Last Thursday was a very exciting day. My Other Machine arrived…Oh Happy Day!!!
This is the story of how I got from here:
After the dust settled:
I can roll my own prototype PCBs. Ah the YIPPEE!!! moments are piling up. FANTASTIC.
The goal is to master making prototype PCBs on my Other Machine. By doing so, I have added a valuable tool that will stream my design to finished product work flow.
Thanks to Those That Went Before
A HUGE thank you to Simone, the wonderful person at the other end of my endless stream of support ticket entries. Simone is knowledgable and evolves the role of support from something that sounds like a necessary prop to what it really is – the role that makes or breaks the success of a business. We’ll see if Other Machine Co. gets my vote as a company I fanatically hope for success for the value and happiness provided – as I do with OSH Park. With Simone they are on the right track. When I got frustrated Simone pulled me out with patience and spot on advice. I also thank the creators of the Other Machine. You have brought us a significant leap forward in our ability to prototype hardware/software efforts.
Kicad schematic and layout files are located at this GitHub location.
I found out through the Ladybug Shield project that there are smaller sub-circuits that could be immediately proved out and tested. Yes, this means there is a new schematic and layout in Kicad that needs to be done. But on the other hand, the schematic side is simple. Doing the layout on simplified sub-circuits is good practice in getting better laying out PCBs.
Oh yeah – I can also make PCB stencils. There is an instructable that outlines the steps. As the Ladybug Shield stabilizes the stencil and the reflow oven will be used more. I tried stenciling earlier (see this post) to see what it was like. It is definitely worthwhile if reflowing many boards. What I just don’t understand is why the copper pads and/or the parts don’t come with sticky solder so that once they are placed they can be popped into the oven and toasted. I mean, we have double sided tape… I assume there is a chemical reason why this is a non-starter…chemistry is fascinating but not something I have knowledge in.
There are so many reasons why I will use both the Other Mill and OSH Park. I’ll cover this in another post. The main reason is the quality and sophistication of the PCB layout (vias, small track widths, finishing) are far superior.
For this effort I am creating a PCB that will talk to the Arduino running the Ladybug shield and turn on/off a pump. In the current Ladybug Shield design, the pumps hang off one side of the Shield. After trying out the shield, I found having both the BNC connectors for the probes and the pump cabling to be very crowded, particularly in the nutrient bath where the pumps and probes would be co-mingling. I decided to split the pumps off the shield. Each pump will include a small board with the circuit I use here. The pump will have a wire running from the Ladybug shield so the Arduino controlling the shield can send pump on/off commands. Here is an image of the schematic (kicad files available at this GitHub location):
It’s simple – don’t you get it? It’s right there on the board…
That “simple” statement was a defining moment in my academic career. It happened when I went up to my professor and told him I did not understand the triple integral expression he had carefully crafted on the board. I give him credit for being enthralled and captivated in his work. However, I did not appreciate the annoyed look and condescending comment.
Why mention this? Because I love learning stuff I am clueless in – particularly in the design and building of hardware and software – and developing some level of mastery. I arrive at this particular learning opportunity with no knowledge of its language, culture, or context. I visit. I learn a new language and a new skill. I had never used a CNC machine. I vaguely remember way (way) back in high school that when I took Home Economics and boys took metal working, there were milling machines…hmmm…but my vision was clearly focused on sugar plum cookies.
Anyways, when I first started using the Other Machine among the many oops! moments I managed to break two of the end mills:
I bumbled through over 11 tries and submitted (too many to count) support tickets before I felt I had enough knowledge to feel confident I could actually create a PCB that I could populate with SMTs.
The goal of this post is to add the Other Machine into my workflow by improving my ability to build prototype PCBs. At the end of this post I will have gone through many trials. Along the way, I will build and test a PCB.
The Workflow from KiCad to Other Machine Milled PCB
Loading the Gerbers and Drill File
- can load Gerber and drill files
- run well on my MacBook
and…wait for it…it worked!…well, good enough. I had one challenge but it was not a showstopper.
When making plots in Kicad’s PCBNew app, I chose to plot the top copper, bottom copper, and edge cut as well as generate the drill file.
While I could use the Other Machine to make double sided PCBs, the Ladybug Pump PCB has the circuit on the top side copper only. So the bottom copper is not needed. I included it because well, what the heck. the Other Machine’s software (OtherPlan) reads it and I will probably do double sided PCBs in the future (those darn vias…). It’s good practice to get in the habit of including the bottom copper early on.
Kicad creates the files with the following names:
where .gbl = bottom copper, .gtl = top copper, .drl = drill file, .gbr = edge cut
Otherplan wants the file names to all be the same. So I renamed them:
I was not able to get Otherplan to recognize the board outline (edge cut) file. I tried renaming the .gbr to both .gko or .gm16..but the post on using Gerbers also noted:
I opened up LadyBugPump.gtl in a text editor but did not find any association with an outline layer…maybe that is the problem. I do not know for sure. I forged ahead because Otherplan kindly created an outline and all the other files read in correctly. Let me pause for a YIPPEE! moment.
Matching the Layout to the End Mills
When learning a new tool and process, my quest revolves around getting into the heads of the folks that built the tool – both hardware and software. All of us come at a work flow from a different background with different ways of approaching getting through a process. Given that I have never used a CNC machine before, it was logical to me the folks that built the Other Machine’s idea of an intuitive work flow in the creation of a PCB would not be initially intuitive to me. Thanks to Simone, I now at least feel I am on the expected path.
What I needed to “get” was to think about spacing between what would become the exposed copper – the pads and tracks. The spacing between the exposed copper parts becomes part of the layout as well as the choice of end mill(s). When I send Gerbers to OSH Park, their constraints:
- 6 mil minimum trace width
- 6 mil minimum spacing
- at least 15 mil clearances from traces to the edge of the board
- 13 mil minimum drill size
- 7 mil minimum annular ring
OSH Park’s minimum drill size and minimum annular ring
allow me to be very flexible and more complex in my layout.
I am next to clueless when it comes to picking the right end mills. For this effort, I relied on Other Machine’s guidance:
Requires a 1/32″ or smaller end mill
Requires a 1/64″ or smaller end mill
Requires a 1/100″ or smaller end mill
Luckily, the “Not compatible” is above my soldering ability!
This is obvious but new to me – end mills come in all sorts of sizes and shapes. These sites helped me get up to speed:
- The Skinny on End Mills – an intro from Make Magazine
- The shapeoko project has a great wiki page on end mills.
I had ordered the electronics bit bundle which included 2 1/32” and 2 1/64” end mills. A challenge I had was figuring out which bit was which size. Obviously, putting the tip under the magnifying glass would reveal the difference. The other way (that Simone recommended) was to use a caliper. I would have prefer the end mills to have a distinctive mark…oh, it could be something fun – similar to a pencil caps:
that cover the cutting side of the end mills. Besides identifying a 1/64” from a 1/32”, the caps would protect the end from breaking off. A sad moment for us, but a ca-ca-ca-chink moment for companies that sell end mills. OK, it could be as simple as the itzy-bitzy writing on SMT resistors (although frankly, that reeks of seriousness).
Given that I will most likely be breaking more of the end mills and those that don’t will wear out, I am a bit concerned about the price of end mills. Ah – the razor and blade business model! Is the razor/blade model a requirement for things that shave off?
Drill bits are such a commodity that I can go to a hardware store and have a choice of high quality at a reasonable price. End mills for us home PCB makers using a CNC machine don’t share this advantage. Well, maybe they do but I don’t know about it. I assume the end mills used for PCBs are used for other CNC carving efforts…
I also do not have a grasp on the features and quality of the end mill. For example, Other Machine’s end mills range in price from $13 to $21 (link to store page). Adafruit sells end mills for $4 (link to one of the offerings). Why the difference? As it was explained to me by a folk at Other Machine:
End mills go for anything from $3 to over $40+ a bit, and they are not all created equal. The ones that Adafruit carries have longer shanks, meaning they’re not meant to travel sideways. They are brittle and specifically for drilling holes. The ones we carry will last up to 2000 inches on pcb before they start to get dull. Also, we special order the 1/64th’s and 1/100th’s with shorter shanks so they will be less likely to break.
Given that the likelihood of me breaking end mills as I learn is higher, I’d probably benefit from getting Adafruit’s. For now I will stick with Other Machine’s. My buying habits will expand to other offerings as my skill and knowledge improve. Perhaps asking folks what end mills they use is a good conversation question at a maker event.
Layout and Size Matter
The PCB’s layout partners with the end mill’s size to determine if there is enough spacing between the pads and tracks so that the copper areas that should not be connected are separated.
Here is a layout I had done where the track and pads are too close together (given the end mills I had):
Track and Pad Too Close Together With 1/64” End Mill
Here is the layout I ended up with:
If I left the Otherplan software to cut this out, I end up with:
I wanted to cut out the copper in the areas that do not have exposed copper so to get to this:
Simone once again to the rescue. She advised me:
- go to Otherplan > Preferences, and click on the BitBreaker Mode box. (yes – bit breaker – sounds like my moniker!) Once you’ve done that, in the imported plan file window for your Gerber you will see a button called “Advanced Options”.
- Click on “Advanced Options”, and you’ll see a line item for Trace Clearance. To make sure all your copper gets cleared, just set Trace Clearance for something huge, like 2 inches. This will get rid of pretty much all the excess copper on your board.
Keep in mind though, that depending on the tool you are using, this could take a while to cut! For example, if you have a 1/16” flat end mill selected, this will clear the big areas faster that the 1/32”.
That did the trick!
I am thrilled to add the Other Machine to my tool set. I must say this electronics learning hobby gets expensive. But on the other hand, it could be much worse. I could be paying what I am for my kid’s educations :-).
I now have multiple techniques to use to prototype circuits from break out boards on bread boards, SMT parts soldered to a SMT-to-DIP board and placed on a bread board, to the Other Mill..and finally, Osh Park. I can see a use for each one in a continuum of prototyping needs.
I guess that is what next, an ongoing prototyping effort that leads to useful “stuff” like the Ladybug Shield.
THANK YOU for reading this far. Please find many things to smile about.