not 1/4 of .0625, but it's easy  ; is it?  Lower tolerance must mean a bigger number.

Not sure what it was set for on the machine that crashed - that was the 17" W7 notebook I take to work - but I'll check.  Too many computers here...

Kirk