The 1" Kozo A3 that never was (a bit of a CAD rambling)

[thunderskunk]

So... it's been a crazy year.

I tried doing my own business. It didn't work out, and I made just about... oh .002% progress on my G5C pacific project. Mainly made a nice pile of scrap in between a lot of other work and the not-so-occasional farmer with broken this and that (you'd think I'd get a break during the 9 months of winter up here...)

I've had Kozo's A3 book on the shelf for a while, and during some "hurry up and wait" scenarios, I went ahead and modeled it in Inventor... 2020... which in 2019 still had a lot of bugs. If you're a CAD guy, read on, you might think it's interesting. If not, and you were hoping there was some hope for Pennsy A3/A5 drawings in 1" scale, there is nothing for you here and I apologize.

First off, why CAD? The answer is because it's what I enjoy. I'm a CNC guy, and most of my experience and skills lend towards that. If it takes you 1000 hours to pump out a Kozo engine on a bridgeport, but takes me 700 hours of CAD and 300 hours of machining or more to make the same product, I still prefer the latter. If you disagree, dial calipers are better than digital ones, bite me

I put my head under a rock from 2015 to 2018 or so; no access to CAD besides OnShape. I drew the entire CP G5C in Onshape, all in a single part file. The web-based program then cut back the data you could use at a given time (about the same time private designs for free accounts were disbanded) and my design became virtually useless. Did I learn a lot? Heck yea, and it gave me skills that helped me on the professional side.

I did the same thing in Solidworks when I had access to it. Then I didn't have access to it, and again my drawings went down the tube.

Finally landed on Autodesk Inventor; which if you're involved with engineering students (ie robotics mentor or educator) you can get an educational copy for free. This time, instead of trying to reinvent a pacific, I wanted to draft the Kozo A3 and convert it to 1" scale. I'm sure somebody out there has one, but I haven't heard of it, it sounded like a fun concept, and thought it might generate a lot of interest. The plan was to combine parametric modeling with this new multi-body part file thing that's become more predominant in CAD techniques. The advantage to this is you can put the data into a table then simply scale every dimension using a macro. All of the part files would self-generate, and I'd be able to catch those pesky linkage errors and fix them much easier. I could also adjust for standard stock sizes without having to adjust several part files as it's all driven by a single dimension.

The pictures you see below are the result, and that is indeed a 1" scale model. So, it was a success and I actually made a few parts; a single wheel and a few linkages. The other big change; I added up what it would cost to make the thing out of brass, and for raw stock it was something like $3000. No thanks, I've got enough steel under the bench, so that's what most of it was going to be.

Then the trouble started. Apparently the beta version of inventor had a memory leak. Every time I made a change and rebuilt the model, it saved itself over itself. What should probably be less than 10mb is pushing 200mb, which is 190mb of bad data. Each change takes about 15 minutes of reloading, and I have quite a few changes to make, ie I haven't even started putting in bolt holes. What I have is a single really big garbage file.

I have access to Solidworks through an employer again, but finding the chance to model something yet again... it's less of a logistical problem and more about finding the motivation. I still have a dead business to revive as well, which isn't breaking the bank but it's hard to justify working solely on a locomotive when I have customers.

Maybe you'll find my rant interesting. Most are wondering why the heck I didn't just start building the bloody thing, but the reality is not that I'm not making chips, just that I'm not making chips working on my engine. The lathe needs a rebuild, the mill spindle has problems I'm pretending don't exist, the wife needs help with the kids at home, farmers keep breaking stuff, onward and upward. I always look forward to the input of our forum, good or bad, so fire away.

[RET]

Hi,
Just my two cents, which is worth even less these days.

CAD and CNC are definitely worth it, especially if you need multiple parts and/or if you need accuracy to tenths. The other given I think is to use software that you OWN and both CAD program and part files that reside on YOUR computer.

On the CNC side, I started out with CAM by buying a little turnkey CNC mill from Sherline. The only thing I had to supply was a monitor. After using the mill for 5 or 6 years, I decided I needed something bigger (twice as large 16" x 9" x 8"), heavier and more accurate, so I built my own CNC mill using 1" dia. Thomson ball bushings, support shafts and 5/8" machine tool accuracy ball screws. While expensive, this machine works very well and is well worth the time it took to make it (about 6 months give or take). I used the digital readout on the Bridgeport to make the mill to the accuracy I needed (all the parts just bolt together with socket head cap screws).

I have 2 CAD programs, Anvil1000 which is a very accurate, intuitive 2D program and Synergy which is a 3D CAD program. Mostly I find I can do what I need with Anvil, but on occasion, I will use Synergy which has a post processor for EMC which the Sherline software uses. Synergy has up to 5 axes, X, Y, Z, A and B and the Sherline software uses X, Y, Z, and A. EMC2 is versatile enough to do subroutines which can make G code programming easier.

The CNC mill has added a lot of versatility to the shop. Now I can use the machine (or combination of machines) that is best suited to the job I am doing.

Richard Trounce.

[datman]

Hi Richard,

I am interested in CNC and have been thinking of making my own, do you have a picture of yours?
I will be building a BR55 hopefully and still drawing it up, I have a thread on it. I am using 3d software to model it.

Thanks
Ian

[kcameron]

A comment about Solidworks, if you are a veteran, they offer a very cheep license. $20 a year as I recall. As you found many of the programs also are cheep/free if you have an education email address.

[thunderskunk]

My problem is definitely not with my CNC, haha.



What I did was pick my favorite brands from working in the industry to use in my own shop; every investment from the machine control to the style burr knife was of considerable debate. I have about 250 square feet for the entire CNC side of the shop, so I needed every trick in the book to keep things at least somewhat organized.

I could spend a few pages explaining it all; the probe you see in there is self explanatory. I programmed my own macro and it’s payed off a hundred fold. You see the file organizer hung on that box? Came from the several hours I’d wasted in another shop looking for files that should have never been an issue. My tools for loading tool holders are all in one corner with the holding fixture right there. Every time something moves, it’s usually less than 3 feet away. I even bought a CMM for a song and dance, though it’s a bit of a project as you get what you pay for.

I’m very proud of my shop, and while it costed some money, it’s not as much as you might think. Most of the best investments I’ve made have costed a fraction of what it did for the shops I stole the ideas from. It costs nothing to have things organized well. If I could do it again, I would have gotten a Trak of some sort. While that robodrill smokes the alternative, my spindle sits a lot longer than I like. I have a beautiful 5th axis that just doesn’t see enough use.

[thunderskunk]

A comment about Solidworks, if you are a veteran, they offer a very cheep license. $20 a year as I recall. As you found many of the programs also are cheep/free if you have an education email address.

So I used that in 2018, and while it was less than ideal it did function for a hot minute. I tried it again recently and haven’t heard anything back. I’m wondering if the person who approves those applications is on COVID staycation.

Solidworks is 500% better for this use case versus inventor. With its excel based design table, I don’t need a bleedin macro to do my scaling craziness. Configurations are also a lot easier in SW. Inventor does have its strengths over SW though. It’s interface for multi body part files is way better and has certain extrusion capabilities that Solidworks just doesn’t have. They both suck at 3D sketching. I think that particular feature has come a long way since I got into this mess, but it’s just not there.

I know somebody somewhere is going to say “what about F360” to which my response is the above. None of these features are even a thing with F360, and the user interface is akin to Fanuc controls; designed by somebody who’s still using windows XP.

[Mike Walsh]

... the user interface is akin to Fanuc controls; designed by somebody who’s still using windows XP.

Ha, you summed it up right there... Which seems fitting since you've experience with a Robodrill

[RET]

Hi Datman,

Yes, I do have a number of pictures that cover the building of the mill pretty well, probably enough for a thread on the subject. If there is enough interest, I could put it's construction on chaski. The travel for the different axes is X=16",Y=9" and Z=8" which is big enough for most model work.

To put things in perspective, I would call it a heavy tabletop machine, with the emphasis on "heavy." The column and overarm are made from 2" x 5" solid steel bar since I had it handy and the table is cut from a piece of 3/4" x 7 1/2" steel plate which I had to buy. Cutting the table slots was the hardest part of the construction.

Richard Trounce.

[datman]

Sorry this a little of topic.

Hi Richard,

I was thinking of using epoxy composite for the frame. It would be great to have a thread on it!

Ian

[thunderskunk]

Sorry this a little of topic.

Hi Richard,

I was thinking of using epoxy composite for the frame. It would be great to have a thread on it!

Ian

Man, talk about gettin hijacked.

I’m pretty confused, why would you want to use a composite to make a mill?

I haven’t made my own CNC machine; I don’t have much a need to. Possibly a plasma table someday, but not any time soon. If I were you (which I’m not) I’d be making any CNC machine out of the toughest, most rigid stuff I could get my hands on. In my case, I’d use steel plate and channels all welded together for the “casting.” I’d get a set of linear rails for ways; if it works for Haas... I would take a hard look at the biggest thing I would ever make, determine if it could be done in a few different clampings, then make the whole thing as “small as possible.” I’d want it to be as rigid as possible.

Even EDM machines, which in comparison to a mill or lathe have little to no forces at the tool tip, have heavy castings. I’m not saying your wrong, obviously it’s been done, but I am curious as to why anything other than steel would be used to make a machine.

[RET]

Hi Ian,

Thunderskunk is right. When you are designing machine tools, you design to minimize deflection, not for strength. That is why machine tools have heavy castings, etc. Cast iron is also used a lot because it won't gall and seize at all, it "self lubricates" because of graphite in the grain structure. While FRP materials are strong, they deflect a lot which is the last thing you want in a machine tool. Even if you used a lot of it, the resin "gives" way too much. You need materials that have a high Young's Modulus (stress vs. strain). For steel or cast iron that is 30,000,000.

All of this is why I used the materials I did. At the moment I have to decide if the CNC thread should be in the CNC section or here. It could be here because it won't be that long. We'll see.

Richard Trounce.

[Harlock]

If you try again in Solidworks, might I suggest using a traditional set of part and assembly files, rather than a single multibody part. Then all your problems down the road will go away. You could probably still do dynamic scaling with tables if you really wanted to. The main issue with that in practical terms is that you would want to adapt some of the parts to standard stock sizes, and a direct scaling may not play out well in some cases.

As a second suggestion for CAD software, give Alibre Design a try. It is a solidworks knock-off that is also a true feature-based solids modeler. I found it similar enough that it was easy to use when I tried it out. There is a free trial as well. https://www.alibre.com/ The hobby tier software is in the hundreds of dollars rather than the thousands as in the case of Solidworks. For a while the product was unavailable, but has since returned to independent status after being subsumed by another company.

I have a standalone bought and paid for (by an Aerospace company I used to work for which is now defunct) version of Solidworks installed on a machine here but without paying a lot more money I can't transfer it to my new machine, so I've been thinking of switching to Alibre. It seems to open my solidworks 2016 part files just fine, so there is a path forward.

Regarding 3D sketches, the only time I've ever needed to use those was modelling a composite airframe with a very complicated shape. I've found little need for 3D sketches in railroad-land. Everything can be made with 2D sketches and resultant features.

Best,

-Mike