The Beating Heart of the Bullet R1

11 min readDec 6, 2018

An Improved Induction Board Means A Better Coffee Roaster

The original proof of concept for the Bullet R1’s induction board

Nearly a half-decade ago, Jonas Lillie built a proof-of-concept for an induction-powered coffee roaster by ripping the guts out of a induction stove and shoving it in a box. It was an encouraging start — and the beginning of an incredible amount of work.

It takes considerable effort to produce a working prototype from a simple idea. But it is far more difficult to take that prototype and then develop a finished product ready for market. Startups around the world will sympathize when we say that the problems never end. The only way forward, however, is to push through them.

One of the very first challenges we faced when moving from prototype to market was the development of the power electronics for the induction board. This board is the beating heart of the Bullet R1, functioning both as a low voltage power source for the fan, drum motor, and control panel, as well as a source of high voltage, high frequency switching for the induction coil. The board is what makes the roaster uniquely powerful, and it is the single most complex piece of gear in the machine. Finding a developer to cooperate with us on its design would prove a difficult task.

A Search And Create Mission.

Once we knew that we could roast coffee in an induction heated drum, we began our search for an induction board developer. Originally the plan was to work with a local firm — one here in Taiwan — to produce it for us. While we were able to successfully source and develop all of the other parts for the machine from a variety of local companies, our search for a board developer was turning up nothing. After a year of looking, we realized we had reached a dead end. In fact, although we’re surrounded by high-powered electronics and depend on them in our daily lives, companies able to build them specifically for induction are few and far between.

We had to search elsewhere. Thankfully, we were eventually able to find what we were looking for in China — a high-reputation manufacturer with many years of experience building electronics specifically for induction-based products. We worked together with them, and also the Technical University of Denmark, to successfully design a powerful induction module that could easily handle 1500W of heating power, enough for a full kilogram of coffee at a time. Truth be told, if not for the space restrictions of the drum and heatsink, it could handle even more. The board really packs a punch.

We were quite pleased with the results of our collaboration, and we would still be pleased if it weren’t for one little problem: It was nearly impossible to get the thing certified.

Certifiably Insane.

If you’re not able to purchase a Bullet R1 in your country, then you probably already know why. It is difficult for some countries to import goods without certification, and version 1.0 of the Bullet R1 is not certified, though certainly not for lack of trying on our part. This has been a source of great frustration in some parts of the world, particularly in Australia.

We knew that it was never going to be easy to get certification. The typical appliance in need of certification — a microwave, for instance — will already have certification tests designed specifically for it available, and it’s clear from the beginning what it will need to do to pass. But there is nothing else on the market like the Bullet R1, and so it took a lot of time to create the right tests for it. In other words, the project was an ambitious one for both ourselves and for our certifiers.

Still, despite some delays, the company assisting us with certification did eventually finish crafting the tests, and we passed every single one that was thrown at us.

You may be surprised to hear that. We've received mails from some customers who assume that the Bullet was unable to comply with some particularly strict safety or mechanical rules. But the real reason version 1.0 did not get certified is actually quite dull: In the end, much to our consternation, it was a paperwork problem.

Each of the dozens of components on the induction board — many of them custom-made — needs its own certification paperwork. Unfortunately, we were unable to obtain those documents from our manufacturer in China, despite them being under contractual obligation to provide them to us. This was where our certification inexperience certainly hurt us — we should have demanded the paperwork from the manufacturer up front. Unfortunately, we were already deep into the certification process before we discovered our need for it. Certainly, it would have been helpful if the company assisting us with our certification had filled us in on this requirement in advance. But we can’t blame them, either. It’s on us.

Undaunted, we began researching the component suppliers from which our manufacturer had sourced its parts. With a bit of detective work in China (not easy) we began tracking down the missing papers, with some success. In other words, we were giving our own manufacturer the certification papers they were supposed to be providing to us. Needless to say, this was all very upside-down and not at all what we had hoped it would be. And, worst of all, there were still some components for which we could not get the paperwork, even after tracking down the companies who manufactured them.

And so a year after shipping out the first pre-order Bullets, we found ourselves in a position we never expected to be: Unable to enter some markets because we didn't have the right papers.


In a twist of fate, as the reality of our paperwork conundrum was just beginning to sink in, STMicroelectronics — the company who manufactures the Bullet’s microcontroller — had developed a new reference design we believed could kickstart our own, brand new induction board design.

We had already been considering this path for some time. The old boards are great and can do almost everything we want. Almost. But we did see some opportunities to improve upon the design, especially after working with them for so long. And now suddenly we had a new, urgent reason for moving the board design in-house: Certification.

We emailed STMicroelectronics’ Italian office, and our requests for info were funneled through a chain of contacts that took us from India to China and eventually all the way back to here in Taiwan, where Giovanni, a leading expert in power electronics, was working for ST.

Giovanni reached out to us, and kindly demonstrated a 3KW induction heater that ST had made. Jacob met with him one-on-one to discuss details, they got a little geeky together, and soon we had the foundation to start developing our own board.

We made great progress during those early stages of development. It suddenly seemed less risky, and made much more sense, to pour our resources into something that would actually make the Bullet better, and that would ultimately offer us more control over our own product. It was the right time to build a better board.

A Better Board.

The New Board

Once the baton was passed, Jacob ran with it. It was a tough project, one that probably deserved a team of experienced engineers, but Jacob was working alone. He wasn't exactly starting from scratch, however. He has a Masters in Electronics, and his thesis project was to develop a high power class D amplifier that would qualify for George Lucas THX Ultra 2. Although that sounds like a non-sequitur, it isn't. An induction module is a lot like a high voltage class D audio amplifier, and Jacob has joked it would be fun to attach a speaker to our module, modify it a bit, and set it up to play music.

Jacob was able to produce a working version quickly. But he also discovered that it would indeed be possible to improve upon the design. As the project became more ambitious, those planned improvements ate up more of his time. The development of the current measurement circuitry to determine the power of the induction was the most challenging part of the design process. This has to be extremely stable at all temperatures, low noise for precise measurements, and also very fast to detect any short circuit in the coil wire.

Other improvements include more stable power output over a larger temperature range. Generally more efficient heating. Simplified wiring, with a board to board cable of only 5 wires compared to previous 14. Better protection for the induction circuit. A more stable power supply. The ability to upgrade firmware. Long life capacitors, a better PCB (now in black!) and a better EMI / EMC filter. And of course, the very highest quality components we could source ourselves.

Sexy in black?

For users of the Bullet R1, the real-world benefits of the new board are numerous. For example, the board will automatically adapt to different line frequencies and voltage fluctuations, so that each module will output the same power at each step, no matter where you are, or what your setup is. Considering how sensitive coffee is to slight variances in heat while roasting, these extremely accurate power outputs will make it easier to repeat a roast profile that you’ve dialed in.

These more precise power measurements and power outputs of the new induction board will pair particularly well with the upcoming Infrared Bean Sensor upgrade, which creates more accurate and meaningful bean temperature data across a wide range of batch sizes. (More on that in a future article!) Through the exchange of truly repeatable roast profiles and associated cupping notes, we envision Roast.World and the entire Bullet R1 ecosystem eventually functioning as a world class coffee lab, a space where we can all learn more about coffee roasting together.

There is more to be excited about. The component upgrades and increased efficiency also allow us to run the induction module with higher power while maintaining cooler internal temperatures, which means that the Bullet R1v2 is going to be a little more powerful than its predecessor, too. Having roasted on a prototype myself, I can tell you that this boost is indeed noticeable, offering more headroom on larger batch sizes.

And in the event that something goes wrong, the new board opens up different avenues of technical support as well. The board’s microcontroller is capable of logging its own data, which can then be sent back to our servers to let us know how each module is running. We've also customized its error codes, making it possible to quickly diagnose problems with much greater precision. In some cases, we will be able to tell you exactly what’s wrong with your roaster before you even discover there is anything wrong with it.

All of this is to say that the new board is a significant upgrade and we’re really happy about how far we've come.

What’s Next? More Tests. And Then A Celebration.

Glamor shot.

Jacob began this project nearly a year ago, and barely a day has passed without him working on it. Lucky for him, we purchased high-tech testing equipment to help him out. The coolest piece is probably the high-resolution infrared camera — the one we like to play with on Instagram — which allows us to check the temperature of each component on the board individually while the system undergoes various stress tests.

So far we have tested the power output up to 2800W, which is twice the power the version 1.0 will use at power 9 (though in the new Bullet this will be limited to about 1600W), and we have stressed tested the line voltage to over 265V and under 200V. In the process, we have done a few very weird roasts with a very low preheat temperature followed by a massive amount of power during the roast.

Did those roasts taste good? Um…

Another cool piece of testing equipment is our HI-POT Ground Bonding and Insulation Resistance Tester which tests safety insulation up to 5000 volts. We used this for pre-compliance testing, so we can be sure that when it goes to the certification laboratory for testing it will pass.

All of this equipment has already paid for itself, as we've been able to diagnose a few issues and resolve them without wasting time. We are at the final design iteration now, and have produced 15 sample boards. Jacob came to Taipei in November to put these boards through their final in-house tests, which included such potentially scary things as deliberately shorting the various outputs to ensure everything shuts down in a safe manner.

Five of the 15 sample boards have been pegged for use in the upcoming certification tests. (The remaining 10 are being distributed to a few trusted users for real world use.) We have hired a certification expert — someone who actually trains certifiers how to certify machinery — to help us keep things moving forward as quickly as possible.

Paperwork-wise, there will be no more problems. We have made sure of that. All critical electronic components are now manufactured in either Japan, Taiwan, the US, or the EU. They can all be purchased locally here in Taiwan, and we have all of their certification papers.

We are told that once we line up all our ducks in a row, the certification process should only take about 8–12 weeks from start to finish. We hope to begin very soon, after we receive feedback from our test users in the coming weeks.

It’s important to keep in mind that there are some upcoming certification tests that we simply cannot duplicate in our office. For example, the Bullet will still need to go through the abnormal run tests, in which they run the machine at full power under extreme conditions (100% humidity, and very high ambient temperatures, for example) to ensure that it is safe at all times. And then there are the EMI and EMC tests, to ensure that the Bullet neither creates excess electromagnetic interference, nor is vulnerable to it. Obviously, if the Bullet is unable to pass one of these tests for some reason, our time line will be altered. Still, we are feeling more confident now about certification than we ever have before.

For those in markets such as the United States, we have already begun developing a 110V version of the new induction board, and its release will follow later in 2019. Don’t worry about getting left behind, though. Anyone with the old version of the induction board will be able to purchase upgrade kits at a cost approximately the same as the cost differential between a version 1.0 and version 2.0 machine. The upgrade is certainly not necessary for roasting great coffee, and some casual users may decide they don’t really need it. But we strongly encourage serious roasters to get their hands on a new induction board when it is available.

As for all of you in markets requiring certification, we’re very sorry for the long wait. Hopefully, you now better understand some of the unique challenges we've faced, and also just how much additional work we've poured into building an even better machine for you.

We are so excited to finally get it in your hands…