Years ago, maybe 7-8 at this point, I finished building my very first CNC machine. It was not named Shapeoko, you could not build it for anywhere near $300, and it was not assemble-able by nearly everyone. It was in fact, the exact opposite of Shapeoko. But, neadless to say, I finished assembling the machine, and wired all of the electronics. After that I installed EMC2 (now linuxCNC). I tuned the motors and ran the test job. Success! I was in business.
Right away, I wanted to cut my own part. I had already drawn it to scale in CAD and saved the file. Problem was… I kept trying to open it with EMC2 and for the life of me, I couldn’t get it to open. Ends up, there’s a step between the CAD part and the cutting part. It’s called CAM.  I didn’t know that. So off to the internet I went, scouring any sort of information I could on this so called “CAM”.

I’ll give you a run-down on the jist of CAM:  CAM is the process where you (or the software) define your toolpaths. The toolpaths are what turn into g-code. And the g-code is the language your machine understands. The overall workflow goes like this:

Draw Something -> CAM Something -> Stream Something -> Cut Something

Pretty easy right? It actually is surprisingly easy once you find a program for each of those steps that you are comfortable with. The problem for most beginners is they don’t really know what software to use for steps 1 and 2.

Here are my suggestions:

DRAW SOMETHING: Start with inkscape for your CAD program. it’s free, there are plenty of tutorials, and it’s pretty easy to use once you get the hang of it. Yes, you are not going to design the space shuttle with inkskape, but it’s a no-risk way of getting started drawing.

CAM SOMETHING: I’ve looked high and low for a nice open source, entry level CAM program. As luck would have it, after YEARS of searching, I found one mentioned in another forum, and tracked it down to find out it had been abandoned but the source was released before said abandonment. It was created by a guy named Jack, and I’m planning to do an entire other blog post explaining how this came to be. For now, just know that Jack’s creation (which he called PartKam), is now available on my github page, and is being hosted at one of my other domains. You can find it here: www.makercam.com. You will find a basic ‘help’ and ‘tutorial’ link at the top of the page.

However, they say that a picture is worth a thousand words, so we’ll assume that a video is worth a million. So, instead of rambling on more about this, here are two videos representing steps 1 and 2 of the process.

Please feel free to join in on the forum discussion regarding makercam, and if you’re interested in helping improve the code base, drop me a line.
There you have it. Enjoy!

When this project started it was no more than a hobby to me. I suppose once the Kickstarter project was launched, and 800% of the goal was raised, some may have immediately excluded it from the hobby category. But I digress. It was still a hobby to me, albeit a successful one.

Then some stuff happened. The kickstarter project ended and I had 20 units to deliver. As noted in the article written about Shapeoko on the MAKE blog, there was a tremendous amount of demand past the inital 20 kits, to which I complied.  In order to produce these additional kits I spent my “free time” procuring the pieces, making some of them on my own Shapeoko, packing them into boxes and shipping them all around the world. It was fantastic.

Then it was not fantastic. It was exhausting. I couldn’t justify the time I was spending on the project with what I was receiving in return. Well, I could justify it to myself, but had a tough time justifying the rational to my wife and family. Probably with good reason.

That’s where Inventables came in. They took the burden of fulfillment and delivery out of my hands, and in return paid a royalty for each unit sold. It was like receiving back-pay for all the time I had put into the project prior to them taking over. Again, it was fantastic.

I really thought that once Inventables took over the project’s fulfillment responsibilities, my life would go back to “normal” (whatever that means). My thought being I would regain all of those hours that I was putting into packing/shipping and be able to split them into additional time to spend with my family, sleep!, other projects, and supporting Shapeoko.

But that didn’t really happen. In hindsight, it seems like I should have been able to see this problem coming, but you know what they say about hindsight.

The problem was that although I didn’t have to pack and ship anymore, because Inventables made the project’s availability significantly higher than I could have made it, the userbase (which I was still supporting) effectively tripled overnight. So yes, I wasn’t packing and shipping, instead I was browsing the forum, helping to answer questions. I was also helping to get Inventables up to speed on how I was managing the project. Shapeoko was a far departure from their typical product, as such there were some growing pains and a significant learning curve. Plus, the transfer of knowledge took longer than I had anticipated. Mainly because I hadn’t anticipated it at all. I had the misguided fantasy that I would just hand over the Bill of Materials, and be done with it. I was wrong.

So, there I was in July of this year, working just as many hours (maybe more) on project Shapeoko. With both the userbase explosion, the supply chain support, and trying to improve the design I was putting in close to 35 hours a week. This was in addition to the 40-45 hours I was putting in at my day job. The difference between the two being that my day job was more or less done (save a few hours a week) once I left the office. But, because Shapeoko has users in all corners of the world, the project never slept. There was always a new email, a new post on the forum, or a new phone call relating to the project. It was exhausting. I was burning the candle at both ends and couldn’t see a break in sight.

Nate from Sparkfun talked recently about something he’s calling “The Pit of Despair”:

These products get far more attention than the creator(s) expected. The product was well-designed but may have been designed for kitchen production, where they planned to build tens of units. When demand grows beyond thousands, the creators are often forced to make tough decisions: the income may not be enough to release them from their day job, and the amount of time required to build the product begins to gobble up evenings and entire weekends. If you’ve ever had to count out 150 bags of resistors and LEDs, you know what I’m talking about. It’s really painful.

This is exactly what was wrong with Shapeoko. Yes, it was generating revenue, but not nearly enough to support me and my family. While at the same time requiring giant amounts of my time.

As a side note, I want to add the fact that Shapeoko was never intended to make any sort of profit. I started the project to release a set of plans to build the machine, I didn’t start it to actually build and sell machines. [We can explore the details of what changed that in another blog post.]

The situation was this: here is a project that I started, champion, love, find both satisfying and meaningful, but I did not have nearly enough time to dedicate to supporting it properly. The proverbial rock and a hard place scenario!

By mid July, it was clear that I couldn’t do both. Because my income (read: security) was being generated by my day job, I felt that the lion’s share of my time and effort should go there. Shapeoko was struggling because I couldn’t dedicate myself to the project completely. Design improvements were stagnating, the forum (although thriving with user content) needed additional attention that I was unable to provide, and I was afraid the cost of the project was going to increase if I wasn’t constantly searching for new vendors or more efficient ways of doing things.

In the end it came down to a simple choice: what did I want to do? Continue toiling away at my day job, settling in with the security and relative non-risk of that career, or take a risk and do something that I love?

One thing that I have failed to mention up to this point is open hardware and desktop manufacturing. Both of those subjects are included when I say that I love Project Shapeoko. Yes, I love the little machine, and the community of users that has built up around it. But from a larger picture I love what it represents for both right now, and especially the future.  I believe desktop manufacturing is something that is not going away. I see it more than just a fad, more than just a hobby. Desktop manufacturing is to products as the mp3 was to the music industry. Desktop manufacturing has the potential to change the world.

In the end, selling a desktop CNC machine is cool, and it might generate some revenue for the time being, but it’s not a sustainable business. Just like many other pieces of hardware, desktop CNC machines will be a race to the bottom with one company trying to undercut the company that came before it. I may have exacerbated that fact by trying to build/sell a unit for only $300. The real value in the project is what comes out of it. Hopefully better, less expensive machines come out of it. Making it even more realistic to believe that everyone would have something like this at their disposal. But also the software toolchain, and most importantly the end goal is for this and all the projects that come from it, to alter the way people *think* about how they acquire objects.

Like I said, in the end it came down to a simple choice: What did I want to do?

Well, after nearly two months of thinking, I decided that I want to help change the world! So, starting Monday of this week (October 1, 2012), I officially joined the Inventables team as an employee!

Now, instead of doing Project Shapeoko on the side, I’ll be doing it full time. Uninhibited by another job, and with the full resources of the company at my disposal, Inventables is giving me the opportunity to make Project Shapeoko the best that it can be. We have a lot of amazing things planned for the coming months, and I can’t wait to share them with all of you.

I also want to say thank you to everyone who has supported this project. I’m going to do my best to deliver on the promise of a better machine and do my part in moving the desktop manufacturing revolution forward.

Here’s to the beginning of the next chapter in Shapeoko’s history!

Take care,

Edward Ford

For instance, here is a simple coaster I drew with inkscape:

I then saved the file as ‘Desktop Cutting Plotter (R13) (*.dxf).

But, when I tried to import the saved dxf into HeeksCNC, I was met with the unpleasant side effect of not all the paths being joined together where they should be. If you cound the number of sketches in the left pain, you’ll see a lot of them. There should be one sketch per path, resulting in 9 paths total:

1.) Outside perimeter

2.) Inside decorative edge

3-5.) Letters: C, S, T

6.) Outside of Letter  ’O’

7.) Inside of Letter ‘O’

8.) Outside of Letter ’A’

9.) Inside of Letter ’A’

It’s important that the paths are correctly turned into sketches because that’s how we control the toolpaths. Think of the letter ‘O’: If we machined both the outside and the inside in the same fashion, you’d be left with a shell of a letter. Same with the letter A.

That issue can be resolved in HeeksCNC, but proved to be a PITA and not something I would want to have to do every time I imported a DXF file. Or ever again for that matter.

Luckily, the people over at Big Blue Saw have a fix for you (and me!) that will result in perfect DXF exports every time.

Here’s the same inkscape file, saved as Big Blue Saw DXF Output (*.DXF) imported into HeeksCNC:

Notice the number of Sketches! Yep, there are 9 there, just as we would expect.

Here’s the situation: To save you the hassle of having to re-input your grbl settings each time you do an upgrade, grbl instead stores your settings in eeprom. That’s a really great feature, and a real time saver until you don’t want those settings anymore! Then it turned into an exercise in frustration as each time I loaded a new version of grbl (v0.8) my setting remained the same, because they were stored in EEPROM.

What’s the solution? Well, for me it was easiest to just clear the EEPROM with a small script, and *then* flash the new grbl.

The script is from the arduino site (link):

——————————–


/*
* EEPROM Clear
*
* Sets all of the bytes of the EEPROM to 0.
* This example code is in the public domain.

*/

#include

void setup()
{
// write a 0 to all 512 bytes of the EEPROM
for (int i = 0; i < 512; i++)
EEPROM.write(i, 0);

// turn the LED on when we're done
digitalWrite(13, HIGH);
}

void loop()
{
}


——————————–

 
One of the choices is to run the machine with an arduino (grbl) or with a PC based implementation such as Mach3 or EMC2. I’ll try my best to give a brief rundown on EMC2, as it is what I have been using *almost* exclusively for several months. I have little experience with Mach3, so I’ll leave that to another user.
First, we’ll cover some basic concepts:

To run a ‘job’ on a CNC machine you need to do a couple of things.

1.) Draw something in a vector format (CAD, inkscape, Sketchup, etc).

2.) CAM what you drew in #1 (cam means to generate the gcode)

3.) Send the Gcode to the machine

4.) Interpret the gcode into step pulses

The difference with grbl vs EMC2: At the most basic level, EMC2 does both sending/interpreting.

- With grbl, you would use something like gcodesender to stream the gcode text file to the arduino. grbl then would interpret the gcode commands and convert them into pulses which are sent to the stepper drivers.  And the job is run on your machine.

- With EMC2 – you load the gcode file into the interface (usually called AXIS). The gcode is interpreted into a toolpath and shown to you on the screen. This effectively gives you a preview of what EMC2 thinks your gcode is going to produce. When you’re happy with the preview, you click run. After clicking run, the gcode is interpreted into pulses and send over the DB25 cable to your controller (stepper drivers). And the job is run on your machine.

Got it? OK, good. So what’s the difference after this point?

- There is a slight difference in the way that gcode is interpreted between EMC2 and grbl. (i.e. EMC2 is more robust with a larger set of understood commands).

- I also think that given the same job, EMC2 will run it faster than grbl due to the limitations of the 328p.

 
In my opinion both of those differences are minute. EMC2 offers a better interface and better (truer) CNC experience than grbl. But, it’s debatable as to whether that matters at all. GRBL was designed to be a no frills CNC controller capable of fitting on an arduino. It’s simple by design. Whereas EMC2 is huge and monolithic in it’s quest ot be all encompassing to all CNC operations (lathes, 6 axis machines, big machines, small machines, etc. A pretty accurate anology would be: Microsoft Word vs Notepad. Where EMC2 would be Word and GRBL would be notepad. They are both word processors, and both very good at what they do, but one has A LOT more features than the other. Some of those features are handy, some are simply unnecessary for all but the super niche users.

Are you missing anything by not jumping straight into EMC2 or Mach3? It’s tough to tell because everyone’s situation and expectations are different. My general suggestion to everyone is get your machine running with grbl. Get used to running jobs, get used to the machine in general and the workflow that accompanies it. Once you have that stuff down, if you feel like venturing out, go for it! The most difficult part of using EMC2 is procuring a PC with a parallel port. If you already have that, or have access to that, then in my opinion it’s a great choice to use EMC2.

A little over 2 years ago, I posted this to the sprakfun forum. In summary it is a schematic for an ‘all-in-one’ cnc controller that I wanted to call grblDuino. The core of the design was based off the arduino chip (atmega 168 at the time) and 3 easy drivers (atmel A3977). I tried my hand at eagle but in the end didn’t make the appropriate effort or time to learn enough about basic board design to get past a simple (and probably flawed) layout/schematic.

After several months of not doing anything with the grblDuino, I met and became friends with some people who were capable of making such a board. The problem was they didn’t have an interest in it. In fact, nobody that I talked to had an interest in making the board. Their argument was that it wouldn’t be cost effective. After all it’s tought to compete with Arduino, considering their quantities.

The issue I had was that I wasn’t trying to compete with Arduino. I wasn’t trying to make anything arduino compatible. The only similarity was the chip (at the time the atmega 168). But, nobody was down with building something that wasn’t directly arduino compatible. Oh well, as it were, I was happy with the easy drivers, then I was happy with the pololus, then I was happy with the stepper shield from buildlog, then I was happy with the grblshield from sythetos.

Fast forward to present day:

Last month I went to Makerfaire in the Bar Area. One of the people that I met was Roy of Panucatt Fame. You know, the ones who make the Azteeg X1 3D printer controller. Roy and I talked about project Shapeoko and he asked if there was anything the project needed, in terms of electornics. My immediate though was back to the grblDuino! Yes, there are lots of solutions currently available to drive your machine, all stated above, and they all work great. But, it’s always seemed more complicated than it needs to be. And expensive. Without building your own solution, you’ll need:
An Arduino – $30 +
a.) Buildlog shield: $25 for board + $36 for pololu chips = $61 + Arduino = $91
b.) grblshield: $60 for board (build in drivers) + Arduino = $90

For a project that’s striving to hit the $300 mark, every dollar counts! At ~$90, nearly 1/3 of the goal cost for the entire machine is being consumed by the controller. I’m not saying those controllers aren’t worth it, they’re great, and they’ve never let me down, but again it seemed like the cost could be shaved considerably by eliminating the unecessary bits and pieces that weren’t being used.

Enter the Azteeg G1 – all in one CNC controller.

Azteeg G1 CNC controller

Right now panucatt is anticipating a price point between $70-$80USD. As I’m told, general availability could be as early as the end of July!

Some tech specs:

- ATmega 328P
- FT231x USB chip (FT232RL on prototype)
- 2oz PCB RoHS compliant
- 3 DRV8811 Stepper drivers
- 12-30V wide input switching power supply(single supply)
- PTC protected USB line
- Fast Acting 5A Blade Fuse
- RX/TX Leds

Here’s my review of the unit they sent me for testing.

First impressions: This thing looks awesome! It’s the same color blue as the Azteeg X1. The silkscreen is white (and super informative) and the markings are in a very cool silver color. Not sure where the idea came from, but they also put a piece of thick (maybe .22″) clear acrylic over the top of the controller. I want to take it off and engrave the Shapeoko logo on it! The acrylic is held in by 3 screws going into standoffs. The standoffs themselves are an added bonus. I can see mounting this thing directly to the back of the machine. (with an additional shroud for protection of flying debris and to prevent possible shorting issues)

Azteeg G1 - top left view

Testing:

Right now there are only about 2 hours of machine time on this controller (time the controller was actually driving the machine). But, from my testing (mostly single line drawings) it’s been as advertised. It runs exactly the same as an arduino + grblshield. As it should, using the same stepper driver chips (TI DRV8811) and firmware.

Testing Overview: The machine I used for testing has a dual drive Gantry. The Azteeg G1 only supports 3 axis. To prevent disassembling the machine, I plugged both my gantry drives into the Y axis control (taking care to reverse the pairs on one of the drives). No problems there. When they sent me the board, it was pre-loaded with the stable version of grbl (v0.7). Instead of loading the edge branch (v0.8d), i ran with the stable version. I’m planning to load v0.8 soon and run some more testing. A couple of features I’d like to test are the limit switches in action and the spindle control. Panucatt placed 2 Mosfets on the board which can be used to drive an external relay (in the same fashion the buildlog 4th axis relay driver works). Some people might disagree, but I like the fact that they left the relay off the board. This saved some cost, yet still left the option available without too much added work.

Azteeg G1 - Front/Top View

Usability Notes: As with the grblShield, heat dissipation seams to be great. The drivers were set to 1A, and I was running my machine at about 16,000mm / minute. I didn’t measure the temp, but there were no overheating issues as seen when using the pololu drivers without heatsinks + active cooling. I’m told with the production version of the Azteeg G1, 2oz copper will be used, adding even more disappation potential to the layout.
- Setting the microsteps individually per axis was a huge plus. They’re manipulated the same way the buildlog shield is done (with jumpers).
- Screw down terminals made connecting my machine a breeze. They seem slightly smaller than what’s used on the buildlog shields, but didn’t seem to effect inserting the wire and screwing it down.
- There’s a jumper on the board allowing you to power it from USB only. This can be used for programming the board. Switch the jumper back and it uses the 24v supply. Both are clearly marked on the silkscreen.
- I inadvertantly plugged in the controller to the 24v source *before* attaching my stepper motors. As some of you may have learned the hard way (myself included), with a pololu setup this would have *blown* all the chips, releasing their magic black smoke and becoming $12 paper-weights (for very small pieces of paper). With the Azteeg G1, it didn’t matter. After realizing what I did, I simply unplugged the 24v source, then connected my motors, then plugged it back in, and it worked fine. No issues. This is more of a testimate to the TI DRV8811 chips than the board itself, but it’s definitly worth noting.

Azteeg G1 - bottom view

Azteeg G1 - Top Right View

Azteeg G1 - Front Left View

Conclusion

I can’t really find anything to complain about. There was a minor mistake on the silkscreen, but even with that, this is the best ‘prototype’ board I’ve ever seen. Had I not known first hand it was a prototype, I would have guessed it to be a finished product. If Panucatt can really deliver this board for the pricepoint they’re estimating, I think it’ll be a winner. It’s clean, simple, and does what it’s supposed to do. I’ll post a follow up review after the edge version of grbl is installed and has been tested.

People who are down with the Shapeoko feel a little bit like David vs Goliath.

Prior to Shapeoko there was a barrier to get into desktop CNC. Mostly the barrier was price. After price, it was complexity and /or a general obfuscation of all the little inner workings of the machine. Intended or not, it seems that one of the key issues we’ve ‘solved’ is an openness about how the machine works and what it takes to get it working.

Having an open dialogue about the machine, from ‘hey this is great!’ to ‘hey, this could be better’ is really important. It seems that everyone in the community is trying to make this machine better. Either by pointing out errors in documentation, or clarifying instructions, or helping with design proposals. I personally take no offense with any of the design critics or suggestions for improvements. In fact, I’m asking for it! From the beginning I’ve tried to be as honest with people as I could be. Whether that honesty was about pricing, or time, or mistakes I’ve made, or the fact that I’m not an engineer by any stretch of the imagination. Part of the openness of this project is the users understanding that we’re doing this together.

We’ve also said “hey, you don’t need to fab anything to make this machine”. Which is a huge bonus. Yes, there are machines available that you don’t have to fab, but those machines *start* at twice as much as a Shapeoko. For all of the machines that you do have to do some fabrication, they are mostly still more expensive than Shapeoko *and/or* they have very poor documentation if any, other than a few pictures.

So, when I hear people who are promoters of Shapeoko, I can only think they feel like someone was listening to them. or that someone was thinking of them when they designed the machine. Which, I can tell you is true 100%. After all, I am one of them So, when the machine was designed I was trying to make something for me. Make something for a regular guy that didn’t have the cash to buy an expensive ‘turnkey’ cnc machine. Make something for a regular guy who didn’t have the skills to drill precision holes or cut precision square lengths. Shapeoko is for us.

Maybe I’ve mentioned this before but the other thing that really makes all of this work is the fact that everyone has the same machine (more or less). Yes, some people have done heavy mods, others have done light mods, but at the core each of these machines are the same: Basic frame, makerslide sytem, and simple electronics. That commonality among all the users, whether the user is in Australia or in Chicago makes this work. When one person finds something new, or learns something, or makes an improvement. That information is DIRECTLY related to every other Shapeoko user in the community. And people dig that. Hard. It’s like they’ve been put on a team. Team Shapeoko. It is very cool.

I’m really proud of this project and really proud of the community that has sprung up around it. I hope with the 500+ people we’ve added to Team Shapeoko over the last few months, we can really kick the project into the next level.

I can’t wait to see what happens next.

-Edward

Shapeoko, ORDbot, printrbot

But, best of all, I met close to a dozen members from the forum! That was really cool getting to actually shake hands with everyone and put a face with a name so to speak.

I also spent some time with Bart of buildlog.net. Bart setup an ORDbot in my booth and spent Saturday and Sunday talking with folks about Makerslide and home CNC in general. He was printing giant whistles and showing off some of the newer 3D printer controllers. It was really fun.

As mentioned in my previous post, I had taken a bunch of orders back in March, just beforemaking the deal with Inventables. As such, when it came time to deliver the orders I was running really low on inventory and had some part sourcing issues, by the time all of the orders were shipped it was the end of May! The entire time those orders were waiting to ship, I found it difficult to focus on anything other than shipping the orders. Now that those are gone I feel like the weight that back orders create on my well being has been lifted. (I realize that last sentence was a little dramatic!).

It is very nice to just be a user again and helping to drive the project forward. I have several ideas for improving the machine that have been dwelling in the back of my head for some time. Looks like I’ll have some time now to actually test/implement those.

In the meantime the community over in the forum has been moving the project forward on their own! It’s been unbelievable to watch some of the things people have come up with. Here are a few examples.

Improbable Construct is making signs:

(side note: checkout his etsy shop. It is awesome!)

Nio made a v-slot table to lock down material.

This is exactly the type of thing I was hoping would happen once the kits were delivered. I can’t wait to see what else happens.

Take care,
Edward

You may be asking yourself what the hell I’m talking about. How could I still be shipping Shapeoko kits? We inked the deal with inventables on April 1st… Well, let’s just say that I had some back orders to take care of.

Unbelievably, it’s taken me this long to complete all of the orders taken from ordering period March 1 – March 14th. First there was a part shortage, then delays getting the parts back in stock, then one thing, then another. The point of the story is this: That’s why I gave up the order fulfillment business. With work, family, ‘events’, and just maintaining the project, I was burning the candles at both ends trying to get everything done. And when your schedule is stretched that tight, when things don’t go as planned (and they seem to never go as planned) one tiny miscalculation or backorder and it screws everything up royally.

I think if I could have buckled down last week, like I thought I was going to, I could have had them all shipped by the time I left for Maker Faire last Thursday. But, that didn’t happen. In typical fashion I ran out of M5 x 30mm bolts with a few kits to go. M5 x 30mm bolts are my most common, most used bolts, and I managed to run out of them… I should have implemented an ERP system back in January instead of relying on my spreadsheet.

The good news is this: That’s behind us now! Hooray for Time. The bad news is that I have a laundry list of catchup items to take care of. But the good news about the bad news is that I’m looking forward to the laundry list

Here’s what’s on the Docket, in no particular order:

1.) Test the RC spindle.
2.) Test all the different controllers I have with a ‘controlled’ (no pun intended) experiment
3.) Finish up the quick start guide documentation for all of the users about to receive their machines from Inventables!
4.) Finish building my ORDbot.
5.) Implement the alternative z-axis setup we’ve been discussing on the forum.
6.) Implement the dual drive (driveshaft) setup we’ve been discussing on the forum
7.) Finish cutting all of the z-axis plates for the inventables order (that should be done Tomorrow night).
8.) Test EMC2Arduino

OK, so that’s not too bad of a list and it all looks pretty fun, so I’m not going to sweat it. One item at a time and it’ll be done soon enough.

In the meantime, I’ve got another project to start.

Take care,
Edward