You Need It, They Build It

David Profitt was a little confused when a developmental biologist walked into his engineering shop and asked if Profitt could make an automated embryo sorter using a Fax machine.

Jeff Minerd(jminerd@the-scientist.com)
Jul 3, 2005
<p></p>

Photography: Jennifer Saul

David Profitt was a little confused when a developmental biologist walked into his engineering shop and asked if Profitt could make an automated embryo sorter using a Fax machine. "I thought, how can you sort embryos with a piece of telephone equipment?" he says.

As it turns out, the researcher meant FACS (fluorescence-activated cell sorter), not Fax. Profitt, an electrical engineer and head of the shop that serves the Stanford University School of Medicine, was no stranger to medical technology, but cell sorters were new for him. Nevertheless, he agreed to take on the project.

For Profitt, getting his homonyms straight would turn out to be the least of his challenges, and what was supposed to be a three-month project would become years of work. Yet such is life in machine and engineering shops around the world, where people with different backgrounds, vocabularies, skills, and working styles...

THE PRICE FOR FIXING

While buying off-the-shelf is almost always quicker and cheaper than having something custom-built, it oftentimes just isn't possible. Researchers turn to machine shops when they need a device, part, or component, and nothing exists to do the job. Plus, most researchers lack the experience in machining, electronics, programming, welding, and so on – not to mention the time and equipment – to do the job themselves.

So, they turn to machine shops equipped with saws, sanders, lathes, drill presses, milling machines, sheet metal equipment, welding equipment, and more – everything one needs to fashion parts from common metals, plastics, and wood. Some institutions have separate shops for making glassware and other more sophisticated services, such as fashioning electronic or optical components.

"We get about twenty projects a week," says Peter Morley, central machine shop supervisor at the Massachusetts Institute of Technology. "I feel we are so immersed in developing the unusual that we take it for granted when really, it's fascinating."

Many in-house shops are also tasked with repairing and maintaining lab equipment and instruments, so researchers often get experience working with these shops early on in their careers, long before they ever need a customized device.

At academic institutions machine shops tend to charge researchers an hourly rate for their work, because the facilities are expected to be financially self-sustaining. In a survey of a half-dozen shops, rates ranged from about $36 to nearly $60 per hour. After sitting down and talking with the scientists about what needs to be done, the shop supervisor makes a cost estimate based on how long the job might take. The cost is paid out of the researcher's budget.

SORTING OUT SOFTWARE

Sometimes, though, that estimate can be far wide of the mark. As he struggled to develop his automated embryo sorter, Profitt was discouraged by a string of failures as the project stretched into its second year.

<p/>

From left, shop supervisor Fred Letterio, client Dr. Paul Nealen, and instrumentation technician Andrew Callahan, at the University of Pennsylvania biology department machine shop.

Because embryos are orders of magnitude larger than cells, Profitt realized he could not merely modify a FACS machine to accommodate them. Instead, he had to build his sorter from scratch. He replaced the FACS' electrostatic charge-based sorting system with a mechanical one to accommodate the heavier embryos. He also replaced the instrument's round optical cuvette, where the fluorescence analysis occurs, with a larger, square chamber. (A square cuvette is more optically efficient than a round one, Profitt explains.)

At last, Profitt had a device that should have worked – theoretically. And yet it didn't: Tests showed the machine was not sorting embryos accurately enough.

The problem turned out to be a software glitch. As each embryo passed through the optical cuvette, the sorter had about two milliseconds to analyze and "decide" whether to open or close a switch that shunted the embryo to a save or waste group. This time limit shouldn't have been a problem for his computer, but after exhausting other possibilities, Profitt found that the software allowed for as much as a 15 millisecond delay in the decision-making process.

"The computer wasn't doing what I had assumed it would do," Profitt says. After making the necessary corrections, Profitt and Stanford researchers Eileen Furlong and Matthew Scott produced a machine that automatically sorts 15 Drosophila embryos per second with 99% accuracy.1 Their invention, which Profitt estimates cost about $100,000 to develop, promises to speed embryological research by supplanting the time-consuming and laborious task of hand-sorting embryos. "Persistence is key," Profitt concludes. "That, and paying attention to detail and not taking anything for granted."

INVENTING ON THE CHEAP

Given the possibility of significant cost overruns, some inventors prefer to use existing materials and components whenever possible, rather than creating everything from scratch. To make the watertight outer casings for his autonomous underwater listening stations (AULS), Cliff Goudey, director of the MIT's Center for Fisheries Engineering Research, used the discarded end caps of natural gas piping he picked up at construction sites.

The National Oceanographic and Atmospheric Administration deploys AULS around the world to detect the distinctive sound of cod and haddock spawning. Such information helps scientists to pinpoint where and when fish breed, data that is crucial to regulating commercial fishing and preventing over-fishing.

Goudey took the pieces of bright-yellow polyethylene piping he found to the MIT Central Machine Shop, where shop supervisor Peter Morley and colleagues sawed them in half so the electronics could be inserted. The shop then machined grooves and added Oring seals so the halves could be clamped tight for an extended stay underwater. More recent versions of the device use a trawl net float (a large, green, hollow plastic sphere) to encase the electronics.

For the listening and recording instrumentation inside, "The typical approach would have been to make our own electronics and write our own software," Goudey says. "If we had done it that way, we would have had a bulkier, higher-cost device." Instead, he found that a commercially available MP3 player, the Nomad Jukebox, served very well. Using this existing component, available for purchase both online and at local electronics stores, has helped keep the cost below $1,000 per unit. "I'm a scrounger," says Goudey. "I find inexpensive ways to accomplish things."

FUNCTION OVER FORM

When designing a miniature, remotely operated underwater vehicle (ROV) that fits in a suitcase-sized carrying case, Brian Abel, an engineer and president of All Oceans of Aberdeen, Scotland, told his engineering team not to bother with aesthetics. "We're not an aesthetic-design company," Abel says. "We did-n't worry about what it looked like as long as it worked."

The AC-ROV is a commercial device used primarily to inspect the hulls of ships, but it can also be used for underwater archeology or to take the place of a diver for almost any application. In the commercial world engineers are too focused on aesthetics over function, Abel says. This caution can also apply to academic inventors, who may be sidetracked by being overly concerned with form.

Abel's only formal requirement was that the ROV should be small enough to maneuver through an eight-inch pipe. This presented a significant engineering challenge, as the thrusters, lights, camera, sensors, and other electronics had to fit into a tiny space but still be accessible for servicing. Abel and his engineering team solved the problem by making all the parts removable and serviceable from the outside, rather than the inside.

The resulting submersible unit measures 203 mm x 152 mm x 146 mm and weighs 3 kg. And, as it turns out, Abel is pleased with the way the ROV looks and says his clients are, too. "It's cute. It's a babe magnet," he quips. "Women love it. But it wasn't designed to be cute," he adds. "You can't design a form and then build the function into the form. I would argue that because it was designed so well to achieve a function, the aesthetics followed."

WHAT TO DO

If you find yourself in need of machine shop services, your first step is to put your concept onto paper. Most preliminary drawings are more like sketches, produced in generic paint programs or even PowerPoint, says Harold "Buddy" Borders, a shop supervisor at the University of Pennsylvania. But he adds, "I've worked with a magic-marker drawing on the back of a cocktail napkin."

The gold standard among machinists is computer-assisted drawing (CAD) software. Many shops use computer numerical control (CNC) technology, which can read a CAD image and then automatically direct cutting machines such as mills and lathes to create the three-dimensional object depicted.

Though good sketches aren't necessary to get the process started – technicians can begin work with only verbal ideas – a drawing will usually have to be made at some point in the design process. "If we have to sit down and draw the part from scratch, that takes time, and time takes money," Borders says. "But if someone brings in a complete design, the process goes faster."

Make the Most of Your Machine Shop

1. Give technicians the big picture.

2. Scrounge for existing materials and components.

3. Concentrate on function over form.

4. Learn to use computer-assisted drawing software.

5. Take a course in machining.

"It's all about communication, and engineers communicate primarily through graphical means," says Abel. "Not enough people appreciate that." Borders recommends that scientists invest time in learning even an elementary CAD program. He also suggests that researchers take a course in machining. "That can be helpful in learning how to communicate with machinists in a way that gets your ideas across," he says.

To maximize your machine-shop experience, be sure to bring the machinists into the project, explaining to the engineers how the part they are building fits into an overall plan, says Mark Milanick, a professor at the University of Missouri School of Medicine. Otherwise, the final product may not work the way it needs to. Plus, the engineers may be able to enhance the product. "If they know the big picture, they can help by coming up with useful ideas," says Milanick, who has used his university's machine shop to create sensors that can detect molecules of interest in human saliva, urine, and sweat.

MIT's Goudey agrees. "Don't hand off a drawing and hold off on the insight," he advises. "The approach I take is to have a conversation about how a particular part fits in with the other parts. That way, the machinists become part of the team."

Though researchers may be reluctant to do this, both Milanick and Goudey say communicating the big picture ultimately saves time. And they needn't worry about losing intellectual property rights. The shop technicians and scientists say that, as per their contracts, the rights to any inventions they make are owned by their institutions. In the spirit of freely sharing information among scientists, Profitt and his group have published online a parts list and instructions for making their automated embryo sorter.2

ESTABLISHING RAPPORT

At least one institution is looking to formalize the interaction between researchers and machine shops. At the University of Missouri, Milanick oversees a new clinical biodetectives program, which began in January and will enroll about eight graduate students per year. The goal of the multidisciplinary program is to train graduate students to work with experienced inventors and shops to create clinically useful devices.

The ideal goal for researchers, Milanick and others say, is to establish a rapport with their machine shops so communication becomes effortless. When that happens, "You are in a mode where you can hand them a drawing and you know it will come out right because you are on the same wavelength," Goudey says.

Achieving that kind of rapport takes considerable time and effort. In the meantime, the tips presented here should help you to get the process started, avoid some common mistakes, and smooth some of the bumps along the way.

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!
Already a member?