When to leave the breadboard behind

"I swear, it worked 30 seconds ago, I just need to wiggle these wires until I find the loose one... 15 minutes later... See it's working!"

I've had this happen to me multiple times, it's very frustrating, in large part because starting on a breadboard or protoboard is the RIGHT choice when handling unknowns in a quick moving company. You need to figure out what you can and cannot do with the sensors or actuators you want to use, see if your goal is possible and frankly, iterate as quickly as possible.

The real stakes

There are two mistakes here, moving to a custom PCB too early, and moving to a custom PCB too late.

Too early and you can prove some of the system works, but don't actually know enough about your schematic and requirements to properly layout the PCB. This will slow down your iteration cycle, since you will have to make big changes to the custom PCB and order, at least, one more iteration of the PCB than you should. Since laying out a PCB and board bring-up is measured in weeks, not hours, one or two additional iterations of custom PCBs can dramatically push back your timelines, especially if you have to change your MCU selection. Changing to a custom PCB too late will result in very different but also significant problems. The use of a breadboard or protoboard has the advantage of speed, at the cost of reliability, if you are still using a breadboard when trying to demo to investors or customers that aren't yet bought in you risk losing their interest through a failed demo. In extreme cases with non-technical viewers, it will lose you necessary investment or revenue. As this is a time-consuming process it is easy to mistime. Experienced founders will often do this transition too early trying to get ahead of their dwindling timelines and runway. First-time founders often under-estimate the time it will take to go from a prototype to a custom PCB and over-estimate how much information they need for this transition, a fear of making the wrong decision.

The signals

There are a number of signals that indicate when it's time to make the move to a custom PCB, and I've split them between the technical forced transitions and readiness signals.

Forced transitions

1. The needed slew rate (also called edge rate) for operation of your prototype are impacted by breadboard parasitic L/C. (Dangerous Prototypes, Hackaday, Sierra Circuits)
2. Your rail currents are beyond those of prototype boards or breadboards. (Circuit Specialists, DigiKey)
3. Your prototype needs to be moved around enough that your biggest concern becomes transportation of flaky connections.

Readiness signals

1. There are no longer significant changes needed to the schematic (not the layout).
2. You already have a few datasets showing the basics of the functionality you need for your product.
3. You are taking the prototype into any type of 'in-field testing' with real-world conditions.
4. You are starting to talk about secondary requirements beyond 'does it work?' and 'can our customers afford it?' to 'where can we pull cost out?' and 'how do we improve the reliability?'.

What the move actually involves

With a custom PCB the key goal is solidifying your product, taking it from conception to mass-production.

1. Design
   1. Make sure you have few or zero major uncertainties remaining in the product, and have a beginning list of requirements for your product (a PRD).
   2. Plan your initial test path. This is often skipped, but you should at this stage know how your engineers will access deeper information about your product.
   3. Create the schematic for the product. You should also be able to get a rough cost estimate from this point.
2. Development
   1. FW should start putting together your lowest level of firmware the HAL (Hardware Abstraction Layer), the BIST (Built-In Self-Test) framework and the initial testing software.
   2. Design the PCB layout, this should involve discussions between your EE team and your FW team. Send the mechanical team behind your product a CAD file of the PCB.
3. Order your PCBs. This will take 1-2 weeks for your first batch. (JLCPCB, NextPCB, Sierra Circuits)
4. Bring-Up and Go/No Go decision
   1. PCB bring-up, this can take up to a month of time including debugging, but you need to make sure this is done right.
   2. Review the PCB status and decide if your current testing is viable on V1 of the PCB, this is based on if your PCB needs any mods or if any components are acting in an unacceptable manner. Complete any testing you can do and need on V1.
   3. You will either have minor changes or major changes needed to the PCB. If minor then you are good to proceed, if major then you will need to go back through many of the above steps (albeit much more quickly) for your PCB V2.

Do it so V2 isn't a rewrite

You are looking to create the foundation, not the final product, with this transition. This will be the PCB that you can send out to remote team members, testers and install in your prototypes that potential customers will use without supervision. By making this transition at the right time you will have the first clear vision of what the product is going to look and feel like to use, allowing you to select the key traits that you want to maximise going forward. A few key pieces to have in-place in your V1 so the V2 transition is small:

1. Test points on your PCB.
2. BIST plan ready.
3. Proper documentation.
4. Modular blocks.

If you need help with this transition, it's exactly where our expertise lies. Feel free to reach out.