Symphony

A bioprinter that helps researchers streamline cancer drug development

DETAILS

Jul 2017 - Mar 2018

Small, robust team of 5:
​​CEO (Full-time, onsite)
COO (Part-time, San Diego)
Senior Mechanical Engineer (Part-time, Boise)
Senior Electrical Engineer (Part-time, San Mateo)

MY ROLES

Mechanical Engineer
Project Manager
Coder
Manufacturing Engineer
Quality Engineer
Packaging Designer
​UX Designer

PROJECT GOAL

Design a robust product development process to get our bioprinter to market quickly.

KEY SKILLS

Low-Volume Manufacturing
Electo-Mechanical Assembly
Firmware Development
Product Management
User Research
Usablity Testing
Quality Control
Packaging Design
Shipping & Maintenance
​

---

Developing for Cancer Cell Therapy

When developing immunotherapy drugs for cancer, it typically takes years of product development & research before committing to even more years of animal studies & human clinical trials.

The success rate for drugs & vaccines is 13.8%, with oncology drugs having an even smaller success rate of 3.5%. This is after all the years of research, animal studies, and human clinical trials. 

So how might we change the product development process of developing drugs for cancer?

​With our bioprinter, we can help researchers cut that decision time down to months and potentially save billions of dollars down the pipeline.

My challenge was to create structure for a fast-moving early-stage startup as they bring their first product to market.

​As a mechanical engineer, project manager, and many other hats, I helped build the future of 3D cell culture by bringing to life 6 iterations of bioprinters from concept to early-stage production in just 9 months. 

The Journey So Far

It's the beginning of 2018. We had already gone through 4 iterations of bioprinters in the last 6 months. It was exciting to see our bioprinters evolve from 3D printing our own printers in-house to finally being able to work with external vendors.  

Proof of Concept

V1 
Pre-July 2017 (before I joined)

Units Built: 1​

​Manufacturing Processes:
​3D printing in-house with Formlabs Form 2 printer

Features:

• Proof-of-concept device
• 3D printed enclosure
• Didn't look very pretty 
• Wasn't very user-friendly

Internal 3D Print

After we were able to show that our proof-of-concept worked, we brought on two senior engineers to join the team to make the bioprinter more manufacturable & marketable.

V2​
​Aug 2017
​
Units Built: 5

​Manufacturing Processes: 
3D-printed in-house with Formlabs Form 2 printer​​

Features:

• 3D printed most components
• Post-processed with sanding & spray-painting
• Off-the-shelf electronic components (button, LCD display)
• Built all 5 units in a week

​

V2.2
Oct 2017
​
Units Built: 5
​
Manufacturing Processes: 
3D-printed in-house with Formlabs Form 2 printer​​
​​
Features:

• 3D printed, post- processed with painting & sanding
• Upgraded constant current source for 24 LED Array 
• Arduino module no longer necessary 
  
• No flash button
• Interlock installed but not implemented yet

External 3D Print

V3.1
Nov 2017

​Units Built: ​2​
​
Manufacturing Process: 
External CNC body
Urethane-casted lids

Features:

• Flash button implemented
• Interlock implemented — countdown won't start + LEDs won't turn on if lid is open
• Added front panel to protect unit from any spills
• Changed LCD screen to one that is easier to read

V3.2
​Dec 2017

Units Built: 2
​​
Manufacturing Process: 
External CNC body
Urethane-casted lids
​​
​Features:​

• ​​Potentiometer replaced by Rotary Encoder

The goal for the next and final iteration is to move away from 3D printing and towards a more efficient process to account for scale-up manufacturing, such as injection molding. Because we're an early-stage startup and injection molding is a very expensive process, we decided on the next step up → urethane casting. 

Design Process

With each iteration, I was able to refine the product development process. I broke down each iteration into four main parts: 

Prototype

Develop & implement new technical features: 

• Design of bioprinter
• Firmware
• Upgrade electrical components
• Implement user feedback & safety features
• Cheaper manufacturing techniques

Plan

Design Review: 

• ​Identify key features to implement
• Assess potential risks
• Create Bill of Materials
• ​Do we have the budget?
• Create a detailed project timeline

​Build

Order custom & off-the-shelf components to assemble bioprinters. Typically takes anywhere from 2-6 weeks. 

Test

Quality Control: Ensure that our bioprinters meet all criteria.
​
User Testing: Test our new functional prototypes out in the real world with researchers to see how our bioprinters fit in their workflow. 

Design Process Challenges

With each part of the process came its own set of challenges:

Prototype

​We only had one 3D printer, so it was difficult to prototype quickly since each part took anywhere from 1 hr to 20 hrs to print.

Plan

It was difficult to work with a remote & part-time team. 
​
A lot of documentation and processes had to be created from scratch.

Build

Parts can take a while to arrive. 

Manufacturing delays due to machines breaking down, figuring out how to machine parts, sourcing raw material, etc.

Sending defective parts back to be reworked. 

Test

Building a QC process from scratch. 

Getting users to use our bioprinter. 

Technical Challenges

Within the Prototype and Build processes, we had some technical challenges: 

Uniform Light Intensity

We wanted all LEDs to have the same light intensity to achieve consistent cell cultures in each well. However, using cheaper LEDs yielded a wider range of light intensity values.

We solved this by visually inspecting the intensity of each LED upon receipt and calibrating their intensity values so that they all fall within the same range during QC. As the company grows, we can afford higher quality and more reliable LEDs, as well as have our distributors bin sort them for us prior to shipping to save some valuable time on our end. 

Working with Different Vendors

Because we often worked with different vendors to get different parts of our bioprinters made, we quickly learned during the assembly process that some of the parts did not fit so well together. Different parts called for different tolerances, so we often found that holes didn't align properly.   

This can be solved by choosing the same vendor for future iterations and having tighter tolerances on our parts. 

Hydrophobic Material

We chose one of the components of our bioprinter to be made of Teflon for its hydrophobic properties. However, we found that there were several factors that affected the material's hydrophobic properties during manufacturing: where the raw material is sourced, how they're machined and how it's treated after.

To address this issue, we had conversations with our vendors to narrow down which vendor was able to help us achieve the best results while staying within our budget.  

Also, because Teflon is an expensive material to machine (~$700 per part), we set out to find alternative materials for future iterations.

Hardware Improvements

​Replacing Potentiometer with a Rotary Encoder
The bioprinter has a knob that allows the user to set the exposure time of the LEDs by intervals of 5 seconds. For the knob, we used a potentiometer but soon found its sensitivity was limited by its range of rotation. To address this issue, we explored several options: creating larger increments in the code, having a second potentiometer for set minutes, using a dial with a larger range, and using a pushbutton. We eventually decided on a rotary encoder because it can spin in either direction indefinitely with no stops and it has a more tactile feel. 

Implementing a Flash Button
When we tested our bioprinters with researchers, we learned that they would like a more visual cue that the bioprinter is in use other than the displayed countdown on the LCD. To solve this, we chose a button that flashes during countdown to indicate that the bioprinter is in use.

Adding an Interlock
Another takeaway we found during user testing is that researchers often forget to close the lid prior to starting the countdown. To solve this, we added an interlock so that the user gets a message if a countdown is started with the lid open.

Refining the Design Process

We wanted to launch our bioprinter in the next 3 months, so we took our lessons learned from building our last 5 iterations to refine the product development process and get to market quickly.

​Prototype

​Prototype quickly and more efficiently by outsourcing to local external vendors where we can.

Plan

Have frequent standup meetings and collaborate with all team members as much as possible.

Find a PM tool that everyone is willing to use and can learn quickly to streamline communication and get tasks done faster. 

• Email → Slack
• Trello → Asana → Google Sheets

Build

Add a 2-4 week buffer in our build schedule to account for any delays due to manufacturing, part defects, reworks, etc. 

​Update your clients about your progress as soon as possible. 

Test

​Learn as much as you can from your team members and your users. Don't be afraid to ask all the questions!

Creating fully functional prototypes for each iteration allowed us to fail fast with every build, quality control check and user test. By the 6th iteration, we were able to streamline product development and unlock new parts of the process to have a fully fleshed-out product for our first big pharmaceutical client.

🚀 Prepare for Launch!

We were finally at a stage where we're ready to launch. With this came unlocking a new part of the product development process. 

Prototype → Plan → Build → Test ​→ ​Launch!

By the time we got to the final iteration, we were able to get down to a process we were happy with so we can run our next iteration more smoothly and save lots of time and money. We knew we were ready to deliver to market when we have addressed most of our users' needs and have gotten our functional prototypes to a point where our hardware and firmware are reliable. 

1. Prototype: ME layout (create the box) > EE layout (design things for in the box) > Design Review
2. Plan: Translate BOM into cost estimates. Get time estimates for building, assembly & shipping.
3. Build: Order parts. Receive parts. 🔨 Assemble!
4. Test: QC
5. Launch: Pack & Ship!​

Packaging

We created a rough layout of all the parts that needed to be shipped (bioprinter, accessories, instructions, power supply) to determine what boxes and foam sheets we needed to ship our first batch of bioprinters to our clients safely.  

Another big challenge was trying to figure out how to store all of our units, packaging and shipping materials in such a tiny lab space!

User Instructions

We worked with an external graphic designer to produce a very professional-looking set of user instructions to go along with our packaging. 

Having produced the first few iterations myself in PowerPoint, it was very exciting to see this part of the process come to life. 

Shipping

To keep track of our inventory,  we created a numbering system for our bioprinters and labeled them accordingly on their respective boxes. 

Because we had two versions of our bioprinters, we had to come up with a numbering system that is consistent across our inventory and supports future parts. This includes smaller parts of our units to have their own part numbering system as well (mechanical, electrical, labels, and accessories). 

Documentation

In addition to creating an inventory system from scratch, I helped create a lot of documentation for every part of the process, from functional design specs and very detailed SOPs for assembly, to detailed quality control processes and checklists for demos and shipping. I like to think of them as very detailed packing lists.

Results

After a long weekend, and a handful of late nights and team dinners, we were able to complete building our first batch of bioprinters and get our bioprinter to our very first pharmaceutical client on time. This was the first product that I've personally ever shipped, so it was a very exciting time for me to be with the company. Over the last 9 months, I watched our bioprinters transform from these slightly fragile units that were 3D-printed, sanded and sprayed in-house to these beautifully anodised bioprinters with glistening urethane-casted lids. 

6

Iterations in 9 months

28

Units built in-house

25%

Reduction in manufacturing costs + reduced assembly time from 1 day to several hours

19

Potential clients in academic & pharma

100+

Cups of ☕

Lessons Learned

Always maintain transparent communication with all of your stakeholders. 

• This includes all end-users, potential clients and manufacturing vendors
• Give manufacturing vendors an earlier lead time than expected
  • Account for some delays in machining & delivery of parts
• Be as communicative as possible with your clients, especially when they have a pre-order and there are delays with your build

When prototyping & testing, always 3D print backups! 

• There's always a chance a piece will warp over time or break

Design for Assembly: Reduce where possible!

• The total number of parts required to build
• Minimize separate fasteners

​
Take packaging and shipping into account when creating your timeline.