Manufacturing an Electronic Component

From Start to Finish

Indie Hackers - October 13, 2020

Michael Lynch (@deliberatecoder)

https://decks.mtlynch.io/indie-hackers-2020-10/

Background

  • Raspberry Pi's annoying power design.
Photo of USB-C port on Raspberry Pi 4B
  • The Pi accepts power on the USB-C port
    • This is the device's only USB-OTG port.

Background

  • TinyPilot needs the USB OTG port to connect to a computer's USB port.
    • Computer USB ports output too little power for the Pi.

Background: Y-cables?

  • A Y-cable seemed like the answer!
Photo of Y-Cable

Background: Y not?

Screenshot of Pi Forums post: I notice you're providing a USB 'Y' cable in your kits, how are you dealing with PSU conflicts and backpowering the host PC?

Goal

  • Split access to the Raspberry Pi's USB-C port
    • Should accept 3 Amps of power.
    • Should also function as a USB port.
    • Prevent incorrect flows of power.
  • Need it fast.

Day 1

The engineering firm begins work on the circuit board for the power connector.

It's a simple enough board that they're able to design it and order 100 printed circuit boards from China the same day.

Day 2

  • I ask 3D printing lab to design a case for the power board.
  • Within hours, they send me a work-in-progress image of the case design.
CAD image of a partially completed case design

Day 5

The 3D printing lab completes their design and gets ready to begin printing a few prototype cases.

3D rendering of case, bottom view, open 3D rendering of case, top view, open
3D rendering of case, bottom view 3D rendering of case, top view

Day 8

The engineering firm receives the bare PCBs from their overseas manufacturer.

Photo of a panel of uncut, unassembled PCBs
  • Still need to solder on the components.
  • 3D printing lab produces first two case prototypes.
    • I ship them to the electrical engineers.

Day 9

  • Cases fit the boards

Photo of a panel of uncut, unassembled PCBs Photo of a panel of uncut, unassembled PCBs
  • Large gaps around the port because 3D printing lab erred on the side of caution.

Day 10

  • I receive the first two board prototypes.
    • They're soldered by hand as the engineers build automation
  • They work!

Day 13

  • 3D printing shop prints their first batch of 30 cases.
  • Still a small gap around the microUSB ports, but not a showstopper.
Photo of 24 completed boards

Day 19

I receive the first completed panel of 24 PCBs from the engineering firm.

Photo of 24 completed boards
  • Manufacturing process was automation + manual fixes.

Day 20

  • The 3D printer finishes the remaining 70 cases.
  • They include 10 experimental cases.
Photo of black case for power connector

I like this new design so much that I switch all future production to black cases.

Day 21

I begin sending out the first completed power connectors as replacements to legacy customers.

Screenshot of replacement orders in Shopify

Day 26

  • I receive the remaining 74 completed boards
  • With 100 cases and boards ready, the first run of production is complete.

Costs

  • Boards: $2,897.70
    • Design: $241.72
    • Materials: $422.16
    • Assembly, testing, packaging: $2,579.04
    • Postage: $76.95
  • Cases: $500.00
  • Total: $3,297.64
Per-unit cost: $32.97

Case costs are after 75% subsidy from MA Innovation Voucher.

What went well

  • Luck
    • Almost everything worked well.
  • Limited the number of vendors involved
    • Prevented miscommunications, friction.
  • Added padding to the time estimates I received

What went well

  • Communicated to everyone that I was optimizing for turnaround time
    • I estimated that I lost $50-100 for each day of delay, so I was willing to pay ~$50/day for options to expedite.
  • Working in parallel with the 3D printer and electrical engineers

What could have been improved

  • Underestimated total number needed
    • After sending out free replacements to ~60 legacy customers, I had only 40 left to sell to new customers.
  • PCB assembly time
    • Biggest bottleneck, most costly part.
    • Possible resolution: after prototyping, outsource assembly to two redundant PCB manufacturing firms.

What could have been improved

  • Functional testing
    • Some units were defective when they arrived to customers.
  • Deadline slippage
    • Three week delays happen one day at a time.

Questions?

Bonus: How'd I find vendors?

  • Luck
    • For both vendors, I was already in contact with them to discuss other physical improvements to TinyPilot.
  • Electrical engineering firm
    • Asked PCB manufacturers for recommendations.
    • FORGE Mass: Nonprfit that connects MA startups to manufacturers.
    • Asked friends for recommendations.
  • 3D Printer
    • Recommendation from a friend.