They’ve published videos that show both the fabrication techniques and some success at flying. While they don’t appear to have great flying technique, yet, they’re well on their way.

This fabrication technique certainly has applications much broader than robotics. I look forward to seeing a breadth of creative micro-scale components in the micromanufacturing realm. Has anyone seen other examples of novel micromanufacturing techniques?

FROM STEEL: The Making of a Soulcraft from michael evans on Vimeo.

Filmmaker Michael John Evans sets out to visually portray “the zone” which one enters when their craft is honed. Sean Walling, owner of Soulcraft, builds top notch custom steel bicycle frames. This short film documents Sean’s fabrication methods: a well choreographed dance of experience and muscle memory producing a seemingly effortless ode to process. From Steel: invites the viewer into Sean’s machine shop for an up close and personal look at the work that results in yet another awesome Soulcraft.

Layer built processes create a single part by building up  a series of 2D cross-sections. Different methods require different layer heights and have different means of supporting the layers that are hollow underneath. Of course, one advantage of this process is that parts can be made that could never be fabricated by traditional production methods, such as hollow spheres or even an assembly of multiple integrated parts in a single build that come out of the machine assembled (such as the links of a chain).

SLA (Stereolithography) – Liquid photopolymer (resin) is cured with a laser in layers. After each layer is laid down, the platform lowers further into the resin by the layer thickness, and the laser cures the next layer of material. The part is then post cured with UV light. SLA was one of the first additive rapid prototyping technologies and is still the gold standard.  It is good for general pupose form and fit protoypes and when parts require high resolution, smooth surface finish, or optical clarity.

A part manufactured by SLA (Photo Credit: Key Tech)

SLS (Selective Laser Sintering) – SLS builds layers similar to SLA, except instead of using UV light and a liquid photopolymer, a powdered material (real plastic or metal) is heated and fused together by a laser as a series of 2D cross-sections. SLS is a good choice for functional testing with real materials when smooth surface finish and fine resolution are not required.

A part manufactured through SLS (Photo Credit: Key Tech)

FDM (Fused Deposition Modeling) – Similar to a precision hot glue gun, long strands of real plastic material (ABS, PC, and others) are fed into the nozzle, melted, and deposited in a series of 2D cross-section layers. FDM layers are generally the thickest of the various processes, which limits feature size, but it usually provides better strength and robustness in comparison. FDM is good for prototyping functional parts without small features where surface finish is not important.

A part manufactured through FDM (Photo Credit: Key Tech)

Polyjet – Using inkjet printing technologies, UV-curable materials are effectively “printed” on top of the previous layer to create a 3-dimensional part. Polyjet can produce high resolution parts with decent surface finish, is generally cheaper and faster than most other processes, and is one of the only additive prototyping processes that can produce flexible parts.  It is a good process for small parts requiring good resolution and a decent surface finish, or when flexible parts need to be prototyped.

A part manufactured through Polyjet (Photo Credit: Key Tech)

Cost

Cost can certainly drive one to go with the off-the-shelf solution. If you’re making a low-volume medical device, you might not reach the economies of scale that enable efficient processes and capital investments to drive down cost. Off-the-shelf solutions can save both unit cost and development cost. Plus, prototyping can be accelerated, making time-to-market shorter. However, as a brilliant engineer, you might be able to come up with a less expensive solution then anything out there, especially if it’s for a very specific application.

Branding

Branding can be another issue. If your product looks exactly like a competitor or it’s just packaged in a gray box, that’s a problem. Often, the case is going to be custom for just this reason. For a custom case, you also get to choose colors, shapes, and dimensions to help distinguish your product in the marketplace. It’s less obvious if you’re using a standard piezo actuator to drive an auto-focus mechanism that is hidden inside an enclosure. Of course, if you’re product is a camera, you may want to spend some time making a high-quality, custom auto-focus mechanism so you can use it as a selling point.

Quality Control and Availability

Buying an off-the-shelf component puts you at the mercy of the manufacturer. Their stated tolerances may be less than you need, or they don’t meet their tolerances for every part they ship you.

Then, there’s the issue of shipping your product. If you have a contract to fulfill orders, that’s great for you, but if your quantities are low, you might just be a fly on the windshield of your part supplier.

• Your order may be pushed aside to fill a bigger order.
• The life-cycle may end and nobody tells you.
• Your part cost can go up next year. Sorry.
• The latest firmware release could wreak havoc on your controller application.

Focus

Unless you’re 100% vertically integrated, you’ll have to rely on some degree of off-the-shelf components. You can be a very successful device manufacturer without making your own resistors, screws, LCD screens, membrane switch panels, or plastic resin. Concentrate your energy on what makes your product better. It’s always smart to know your limitations.

Does it exist?

The primary reason to build a custom component is that it doesn’t otherwise exist at or near your target price point. It’s pretty compelling, and probably the biggest factor. If what you need doesn’t exist, custom is the only way to go.

Inspiration for your own custom fabrication

Dave sent me a link to Will Urbina’s video showing the build process for a custom 13TB RAID5 Network Attached Storage device to hold exceedingly hard-drive intensive video editing projects. He could have bought a bunch of external USB hard-drives and made do with a fragmented storage solution or purchased an enterprise product for $10,000. Instead, he’s created a high-quality device somewhere in between. Kudos to the video production quality, too.

Will Urbina is a computer designer and a “video production guy”. You can find out about more of his mod projects at www.willudesign.com. If the videos don’t load correctly, you can find them on YouTube – Part 1, Part 2.

You can read the article in the Summer 2009 printed publication or catch it in the online version, Meeting the challenges of micropart design.

The traditional way to make a product cheaper has always been subtraction – in essence, minimizing the size and complexity of a device without sacrificing its performance.  Size and complexity reductions can drive down costs on materials, packaging, and shipping, while also favoring higher-throughput production and the use of disposable parts – an increasingly important consideration in biomedical applications.  With that said, the simpler and smaller approach is not without limitations, and these limitations are being tested now by “hugely small” applications.

In the case of micro-electromechanical systems (MEMS), microfluidic chips, nano-sensing technology, and numerous other scale-intensive fields, reduced size is actually a profound contributor to increased complexity.  And while these innovative fields show tremendous promise for the future, they currently pose costly manufacturing hurdles as a consequence.  The cost of prototyping and manufacturing micro-parts should be carefully weighed when considering whether or not to pursue an otherwise-avoidable micro-approach.  As of now, these costs can quickly consume the benefits of implementing questionable technology since this often requires several iterations of low-volume custom components.  Lab-on-a-chip devices are a good example prone to this paradoxical limbo.  Even a relatively straightforward microfluidic component can require robust interfaces and innovative prototyping and assembly processes to ensure proper functionality.  Before long, the microfluidic system isn’t so “micro” anymore – in size or cost.

So what can designers and our manufacturing comrades do to advance the cost effectiveness of these emerging technologies?  For starters, let’s abandon subtraction and opt for addition;  additional measures to define and achieve design tolerances, additional manufacturing techniques for creating repeatable micron and sub-micron parts, additional design features for ease of alignment during assembly, additional quality assurance measures to assess as-built dimensions, and – most importantly – additional communication between manufacturers and designers for continued success on the field of dreams we now find ourselves playing.

To me, “rapid prototyping” refers to any fabrication method that can approximate the production part while shaving weeks or months off the delivery time. Three-dimensional printing processes like SLA, Polyjet, and SLS are just part of the product designer’s toolbox. I’ve used processes such as cast urethane, wood tooling (thermoforming), or even machined plastic to quickly create parts that I can hold in my hands, check the fit, and even test depending on the circumstance. In most instances, prototyped parts cannot be subjected to the full spectrum of real world conditions, such as drop-tests or thermal cycles, but it can happen. I’d consider any of these processes to be rapid compared to the production process.

I was recently interviewed for an article regarding the state of rapid prototyping as it pertains to micro-scale manufacturing and product development. From what I’ve seen, the prototyping processes are just not down to the micro-scale, yet. Granted, the micromanufacturing industry is still pretty young, and it’s been growing so quickly in size and capabilities that I expect to see more rapid prototyping solutions soon.

You can find the article and my comments online at MicroManufacturing Magazine.

The organic shape of our MEMS-based windmeter

By now, most mechanical design engineers are used to sending off a solid model and having plastic parts show up in less than a week. While this technique is perfectly suitable and cost-effective for rapid prototyping, it doesn’t quite make the grade for production.

What’s in a model?

3D CAD models can provide a manufacturer with every dimension and feature necessary to make a part and importing the model directly eliminates any errors of translation between paper drawings and the computer-controlled tooling or machining software. Additionally, generating fully dimensioned drawings of plastic parts with complex surfaces is impractical and may even prohibit certain design features as a result. However, part models need the context of a 2D drawing package to fully define the part and its interaction with the whole product.

Highlighting The Good Stuff

2D drawings are effectively the contract between the designer and the manufacturer. In addition to part dimensions, 2D drawings can describe variations in tolerances throughout the part, surface textures, flatness, material specifications, and many more characteristics that the manufacturer must meet. Expectations are well understood and parts can be either accepted or rejected based on this document.

Formerly, 2D drawings included at least three views and dimensions for every feature. These are known as fully-dimensioned drawings. These drawings were required prior to the existence of computer-controlled machining capabilities and were the only way to communicate with manufacturers. As machines become more integrated with 3D CAD software packages, designers enjoy more flexibility, including the ability to design part features with organic, complex, three-dimensional surfaces. With this added functionality comes the complication of adequately communicating those features.

With a single drawing note similar to, “All dimensions shall be held to within ±0.010” of the provided 3D part model unless otherwise shown,” a drawing can be limited to just highlighting the most critical features. This is known as a critically-dimensioned drawing, and it is generally easier to produce and far easier to read. The drawing can include dimensions to features that require a tighter tolerance than the standard, surface callouts, and manufacturing instructions such as limitations for regrind and minimum cure times. A standard dimension can still be included if it is particularly important just to call it to the manufacturer’s attention or to show that it should be inspected for acceptance or statistically monitored as part of the quality assurance program.

Put it all together

In addition to the 2D part drawings, 2D assembly drawings show how all of the parts fit together and list the part numbers and quantities required. This drawing brings the pile of unrelated parts together into a cohesive product. The assembly drawing may also include instructions regarding screw torques, adhesive cure times, or any number of other product-level requirements.

For effective and flexible communication with the manufacturer, the best solution is a combination of solid models and drawings. The goal is to limit the possibility of misunderstandings or miscommunication while recognizing that complicated, fully-dimensioned drawings just are not necessary in most instances. Used well, critically dimensioned drawings combined with 3D models can convey the information required to make a high-quality part as well as or better than our predecessors could.

The conference focused on three aspects of manufacturing – micrometrology, micromachining, and micromolding. Almost every noun is preceded by “micro” in order to differentiate it from the conventional process that spawned it. This is an important distinction because although the processes have high-level similarities, very few of the conventional guidelines scale directly from 100mm to 100um. Generally, the term “micro” is used to categorize processes and components with features ranging from .5um (0.00002”) to 500um (0.0200”), although this is by no means a hard and fast rule even among the conference attendees. The boundaries may be a bit fuzzy because the processes are relatively new and the limits have not been fully defined.

MICROMETROLOGY

Micrometrology involves the technology of measuring parts and features as well as monitoring the processes. In my opinion, it is probably the most mature of the technologies. After all, “if you cannot measure it, you cannot improve it” – Lord Kelvin. Therefore, it’s difficult to know whether micromachined parts, including tools and molds for micromolding, are built to spec without a means to measure them. Manufacturers are creatively utilizing off-the-shelf components such as mirrors, cameras, lasers, and other common, conventional non-contact devices to achieve higher resolution. Additionally, they are creating newly developed technologies, such as interferometers and microvoltage detection, to improve the integration of metrology into the machining and molding processes. For example, one manufacturer presented work that measures the tool-tip location of their µEDM machine based on very small detected voltages so the user can monitor the process and measure the part without having to re-register it to a different machine. This type of innovation is representative of what many of the manufacturers are doing to bring new tools to the industry.

MICROMACHINING

Micromachinists are definitely thinking outside the box when it comes to creating features that are on the order of just a couple of microns. They’ve developed 30um wires for Wire EDM and 5um spade bits to remove material. Additionally, they’re requiring very specific machinery to even use their tools. A 5um bit needs a feed rate around 50mm/sec and a spindle speed over 50,000 rpm. A standard CNC spins closer to 15,000 rpm. Additionally, if the runout for a conventional machine is 1 or 2 um, most conventional machinists would be overjoyed, but that would destroy even a 10um bit on contact. I found that most contract micromachinists were focused on low-volume (1 – 30 parts) with the possibility of higher volumes entertained but not necessarily preferred. Setup time and part inspection can be difficult for such small parts, but I expect there are people interested in the higher-volume runs out there.

MICROMOLDING

Although the processes are far from standardized, the representative micromolders presented an impressive array of precision parts and capabilities. Many of the plastic parts were much smaller than a single plastic pellet, with wall thicknesses and features on the same order as the tolerances of most conventional parts (130um (.005”) to 260um (.010”)) with the limits reaching down into the single-micron range. At this scale, it’s notable that the annual requirement for raw material might only be a handful of pellets! Therefore, the micromolders are limited, with just a few possible exceptions, to stock materials that were originally designed for molding much larger parts on much larger machines. However, micromolding processes require much smaller components, for instance feed-screws are much smaller and internal pressures are much higher as a result of the smaller gates and runners. Therefore, the processes won’t be fully optimized until raw material manufacturers start to find the value in supporting the industry. With some specialized conventional materials costing $3,000 to $20,000 per lb, it may not take long for them to catch on.

With regard to rapid prototyping technologies, I found one vendor that quoted a one to three week turnaround on precision molded parts. I’ve been spoiled by the one to three DAY turnaround on conventionally sized parts, so it wasn’t exactly what I was hoping for. However, it’s much better than ramping up to full production over the course of two months.

Recent work in microfluidics has also required me to look into high-volume molding production capabilities. Of course, the problem with molding very small parts in high quantities is that it is difficult to get multiple cavities to fill evenly, precisely, and reliably with such small tool features. I was pleased to find two vendors with very different yet promising approaches to the problem.

MICROMACHINES

A notable advancement in the micro-scaled industry is the miniaturization of the machines themselves. As parts have grown smaller, manufacturers realized it was wasteful, with respect to floor space, energy, fixturing, and cost, to have conventionally-sized machines performing work on parts that only require 1/10^th of the workspace. This is true almost across the board, as milling, EDM, and molding machines become small enough to fit in your office.

A SUCCESSFUL CONFERENCE

At Key Tech, we’ve been designing micro-scale parts for some time, but finding vendors capable of manufacturing such parts has been difficult. I went to this conference to better understand the state of the entire micromanufacturing industry as well as identify vendors with specialized capabilities. I think the conference was a success on both counts.

THE LIST OF EXHIBITORS (FROM SME.ORG)

• Accumold
• ALBA Enterprises
• American Swiss Products
• Atometric
• BIG Kaiser Precision Tooling
• Carl Zeiss IMT Corp.
• Datron
• Deringer-Ney
• DRC Metrigraphics
• ESPRIT/DP Technology Corp.
• GF AgieCharmilles Corp.
• Kern Precision
• Kleiss Gears Inc.
• Kyocera Micro Tools
• Lion Precision
• Makino Inc
• Microlution
• microPEP
• NanoRite: A Center for Innovation
• National Institute of Standards and Technology
• National Nanomanufacturing Network
• Phillips Plastics
• Sarix SA
• Small Precision Tools
• SmalTec International
• SMC Ltd.
• Sodick
• Sodick Plustech
• T Bryce—EDM Tech Center
• Teamvantage
• Third Wave Systems Inc.
• TNI Manufacturing
• Top Tool Company
• University of Minnesota
• Vacco Industries
• Veeco Instruments
• Whittmann Battenfeld Inc.
• Yasda Precision America

Some MicroMachines you might remember from decades gone by,