The Evolution of the Novatropes Circuit Board

May 10, 2021

Working on circuits with breadboards is fun and, in some cases, as easy as following a tutorial online. However, going from a breadboard to a production board as a startup is a different experience. Tutorial videos on this topic are great, but even the best tips do not account for your own mistakes. In this blog post I’ll share my journey in bringing our Strobe Dock circuit board from prototype to product-ready status in an effort to help shine a light on some of the real-life nitty gritty details along the way.

Quick glossary:
SMD - surface mount device. SMDs are circuit components that can be soldered on the surface of a circuit board.
PCB - Printed circuit boards. It's like a chassis that contains all the connections necessary between electronic components in a circuit board design. This avoids using cables to connect electronic components.
ESD - Electro Static Discharge. Electricity that can accumulate in the human body for example when we rub our feet in a carpet. This electricity can damage electrical components when we touch them. 

To give an idea of where we started, I’ve included a picture of our first breadboard solution that kickstarted our campaign. At the time I had very little to no experience building proper circuit boards, soldering circuits, or even programming them. The breadboard prototype version was fantastic, mainly because it gave us a minor victory and concept validation; it proved our idea was possible. It still lacked some of the functionality needed to be product-ready, but it had the most critical part: a working strobe light that animated a spinning sculpture.

Once we had concept validation on a breadboard, I stepped it up a notch and soldered the circuit onto a protoboard, which is an empty perforated circuit board with holes spaced at 2.54 millimeters. This meant soldering all the components by hand in a complicated maze, which quickly became packed and messy. I wish I could go back in time and tell my past self that there was an easier way, but at the time it made sense (more on this later). The most annoying thing about this approach was the gross amount of time it took to solder all the components and the time it took to fix shoddy solder joints. In retrospect the repetition of this overcomplicated soldering process provided me with an opportunity to gain a better understanding about each of the components in the board, but in the long run it probably wasn’t worth it. On the other hand, we were coming up on a self-appointed deadline of shipping out prototypes by Christmas and didn't have the time to deal with the learning curve of electronic design automation (EDA) software like Eagle, or the long turn-around-times from a supplier. Once we had a functioning board, we knew the design needed to improve but learning circuit board manufacturing and supply chain management in less than a month is not realistic.

Once christmas prototypes were out, the clear next step was to level up to professionally assembled PCBs, which would reduce the time soldering and fixing mistakes my orders of magnitude. So like any kid born in the 90s, I went down a google rabbit hole to get a better idea of how to take our operation to the next level. I was able to find a number of tutorials on building a PCB in design software, manufacturing it overseas, and assembling it locally. I remember going through the trouble of testing out multiple circuit design platforms until I finally settled on Eagle, which I still use today. It has its learning curve, but it was way more user-friendly compared to other platforms. As for the manufacturing company, we eventually landed on JLCPCB, which offers unbeatable low prices and fast turnaround times. I still use them to this day, and they have been excellent for prototyping new boards. On average, it takes about a week and a half to receive a circuit board in the mail after hitting the "submit" button.

Our first PCB design looked like this. It was much cleaner than the previous iteration, and it had way more functionality. However, there were still many unnecessary components, and my understanding of Electrostatic Discharge (ESD) was not as solid back then. ESD ended up being responsible causing many issues with the ATMEGA, which was the brain of the whole board. Nevertheless, the resulting circuit board was a leap forward in quality. We reduced our soldering time from 2-4 hours (including corrections) to 1 to 1.5 hours. It also reduced the amount of improperly placed components.

This circuit board iteration was designed for the pagoda model, which had its main LED chip mounted on over the sculptures rather than underneath the sculptures.

Another particular quirk of this board is that it used more boost and buck converters than was needed. We had an initial desire to have a small screen show the customer the driving strobe frequency, which required a voltage step-down to 3.3 Volts. We later removed the screen after learning that it was not a value-adding feature, and caused more trouble than it was worth.

What people enjoyed about our builds was mainly the controlled effect at the primary observable frequency (40Hz) and having the ability to switch colors and sculptures.

Unfortunately, the Pagoda model was overly complex and costly to assemble, and didn't get a very positive customer response. We decided we needed to simplify the form factor, and rethink the product from the ground up.

To make that happen, we had to redesign the whole structure, as previously mentioned in a previous blog (link below). The LEDs were integrated into the board directly rather than being an external component. The long-term vision was to make the manufacturing more centralized and streamlined when ordering a high volume of boards.

The Story So Far 3/3

The redesign pushed us to set SMD LEDs on one side of the board and the rest of the components on the other side. It also meant that we needed a quick way to prototype with SMDs without the budget to buy professional circuit board ovens A.K.A. reflow ovens. Reflow ovens are allow you to use surface-mount devices (SMDs) rather than being confined to using through-hole devices (THDs). Basically, an empty circuit board is ordered with small conductive pads, and the SMDs are placed with the help of a litter solder paste, and placed into the reflow oven with runs through a temperature profiles that binds the SMD components to the board without burning them. Being a cash-strapped startup, we were able to find a kit which let us build our own reflow oven from a small toaster oven rather than purchase an expensive reflow oven online. The kit we used was called a Tiny Reflow V2 which is available for purchase at the link below: Rocket Scream
While it wasn't the prettiest thing ever, this bad boy ended up cranking out around 250 circuit boards for us and is still alive and well today. In fact, this oven still helps us to reflow our prototype boards, which saves us money by not needing to order fully assembled boards from JLC.

For the first 250 boards, we were still soldering on all the components ourselves. Even with the increased efficiency afforded by the reflow oven, this was a big time burner. Between having to place the LEDs by hand, baking the parts in place, and soldering the remaining components by hand, it still took around two hours to manufacture a single board.

In order to decrease time spent per board, the only solution was to once again level up our production process, and learn how to get JLC to actually assemble as much of the boards as possible before sending them to us. This meant I needed to redesign the board with as many of the components as possible being SMDs (JLC uses robotic pick-and-place machines for SMD placement, which decreases cost per board). Since this is a rabbit hole of its own, I will write another post on this later. Getting JLC to assemble our boards had a learning curve, but one I did, we had successfully outsourced the single most time-consuming step in our production line. While most of our board is now SMD, there are still a few THD components we solder in-house. That being said, we are very close to our goal of a completely SMD board which requires no in-house soldering, and will most likely have that ready in time for the next batch of boards.

The world of electronics can be extremely intimidating, but as you can see from this blog post, it is a constant iterative process that doesn't require perfection on the first try. Small, gradual improvements compound over time to eventually result in a product that meets all of your requirements. While our circuit board still hasn't met all of our criteria, we will continue gradual improvements until our circuit boards are the modern-day feat of techno-wizardry we know they can be.