Inside the Build: 18 Months to FDA
Featuring: Kera West, Engineering Solutions Director, Nate Bolyard, Principal Electrical Engineer, and Sean Sawdon, Principal Systems Engineer
“There’s a human on the other end of that device for the first time, and you have to get it right.”
Welcome to Inside the Build, a podcast series that dives deep into how Forj Medical solves complex manufacturing challenges, told by the people whose knowledge, expertise, and ingenuity make Forj Medical the partner medtech companies trust when the technology is complex and the stakes are high.
Where the project started
The customer had a therapy and a market. What they did not have was a team that could turn a complex, life-sustaining therapy into a portable device. They were good at the pump, the disposables, the procedure, and building the market. The electrical, mechanical, and software work of packaging all of it into something a clinician could carry into a room and set up quickly was the gap.
The devices already on the market were part of the problem they were trying to solve. They were complex enough that hospitals were assigning a nurse to monitor a single patient full time, which does not scale as a business model for the therapy.
Simplification was a core requirement, not a nice-to-have. Fifteen separate parameters and a wall of notifications became a single clear indication of whether things were good to go. Setup, meaning how the device connects to the patient and gets ready for use, got the same treatment.
The program ran in two parts. An early alpha effort defined the architecture and the broad direction, then paused while the customer worked through their side. When it restarted, the team ramped quickly and did not slow down. Concept to FDA submission was 18 months.
Building at the wrong size on purpose
In the first weeks the team worked off off-the-shelf dev kits to prove basic feasibility. Then it did something that looks backwards. It built a complete, functioning version of the design in a deliberately oversized format. Circuit boards were ordered at roughly double their final dimensions carrying more or less the same circuits. Sheet metal frames were oversized to match.
Two things came out of that. Software engineers were writing against real, functioning circuits months earlier than they otherwise would have been. And when a surprise showed up, which it always does no matter how many hours go into schematic review, reworking an oversized board was straightforward. Whole sections of circuits were deleted and reinvented on those boards without waiting for a spin.
The board count moved around during development. It settled at roughly three large, complex boards plus three or four smaller ones, eight total inside the enclosure. The multiple processors that drove some of that board count also served the architecture, keeping critical code segregated from less critical code.
Buying parts before drawing the schematic
The program landed in the middle of the post-COVID component shortage. Once it was clear the shortages were going to get worse, the team canvassed everything the design would need and bought ahead. Alpha, beta, and DVT quantities were ordered at risk, well before anyone knew which parts would survive to the final design, which meant building a small inventory management practice inside a development program.
Electrical engineers worked with multiple displays open, sourcing a part first and then placing it in the schematic, so nothing got designed in that could not be bought. The customer’s willingness to fund long-lead parts at risk is what made that possible, along with the willingness to ramp a large team fast and make decisions quickly.
The sealed box problem
The technical requirements fought each other. The device had to be portable, so the box was small. It had to hold an IPX rating, so it could not breathe. It had to survive helicopter-grade vibration, which is close to the worst case that exists. And eight circuit boards inside it generated heat that had nowhere to go.
The team attacked the thermal problem before the first prototype existed. Crude concepts came first, just to check assumptions, followed by full fluid and thermal simulation against increasingly detailed mechanical models. What came out was a thermal transfer plate approach that maintains the seal, holds good component temperatures in the worst-case operating condition, and is still straightforward to assemble.
Simulation is not the right call on every program. Here it removed roughly a full design iteration, and the program ran two iterations instead of the three or four it might otherwise have taken. On a device at this complexity, an iteration is months.
Designing for reliability instead of adding it
Function is the easy part. A device that is safe and reliable at this level cannot be assembled from a functional device late in the program.
That shaped the architecture from the beginning. Redundant power sources and a power architecture that can draw from two supplies without a single point of failure. Multiple processors with critical code kept away from less critical code. IEC 62304 for the software. This was also one of the first Forj programs to go through the FDA’s cybersecurity guidance, which arrived partway through a device effort that had been running for years.
The front end of the project reflects that. Requirements clarification, risk analysis, and mitigation planning happen before anyone touches a keyboard. Projects succeed or fail on requirements, and rushing into implementation to save weeks at the start tends to cost months at the end.
Moving it into production
Manufacturability was part of the architecture conversation, not a later review. How circuits get divided across boards, how boards mount, how cabling routes, and how the whole thing splits into subassemblies all got decided with the build in mind.
Manufacturing came in early, aligned the build cell to those three main sections of the design, and set up subassembly testing so problems surface at the section level rather than at final test.
Production also ramped ahead of FDA clearance so units would be on the shelf the day the customer had approval. The first 25 devices took rework loops, engineering support, and red lines to the procedures. Today the line runs a Kaizen cell producing roughly 100 a month with low rework and good yield, while the design team continues to work through customer and agency feedback in parallel.
The handoff that wasn’t a disruption
The customer was a startup that always expected to be acquired, and the likely timing was right around submission. That is exactly what happened. A major medical device company came in at the point the submission went to FDA.
Because the acquirer had been close to the startup throughout, the transition was largely seamless. New stakeholders with different expectations arrived at the moment the program was shifting from development speed to production discipline, and the manufacturing work stayed at Forj. European approval came through on schedule. The device is marketed and sold globally.
What made the pace possible
The pattern is consistent across Forj programs. Go hardware-rich. Build a lot of prototypes and build them fast through quick-turn board suppliers. Assume a couple of boards get destroyed on purpose, because finding a weakness at the bench is worth far more than finding it in the field. A spin at the beginning of a program costs a fraction of a spin at the end.
The other factor is harder to schedule. Eight electrical engineers, roughly three mechanical engineers, and as many as 17 software engineers at peak, most of whom had solved versions of these problems before. Nobody was starting from a blank page on motor control, on single-fault tolerance, or on what the regulatory path would demand of the architecture.
The day the first press release went out, the team saw a photo of the first patient treated with the device. Everything got real at once.
This has been Inside the Build, a podcast series by Forj Medical, where we’re shaping the future of life-saving devices, one build at a time. Thank you for listening. For more information, visit us at forjmedical.com.