Why we need a rapid prototype model where the end-product is “grown,” not built from disparate parts.
I’ve been working in the electronics product design and fabrication industry for 18 years, and over that time I’ve seen a lot of technology evolve. The resounding issue that occurs to me every time I implement DfM or DfT techniques in a new design (I do about 130 a year) is the true lack of integration of manufacture with regard to the mechanical packaging of electronics products and overall function.
My colleagues grow weary of my incessant emoting concerning the need to move from planar PCB fabrication and subsequent tedious mechanical mounting of components, subsequent interconnection and packaging, and the disparate steps involved in final assembly of a product toward a fully intrinsic/monolithic approach – one based on techniques such as those used for rapid prototyping of mechanical parts, similar to fused deposition modeling (FDM), whereby a device is a true intrinsic unit basically grown from the various materials required for form and function.
For example, instead of purchasing individual components such as tantalum capacitors, one merely purchases the raw tantalum compounds, licenses the fabrication formula from, say, Kemet and integrates this into what I describe as a “liquid chain,” one in which all materials in their raw form are combined on a molecular level to grow the desired product. This includes all the semiconductors/conductors, other passives, displays, and so on, which are enveloped in the housing materials. As in FDM and stereolithography (SLA), this could be a layered process combined with other processes analogous to those now used in pure semiconductor fabrication.
I use tantalum capacitors as an example because, a few years ago, there was what many trade publications deemed “a shortage of tantalum capacitors” that was incorrectly attributed to a shortage of tantalum. I knew some people associated with the Kemet factory in North Carolina and called one up, mentioning what I had read. He unequivocally stated that his facility currently had plenty of tantalum material in storage bins, the issue being one where he was unable to get the computer-controlled machinery required to make the actual parts. I jokingly suggested that it might be possible that the manufacturer of this specialized equipment wasn’t able to allocate enough tantalum caps from their distributor/supplier, as they were a bit of a niche buyer. Long silence on the other end of the phone followed.
Another issue is that if you look at the amount of material waste in just packaging printed circuit board components for shipment and delivery to EMS facilities for use on automated pick-and-place machines, the advantage of a monolithic approach is very apparent. Just look at any TI datasheet; at least three pages are devoted to reel/tube orientation and packaging, materials that themselves have to be tooled and eventually are discarded (or hopefully recycled), enveloping more mass, process and time than the actual component fabrication itself.
A distinct advantage in striving toward monolithic methods is the apparent advantage in high-speed and RF/wireless design, where as a designer, I now can use a 3D space for interconnection of controlled impedance waveguides, instead of being limited to a simple 2D planar approach trying to fudge microstrip and stripline topologies into a working subsystem that will still be fraught with failure due to interconnect processes, thermal issues, and subsequent issues involved in using soldering processes that really haven’t changed since the early 1950s. And now with the problems created with the higher reflow temperatures required for processes needed to comply with RoHS initiatives, as outlined by Dr. Howard Johnson and others at NASA1, further refinement of current production methods is just simply putting a Band-aid on a broken leg.
With some modification, current EDA and parametric mechanical software tools could be tailored to provide post-processing for these fabrication techniques, but with considerable advantages over the current methodology that relies on distinct and discrete fabrication steps – chip/passive component fabrication, PCB fabrication, mechanical tooling and fabrication – resulting in a much tighter, more efficient production process with a significant decrease in environmental impact, less waste, and much higher yields.
Imagine your cellphone as being a single block of function, with a minimum of custom components, replacing all mechanical actuation with touchscreens (as is already being done), with no need to use all these secondary operations for sub-assemblies. This is what we need to strive for.
We’re heading toward another “tyranny of numbers,” similar to the one Jack Kilby faced when presented with silicon micro modules, a situation that led him to design the first monolithic IC chip.
References
1. Howard Johnson, “Rolling Back the Lead-Free Initiative,” High-Speed Digital Design Online Newsletter, vol. 10, no. 1, sigcon.com/Pubs/news/10_01.htm.
Wayne Mitzen is a cofounder of Fast Product Development (fast-product-development.com), and holds US patents for RF devices, imagers and wireless intrusion detection systems; This email address is being protected from spambots. You need JavaScript enabled to view it..
Question: What is the difference between a flexible circuit and a membrane circuit? Are they both considered flexible circuits? How do I know which to choose for my application?
Answer: Historically, the term “flexible circuit” or “flex circuit” has implied a circuit with photo-etched metal conductors (generally copper) on a film substrate (generally polyimide). Many of the manufacturing processes involved in building flexible circuits are very similar to those used in manufacturing rigid printed circuit boards. The main difference is that virtually all the insulating and bonding materials used to make flexible circuits are much thinner and generally unreinforced (films), as opposed to the glass-reinforced laminates used to construct rigid PCBs. The term “membrane circuit,” on the other hand, typically refers to a circuit that has printed conductors (usually screen-printed) formed using conductive inks. Membrane circuits are created using additive processes, whereby all conductive and insulating materials are added in layers to the base material. Many conductive inks can be used to create the conductive patterns, but most rely on silver as the conductive component. The base insulating materials that membrane circuits are printed onto are typically polyester films.
Determining which of these variants is best really depends on the application and how the circuit will be processed and used. The decision to use membrane circuits is typically driven by cost. Polyester materials are much less expensive than polyimide, and the cost of forming silver ink conductors is significantly less than the photoetching process required to form copper conductors.
However, with reduced cost come many significant limitations. Polyester materials cannot tolerate the high temperatures that polyimide can withstand. For this reason, the upper service temperature will be limited to under 105˚C with PEN, and even lower with other polyester materials. When screen-printing conductors, it is not possible to define features as small as can be done with photo-etching. For this reason, the wiring density of membrane circuits is significantly less than photo-etched polyimide/copper flex circuits. Connections to connectors and SMT components are typically limited to ZIF style connectors, insulation displacement contacts, or silver ink in lieu of solder to attach components. Silver conductors are designed to handle small signals, so electrical current must be limited to 100 MA or so. Last, the additive processes used to create a membrane circuit limit the construction to a just a few layers. The vast majority of membranes in use right now are just one or two layers, with a very small number that add another two to three layers. Since it is not possible to use copper plating to interconnect layers, the layer connections are made by forming openings in the dielectric layers that permit the conductive ink from one layer to connect to the layer above or below.
As stated, copper/polyimide flex circuits are more expensive than polyester membrane circuits. But with the added cost come additional improvements in capabilities and performance in some important areas. For instance, most flex circuit manufacturers can easily make copper/polyimide circuits with 8-plus conductive layers. In addition to the extra layer capability, many vendors can supply circuits with conductor widths and spaces that are well below 5 mils. This results in the ability to pack much higher wiring density into the same amount of space as a comparably sized membrane circuit. If even higher wiring density is required, the flex circuit manufacturer can move to a rigid-flex design, permitting the layer count to increase to 20-plus conductive layers. (Caveat: 20-plus layer rigid-flex circuits are really expensive.) The other significant benefit of copper/polyimide flex over a membrane circuit is a much higher temperature rating. While a flex circuit will not tolerate temperatures as high as those a rigid PCB is capable of, they can withstand soldering temperatures for short periods of time. This permits connectors and components to be assembled using standard pick-and-place and reflow equipment to populate the circuits. And copper/polyimide flex circuits can operate for extended periods at temperatures well above what a membrane circuit can tolerate. Finally, copper conductors can carry vastly more electrical current than similar-sized silver conductors.
When weighing options in choosing your flexible interconnect, consider cost and also performance factors to determine the best fit for your application. Membrane circuits and copper/polyimide circuits occupy their own unique niches in electronics interconnection. Choosing the right one for your application will ensure the best overall value. As always, do not hesitate to contact your friendly neighborhood flex circuit (or membrane circuit) manufacturer for advice. Having seen thousands of applications, they are uniquely qualified to guide you to the right choice.
Mark Finstad is a senior application engineer at Flexible Circuit Technologies (flexiblecircuit.com); This email address is being protected from spambots. You need JavaScript enabled to view it.. He and co-“Flexpert” Mark Verbrugge from PICA Manufacturing Solutions (This email address is being protected from spambots. You need JavaScript enabled to view it.) welcome your questions.
Undetected mask remnants can upset the final finish.
The arch-nemesis of final finish is solder mask. I am an expert in surface finishing. I understand the need for solder mask, but it is my right to dislike it. Truthfully, it makes my life more difficult, but those challenges help pay the bills. Solder mask brings out the ugliest in a surface finish, or so that is the perception. Research continues on how to build the widest operating window for a final finish, even when the problem falls outside the final finish line. I like to tell our development teams, “If research were easy, everyone would do it.”
Two main issues are associated with the solder mask process that later affect surface finishing: improper exposure and solder mask residues. Improper solder mask cure can affect the foot characteristics at the interface of the trace. If undercured, the solder mask will suffer chemical attack from the plating baths. This could result in one of two defects. Solder mask constituents leaching into the plating bath cause instability or plating-performance defects. Or, the chemical plating bath attacks the adhesion between the copper trace and solder mask. This creates a crevice for chemistry to become trapped. If overcured, solder mask can become brittle and fracture. Solder mask residues cause plating defects that may go undetected until the panel is post-assembly. Overall, improper solder mask application can wreak havoc on the final finish process.
The most infamous example of undercured solder mask effect on surface finishing was discovered in the early 1990s. Industry found a new post-assembly defect called “black pad.” Hot electroless nickel baths are the perfect environment to leach sulfur-bearing materials out of solder mask that has not properly cross-linked. This causes an increase of the “wrong” sulfur into the EN bath, which then co-deposits and alters the desired phosphorous content. The higher the sulfur in the EN deposit, the lower percent phosphorous included. Also, the type of sulfur included in the deposit can result in less corrosion resistance. This leaves a weakened nickel deposit that is more susceptible to corrosion in the electrolyte of the immersion gold bath. As the immersion gold bath corrodes the nickel, it leaves behind a phosphorous rich layer, which has poor solderability characteristics. The solder joints are weak, and the potential for components to pop off the PCB are greater. I still cannot resist saying, “Oh, there goes the BGA,” every time I drop my cellphone.
The second scenario observed when solder mask is undercured at the foot is a galvanic corrosion of the copper trace. This can be observed after immersion tin plating. All immersion tin baths contain thiourea; it is this material that drives the reaction. Without it, tin would not naturally plate on a copper surface. So the thiourea and acid in the immersion tin baths solubilize the areas of mask that are not cross-linked, leaving a negative foot. This crevice has a small geometry, which is difficult for solution exchange. Plating chemistry gets in and quickly depletes the metal, but then cannot get out for fresh replenishment of more metal. In essence, the solution trapped in the crevice has a high acid and thiourea content, which accelerates corrosion. As the uncured solder mask is removed from the foot and the copper trace corrodes, there is no longer an intimate connection between the mask and trace. This leaves a ledge of mask. Tape tests reveal solder mask adhesion issues around the pad edges at this interface. It is suggested to UV-bump panels prior to immersion tinning. This will help to cross-link any areas that did not go to completion in imaging. It should be noted that the darker the solder mask color, the harder they are to properly cure.
The most problematic defect coming out of imaging is residues on the pad surfaces. This is not cured solder mask; it is normally junk from the developer that redeposits on the board’s surface. The culprit is poor control or maintenance of the developer. Unfortunately, once the panels go through post-bake, these residues do cure and become extremely difficult to remove. The fun part is that solder mask residues are normally very difficult to see by eye, so they go unnoticed until after surface finish. Solder mask does not discriminate; it will prevent any non-HASL surface finish from coating properly. Depending on the degree of residues, it can go undetected even after the surface finish. But rest assured, it will cause a solderability problem.
So there are a few ways to address a solder mask residue issue. You can use a good cleaner designed to remove solder mask residues, but depending on the quality of the exposure, if the foot is soft, you will remove it and create the same crevice discussed above. Another option I really like is jet pumice. Unfortunately jet pumice gets a bad name, as it is associated with its cousin pumice (slurry) scrub. Pumice scrubbing utilizes a nylon brush that mechanically abrades the surface with the pumice slurry. The problem associated with this process is that the brushes can embed pumice into the copper. Once in, it’s not coming out. But jet pumice has no mechanical component; cleaning is driven solely by impingement, and it is a great tool for undermining solder mask residues on the copper surface.
As always, proper process control will lead to successful quality product. Don’t discriminate: There are multiple ways to create good product. Understand your options.
Lenora Toscano is final finish product manager at MacDermid (macdermid.com); This email address is being protected from spambots. You need JavaScript enabled to view it..
Designing pretty boards doesn’t have to take longer.
Under pressure to reduce costs and improve time to market, the PCB design process is often a target, and rightly so. Not only is software continuously enhanced to reduce cycle time, but there is pressure on designers and design teams to increase productivity. As someone who has been in the PCB design industry for over 45 years designing and working with software engineers to develop new methods and tools, I believe I can offer some insight into making each PCB designer more productive.
I grew up crafting boards with a desire (or, some would say, an obsession) labeled today as the “Artist Syndrome.” Time was the artistic side of the PCB designer was considered a badge of honor. Today it is sometimes stereotyped as a drag on productivity. Hence, over the past decade, the term “syndrome” has been tacked on to describe those of us who apply our artistic talent to design.
I know times are changing, and those of us who still appreciate and respect the art of PCB design need to change with them. The good news is we don’t have to lower our standards. Rather, we need to apply new methods that enable us to be even more productive without abandoning our artistic edge.
There is a secret about the art in PCB design. It is not about creating artwork; rather, it is about the designer’s efforts to obtain a quality result through precision and efficiency. As much as the first-generation PCB designers (and those who have followed in their footsteps) like having the artist label, it is actually a misnomer due to a misunderstanding of what we are trying to accomplish. Yes, the design looks like a work of art, but that is only because the designer became obsessive about locating every component, trace and via in a manner that fulfilled their objective and subjective criteria for quality.
Objective quality means the design fulfills the behavioral requirements for the circuitry, and it is optimal for assembly, fabrication, test and reliability. Subjective quality means the precise and efficient location of all the PCB objects in order to fulfill the objective quality requirements turns out to look like a work of art. The artistic appearance of a design is a secondary effect of designing with a goal of quality.
Each year Mentor Graphics conducts the Technology Leadership Awards. [Ed.: UP Media president Pete Waddell is a judge.] Customers submit designs to be judged by a panel of industry experts, with awards presented to the best designs in various technology segments and one singled out for the best overall design. The criteria evaluated include efficiency (component and route density) and complexity (multiple types of circuits, difficult to route components, high-speed requirements). What’s interesting about the winning designs is that they all look beautiful, like art. Sometimes a very complex design was done so well, it looked simple. So let’s remember that the goal isn’t art; the goal is quality, and the result often looks appealing as well.
Quality as the goal. We should be able to agree that a design with objective quality is a fundamental requirement. Now we need to address the issue of productivity. How can a designer used to being very precise to make an efficient design complete it faster and still maintain the quality?
Answer: Trust the software. This is not like Luke trusting the Force by fully surrendering to it. It is more like trusting the software when you can. I certainly understand that designers have been unwilling to surrender to software automation for one reason: quality. Historically, designers often say they don’t use autorouters because it takes more time to clean up after them than it does to route it manually. This has been true for years, but again, times are changing. Obtaining quality results is a requirement in software development. Developers now understand that unnecessary vias, excessive meandering, poor pad entries, or anything else that compromises fabrication or performance is undesirable.
Designers should evaluate the latest software on a regular basis and determine if they can be more productive, while maintaining the desired result of precision and efficiency. The next generation of designers will also benefit from tools that enable quality results without requiring an “artist” moniker on their résumés.
Charles Pfeil is engineering director at Mentor Graphics, Systems Design Division (mentor.com); This email address is being protected from spambots. You need JavaScript enabled to view it..
Nick Martin, the founder and, until last week, CEO of Altium, is fighting back against the board that tossed him out.
But the real question isn't whether he will regain his spot atop the CAD tool developer. It's why the board saw fit to relieve him of his duties in the first place.
Some contend privately that at least one board member wants to sell the company but that Martin, who is the company's largest shareholder, has been reluctant to go along. If so, pushing him out would mean removing, in part, one big barrier. For its part, the board has publicly stated that the decision to leave was Martin's -- something he vehemently contests, and which seems unlikely on its surface -- and that the company has not returned the type of shareholder value the board seeks.
So while it's true the move to Shanghai coincided with an improved bottom line and a higher share price, it's also true the stock hasn't topped $1 in years (see chart -- the current price is about 80 cents).
No one is getting rich owning Altium right now. If the board is getting antsy, it's understandable. Whether that merits replacing Altium's answer to Steve Jobs -- a design visionary who, according to many we've spoken with, has always put the technology first -- is for the historians to determine.
