4. Electronics Production -working on it

This week we were tasked with making and testing a Microcontroller Development board. In the new Fab Academy 2024 schedule, this is the first (of many) weeks dealing with elctronics. This week’s assignment is all about the physical construction of electronics boards, meaning cutting out traces on a Copper Clad Board and attaching electronic SMD components. It doesn’t deal with the aspect of designing a board or programming a SOC (system on a chip). Those will come in later assignments.

Group Assignment

This weeks group assignement is to “Characterize the design rules for your in-house PCB production process”. And also “document the workflow for sending a PCB to a board house”.

I am a continuing Fab Academy student; me and my group all worked together on the group assignment last year, and the results are here: Electronics Production group assignment.

However 2023’s group assignment didn’t have the PCB board house component, so I documented it one the bottom of this page: Workflow for sending a PCB to a board house

individual Assignment

“Make and test a microcontroller development board”.

development board

I won’t be designing the board myself. Instead I made the Fab Academy’s recommended board, the Quentorres RP2040 hello board, a development board based on the XIAO RP2040 microcontroller. I won’t be making the Quentorresaurus board just yet. The purpose of a development board is to be able to easily connect “things” to your microctroller to try programs on it. The “naked” SEEED rp2040 board would be difficult to use without building a larger board around it since its connection points are small and very close to one another. Making a development board helps allleviate this problem.

Here are the original files given for this test board:

milling the board

These are the I had allot of problems developing the toolpath for milling the board. It should have been a fairly straightforward process given the design.

Problems with toolpath

adding components to board

• List parts swapped out.

Lessons learned and future plans

• The default trace lines given in most design programs are often thinner than needed, and sometimes closer than needed. This can lead to:
  • broken traces when milling, because the traces are so delicate and/or your fine endmill is a little dull.
  • Areas not being milled. This because they are so close that the endmill can’t cut the space between two traces without cutting the traces themselves, so they just don’t cut at all.

• When using the design software in future weeks’ assignments, I intend to increase the trace sizes wherever viable, and increase the gap between traces if I can. If this can’t be done automatically in settings, I wil have to do manual edits.

• Although I didn’t design the board used this week, I should have edited the trace schedmatic when I saw certain issues pop up:

  • 1 gap didn’t get cut. This was the gap between the pad for PIN 1 of the Xiao 2040 board (top left) and the diagonal trace above it. In making the board I left this uncut by the machine, and instead cut it manually with a hobby knife afterwards. In retrospect, it would have been easy to move the diagonal trace to the right and give more space so that the endmill could have cut it.
  • There was a bigger issue with the milling software deciding to mill over 2 sections that were not supposed to be cleared out. I spent a day and a half trying to figure out why, but to no avail. We run our Fab Lab here, so if we intend to teach this topic I wanted to be able to solve this problem rather than use a workaround (which is what I had to end up doing). I ultimately couldn’t find out what went wrong without spending even more time on it, and I had to resign to using a workaround by drastically changing the shape of the problem traces, while still being functionally the same. I still never figured out why Carbide Create pocket toolpath wasn’t detecting a few zones of the design properly, because contour and texture toolpaths detected them fine.

• Use bigger pads and contour toolpath instead of pocket toolpath. Initially I chose pocket toolpath because I don’t have much faith in my fine soldering skills. Pocket toolpaths are more forgiving for poor soldering, because if you spill over, you are likely spilling into empty non-conductive space; with contour toolpath, if you spill over you are likely making contact with conducting copper and creating a short circuit. The problem with pocket toolpaths is that they take allot longer to cut (plus I was having pocket toolpath issues in Carbide Create as indicated above), wearing down the expensive precision endmills (1/64 inch tip and smaller). The compromise here is use contour toolpaths with larger solder sites, so the baord is still more forgiving for mediocre soldering skills, while not overusing the expensive endmills.

• Use a “roughing” toolpath. For my board, I used a 1/32 endmill for a pocket toolpath, then a 1/64 bit for a contour toolpath. This worked well, however there were long sections of certian traces that the 1/32 bit did well, but doing the 1/64 contour was superfluous. My plan for the future though is to refine this method. I don’t need to do pocket toolpaths if my main endmill is 1/32, since this gives me enough clearance for my mediocre soldering. Once the software plots the 1/32 toolpath, it SHOULD be easy to edit the vecotor sections for the trace layout to be able to specify only the areas that the 1/32 pass missed (or that are good enough). Then I just need to contour those sections alone with the 1/64 bit to get my final layout. I just need to be careful not to cut across existing trace lines when doing my finishing pass.

• Hopefully use a solder paste stensilin the future. As I keep saying, I am worried about my solder skills. I know that solder paste stensils are a great solution. After a fair amount of trying, me and the rest of the students here at our lab have not had success getting solder paste to work (there is a chance that the problem is with the paste we have on hand itself). I hope to be able to solve this and use a solder paste stensil for my future projects.

Sending to a Board House

In the real world, you are very unlikely to mass produce your designed circuit boards in a Fab Lab, or even is you are a small company, in-house. Fab Labs are great for prototyping and design, but when it comes to mass production not so much. For many product types you need to create a specialized manufacturing system or assembly line in order to produce your item at scale and with consistent quality. For circuit boards the cost of creating an assembly line are so high that in order to mass produce the only financially viable option is to outsource the board production to PCB board houses.

The process of sending to a Baord House is as follows:

1. Design your board 232pm test

   • The first step in mass production of a circuit board it to design the board itself. You need to pick the components you would need and design your trace layout in software like KiCAD or Fusion360. Usually your first working design is more powerful and complicated than it needs to be. eg:

   • You pick a SoC that is way more powerful than you need and has more input/output ports than you need.

   • You have connectors for programming that are only needed once, or ports you don’t need.

   • You have extra ports and connections dedicated to troubleshooting.

   • Your design is likely too big for the space required for the project as a whole.

2. Finalize your board

   • Once you have your first working design, you need to “trim the fat”. All those extra components, connectors and traces from your initially design that aren’t strictly needed for your production model should be removed.

   • Finalizing your board is an iterative process, where you would produce multiple versions, each trying to hone in on the cheapest reliable design that still works as intended. Common things to change:

     • Downgrading your chips. For troubleshooting convenience we often design our circuits using components that are way more powerful than needed. Downgrading these chips to the lowest power that doesn’t affect final performance means much cheaper production costs.

     • Removing unecessary ports and connectors. This includes the connectors we used to program the onboard chips during design. In your final product you won’t need to reprogram your chips, so it’s easier to program them before soldering them to the board. Fewer ports (both chip pins and as external connectors) means saving money.

     • changing the shape of your board to fit the cavity or space it will be ocupying in our final product.

   • IP Protection. If your board design is novel enough and you think that people stealing your design can significantly hurt your business, you may want to file for a patent, however this is an expensive and lengthy process. Getting copyright protection for your design is much simpler and cheaper but offers allot less protetion against “bootleggers”.

3. Find a Baord House

   • The next step is to find a board house that you can hire. There are quite a few reputable ones out there like:

     • JLCPCB
     • PCBWay
     • Osh Park
     • Aisler

   • Which one to pick is personal choice for your business. You may have factors that limit your choice such as:

     • Not wanting your boards made in certain contries. Perhaps for moral or legislative/tarrif reasons.

     • The reputation of the supplier for reliability and quality.

     • The turnaround to get your product.

     • Tha capabilities of the company especially if you have a particularly complex board.

   • Usually once you find a PCB house that can meet all your needs, the deciding factor is usually the price that they can produce and ship your circuit board for. But again, there can be allot of factors in picking the right PCB house for you.

4. Send your Design to your chosen board house

   • Once you have selected your Board House, you need to send them your design files for them to analyze and certify that they can make it.

   • Most Board Houses can accept multiple design file types, but all should accept Gerber Files.

   • There are design choices that you must stipulate that won’t be included in your gerber files like: board thickness, strength, colour, and substrate material.

5. Finalize your order

   • Once the Board House verifies your design and selected options its time to finalize the order.

   • The board house will likely give you a quotation with differnet options. The cost per unit made will be lower the bigger your order is, so bigger orders work out cheaper than multiple small batches.

   • You will likely get options for priority with rushed orders costing more. As well as different options for shipping.

   • After selecting these final details for your order, there will likely be fine print on the contract that you may want to look over. These are things like quality control, how delays are dealt with, responsibility for customs and duties etc.

   • Once you are happy with all the details, you can agree to the sales terms and arrange payment to the Board House.

---

Last update: June 26, 2024