X-Carve's capacity for PCB manufacturing

I’ve heard through the grapevine that some people at Inventables are concerned about the X-Carve’s ability to accurately mill PCBs. I’ve only ever seen the X-Carve machine wood, so I’m concerned that the Inventables team may be hiding a dark secret- that the 8020 body isn’t rigid enough for anything but art projects. I would have made this a private message to their technical team if there was a way to do that, but only found this forum. Inventables team, I like your stuff, I want to trust you. But there are quite a few people here interested in manufacturing PCBs with your machines, myself included, that deserve to know whether this can route the solder pads for a SSOP package or if it’s mostly good for carving “Home Sweet Home” out of balsa.

Hope I didn’t come off as too rude, it’s just disappointing to see these being sold for a thousand bucks and only having released videos of it cutting wood.

1 Like

Hi @NathanButler @MichaelUna has the most experience making PCB’s at Inventables and would be happy to answer your questions.

Hi Nathan. I’m the one here at Inventables who’s spent the most amount of time working on milling my own PCBs. When I’m not here, I build little synthesizers and noiseboxes under the name Unatronics :smile:

I’ve been working on other projects lately and haven’t had the time to document my latest efforts. But I can tell you that it does in fact work, and there are some awesome new software tools to help you do that.

Regarding machine stiffness- it may seem counterintuitive, but it requires less rigidity to mill a PCB than it does for wood. Part of the reason is that you’re only removing a very thin layer of copper from the surface of the board- less than .01". And if you want clean traces, you do that very slowly.

Which also means, if that’s what you’re mostly planning to do, the actual cost of that machine could go much lower- $799 for the basic kit since the smaller motors and etc. will perform just fine for that task.

There is a great write-up by one of our customers Matthew White on his experiences milling PCBs using his Shapeoko 2, which is the previous-generation model (and is less stiff, for those keeping score):

He discusses using the open-source program FlatCAM to generate the gcode from a Gerber file. I have used that myself and it works well. I have also heard good things about the PCB - GCode plugin for EAGLE.

I’m overdue on publishing a new blog post documenting using X-Carve and these new software packages to make a PCB, so please excuse the delay. I’ll try to get that documented and written over the next few weeks so folks have a comprehensive guide to follow.

1 Like

Thanks for the quick response, @MichaelUna. would recommend either a 12"x12" or 31"x31" X-Carve for limited run of PCBs? I happen to have large-area projects in mind that would benefit from the large size platform, so that’s what I’m leaning towards right now. I have a LQFP48 package chip I’d like to use on a homemade PCB. Here’s a link to the package footprint: http://www.nxp.com/documents/outline_drawing/sot313-2_po.pdf
The pins have a 0.5mm spacing, meaning the cut between would need to be like 0.2mm. In your professional opinion, is this kind of precision best left to OshPark?

1 Like

In my opinion, here’s where these types of machines fit into the prototyping/small run manufacturing process:

Quickturn PCB makers like OshPark are amazing. I use them all the time. But there are two problems I find with relying on them solely-

  1. It takes time for them to turn around a board
  2. You pay for it first, then wait for it to be fabbed and shipped, then get to test it and figure out all the mistakes you made.

While the cycle times are fairly quick, I find that I am out time and money, and if a board requires 2, 3, 4 revisions that can all add up really quickly.

I use my X-Carve to work out all my mistakes, iterate through various versions of the board, check footprints and routing and component placement, in the same workshop that I design my boards in. I then produce a fully functioning & accurate board, sometimes a few of them. I use those functional boards to then design my enclosures around, work out the firmware, user testing, etc.

Then, when everything is set and I’m ready to scale up to ~100+ of something, I call up a board house and get a bunch made. At that point, I know my designs are ready to go and I’ve worked out all the errors and bugs. I can then spend money getting professional boards produced and know that it’s money well spent.

What’s the downside of this? I had to spend the money up front to get the machine. I had to learn how to use it. I had to get the special PCB milling bits and the copper board blanks. And I worked out a quick process to mill out a board.

What I get in return is that errors and bad design choices cost me $2 and an hour, rather than $40-100 and two weeks. I can work through 3-4 iterations of a board in a single evening if no one bothers me.

And then I can use the same machine to cut a custom acrylic faceplate for the controls, then a wooden box with tabbed slots and cutouts for the external connectors and put it all together. In a single evening.

4 Likes

Thanks Michael, I appreciate the quick replies and honesty!

Anytime!

I was actually one of the first customers when these machines were on Kickstarter, like 4 years ago now, which eventually led to me being in touch with Zach and joining the company. And the primary reason I wanted one was to make my own PCBs for precisely the reasons described above.

I can tell you that in those early days, it was pretty rough. The software was nonexistent and you had to kludge a lot of things to get it to work. It took a lot of patience and probably actually didn’t make sense from a time perspective. But it was fun and I learned a lot.

Now, with a PCB–>Gcode plugin directly in Eagle and a much simpler and more functional machine like X-Carve, you can finish a PCB design and then hold the finished board in your hands in about half an hour. It’s kind of unreal. That scenario I first envisioned four years ago is finally happening.

We’re also working on some new architecture for Easel that would make this process even easier, but that’s a few months out :wink:

1 Like

That’s really exciting. Thanks Michael, you’ve been a great help. Guess it’s time to take a good, hard look at my finances haha. Are you still looking at roughly 4 weeks to ship?

Yep, that’s currently the pace we’re going at, but we improve our efficiency every day. Our warehouse staff are really hustling to close that gap.

Oh and to answer your earlier question above- I have a small 3-season porch as my workshop and I use the smaller 12"x12" machine to do all this. That’s way bigger than any PCB I’ve worked on, and still big enough to make wood and plastic enclosures and faceplates for boards that size. I usually use the 4"x5" copper boards we carry in the store.

@MichaelUna When doing fine traces do you ever have issues with how level the work surface is or changes in thickness of the copper? I noticed ChilliPeppr has a feature for probing the surface of the copper and offsetting the gcode to account for variances in material thickness. Have you tried that?

I really like the idea of being able to design my own boards and iterate through designs as you mentioned. Especially as I’m new to EE and plan on making my fair share of mistakes. I’m a big fan of iterative design.

@MichaelUna I’m very sure I’ll never need to mill a PCB bigger than 12"x12", but I’m curious about how the extra 19" in each direction on the larger platform would affect the accuracy for milling PCBs. Like you said, PCB milling is taking of just a tiny top layer of copper and shouldn’t require a huge amount of rigidity; but, with the longer rails, I’d be concerned about vibrations being a greater issue. Has there been much, or any, testing on your or the teams part in machining fine PCB traces with the 31"x31" platform? I ask because I’d love to have the flexability to use my X-Carve for larger woodworking and art projects and have the option of prototyping precision PCBs.

I’ve just done a couple of small single side test boards with Chilipepper, but the results have been great, easily making 0.5mm pitches. The autoleveling works well, and allows you to take very shallow passes which helps with lateral resolution too. It also helps to have a reliable size v-bit. The inventables ones I received were actually sharper than advertised, and broke fairly quickly due to being too sharp. Honing the tip surface to 1 mil helped a lot.

@MichaelUna What is your preferred method for double sided alignment?

Wait, so Chilipeppr uses GRBL, which is exported by Eagle, to export G-Code, to import into Easle? Or can Chilipeppr run the arduino/g-code shield that comes with the X-Carve directly?

Chilipeppr works with grbl. Not sure yet of any tweaks necessary with the x-carve, but, I did play around with it (just jogging).

I use Chilipeppr for most of my projects, it works well with grbl/gshield. You’ll want to go through the Easel setup at least once, as that will set some important grbl variables I believe, but after that you can switch between them at will (just make sure that you only have one open at a time though).

1 Like

One techninque that helps you get a very level surface is to clamp down a piece of material, then use the machine to mill a pocket that’s slightly larger than the copper board. Then you have a surface plane that’s perfectly level to the machine’s path of travel. The board may have some slight irregularities, but those are usually quite small.

But, if you want to be super precise about it-the 0.9x version of GRBL, which is what all X-Carves are shipping with, there are some cool new features for surface probing an uneven surface and deforming your gcode to match. Check out this tutorial and bit of simple software from Martin2250 that does this automagically: https://www.youtube.com/watch?v=kzXzvcUAuus

The program is on github here: GitHub - martin2250/GrblHeightProbe2: PCB autolevelling toolchain for GRBL based CNCs

For two-sided boards, I use a two-pin registration system like this: Shapeoko CNC Router, Rigid, Accurate, Reliable, and Affordable

You just have to make sure you’re representing the holes for the pins in your designs so you don’t accidentally put a trace through them. The easiest way is to make your outline smaller than the actual board and put the pins outside that border. I hope that makes sense?

1 Like

So, I bought a Shapeoko 3 because of what appeared to be very rigid extruded rails. I sat on that purchase for about 24 hours and then just now canceled it for a few reasons:

  1. The Shapeoko looks hella difficult to use compared to the X-Carve. This is my first CNC, if I’m going to be paying roughly the same either way I’d rather put a few hours into troubleshooting and then be making PCBs, engravings, etc.

  2. You guys are awesome. You remind me of Arduino and I like that. Michael, you’re probably reading this 10 minutes after I post it and you’re on the Inventables staff. Big respect for that, dude.

  3. Shapeoko 3 has a preeeetty long lead time.

  4. Shapeoko 3 has pretty sparce info on assembly and usage.

I’m between a rock and a hard place here, Mike. Honestly, I’ve had my mouse on the purchase button on an X-Carve more than once with all my shipping and billing filled out. If it’s worth it to you to get a sale, I would love to see some pictures of what the X-Carve is capable of in terms of finer pitch SMD PCB milling. Even a picture of the finished product of milling out the solder pads for an SSOP package. I know your whole team is busy, but this is the capability I need to know the X-Carve has before detonating my bank account.

Thanks for your continued help, Michael.

2 Likes

Hey @NathanButler I just put out a “Tip Jar Bounty” on your PCB project to try and get you an example project and video faster.

2 Likes

Oh wow, thank you so much. You guys are the best.