Reading time ( words)
Editor's Note: This column originally appeared in the July issue of The PCB Magazine.
Printed circuits boards have essentially been made the same way since the 1960s. A copper-clad substrate is prepared before being fully covered in a photosensitive resist. A phototool of the circuit image is then used to mask parts of the board from light in an exposure unit. The unexposed areas of the resist are then developed before the panel moves on to etch and strip.
There have been several changes in imaging techniques over this timeline. Taped artworks were still being used throughout the industry into the 1980s, and a significant portion of revenues came from photographic reduction of these hand-taped artworks. Once reduced to size, other photographic techniques were used to merge positive and negative films into power planes, and then step and repeat the images at required spaces. This was a laborious process and prone to error--hence the use of photoplotters.
These early plotters were vector based and driven by Gerber data, a simple code containing X-Y co-ordinates and size, flash or draw instructions. Photoplotted artworks were of the highest quality, but the process was slow and it still often required photographic techniques for pos/neg merges for power planes. In the mid-1980s things changed quite radically with the advent of laser plotters. This new technology tackled all of the historical problems of time and cost and was able to produce pos/neg merges as a single plotted image; the merge having taken place in CAM beforehand. As with any new technology, laser plotters were, for a time, the subject of scorn from traditionalists who argued that laser films would not be accepted by manufacturers because the image, particularly of angled tracks, was not a smooth line.
Fortunately, pragmatism prevailed and laser plotters are now the imaging workhorse of the PCB industry and have been for the best part of 25 years. Most statistics suggest around 95% of printed circuits are produced using laser plotted artworks. The remainder are being imaged by the last of the technology shifts--Laser Direct Imaging, or LDI.
LDI brings with it a couple of clear benefits over artwork, namely accuracy and registration as the image is not subjected to film movement. Consequently, the technology has found its niche at the very high end of PCB manufacture where track of 3mil and below are used to produce boards for smart phones and the like. Though often considered a digital technology, one school of thought argues that LDI is an extremely accurate and very expensive exposure unit in that, with the exception of artwork production, all other aspects of process remain intact.
There we have it: A plotted history of imaging techniques adopted by the industry since the 1960s. Perhaps the most striking aspect of this story is that while there have been dramatic improvements, there have been few step changes. Intelligent imbedded data forms like ODB++ are gaining prominence, but the large majority of PCBs are still manufactured from simple Gerber data. High accuracy LDI technology is available, yet most PCBs are produced using films imaged on a laser plotter--technology that has been around since the mid-1980s.
There’s no doubt in my mind that LDI will remain a key technology for tight geometry PCBs, the sheer volume of LDI’s sold over the last two or three years underlines that. Additionally, the 95:5% ratio is likely to remain unchanged for the near future; a consequence of price erosion and increased layer count on mainstream PCBs.
Against this backdrop, there have been several attempts to bring apparently cheap and easy inkjet techniques to PCB manufacturing. These offerings have typically been introduced by equipment vendors attempting to increase their product portfolios by adding print heads and ink delivery systems to a modified X/Y table. Most have at some stage offered inkjet for legend, solder mask and etch resist, but in the main, inkjet has only been successful in legend applications, hence the wide choice of machines offering this particular solution.
It’s worth taking a few moments to ask the question, why is this? Why has inkjet only found a niche in a relatively low technology application, imaging text, numbers and outlines? The simple answer is that, despite its general perception of a simple technique used and understood by everyone, inkjet is actually a notoriously complex and inherently unreliable process to implement successfully in an industrialized setting. To date, no one has really grasped the nettle and tackled these issues. This is about to change.
Later this year, MuTracx will release Lunaris, a fully industrialized inkjet printer for the imaging of printed circuit etch resist. The technology base of Lunaris is Océ Crystal Point Technology, a project focused on the Engineering Drawing Market, in which Océ has a significant market position with the ColorWave 600. The product is based on Océ’s own proprietary Print Head/Toner combination. High productivity is achieved through a combination of an accurate print head and dot positioning on the media. The goal of Lunaris is to produce the first reliable inner layer printer based on inkjet technology. It will offer high yield and productivity with the goal of 100% correct panels into the subsequent process steps that will be enforced with the introduction of an inline process validation.
Lunaris: A fully automatic, double sided printer of PCB inner layers.
As mentioned earlier, PCBs have essentially been made the same way since the 1960s. In all, there are 15 distinct process steps between CAM and inner layer preparation and inspection.
- 15 distinct processes to layup and bonding;
- Yield constraints;
- Registration constraints;
- Costly and complex;
- Infrastructure constraints (clean room, photomech, etc.); and
- Environmental constraints.
Despite the heavy and ongoing necessary investment, fabricators constantly battle yield and registration constraints, not to mention the environmental impact of specialist chemical disposal and product waste.
Lunaris eliminates 11 of 15 process steps and makes zero waste a reality. Historical yield and registration problems are addressed, as is the environmental impact, which is removed completely.
- Replaces 11 of the 15 process to layup and bonding;
- Removes environmentally problematic processes (microetch and developer);
- Removes need for costly and specialized infrastructure (Lunaris is a clean machine);
- From data to print in minutes;
- Tackles registration problems;
- Removes the developer process, a significant cause of inner layer short circuits; and
- Resist not in end product therefore no approvals required.
Lunaris is a fully automatic machine, with a throughput of between 60 and 120 double-sided cores per hour, dependent upon chosen resolution. Side-to-side registration is guaranteed by using high accuracy fiducials applied to each substrate upon entry to Lunaris and the systems own optical inspection unit to detect their positions.
Each core is moved from stage to stage on an air cushion to prevent handling damage and held in place under vacuum. Each stage is temperature controlled to eliminate thermally induced distortions. To achieve real-world production throughputs, Lunaris uses 60 print heads, which are secured in the static print beam above a perfiplane table, utilizing leading-edge mechatronic techniques and high accuracy sensors to control the position of the print surface relative to the print heads within a micron envelope.
Lunaris is unique in that it is the only product offer in which the print heads and the etch resist are made by the same manufacturer. Océ’s print heads have a failure rate of 1 in 1 billion droplets without cross talk or satellites. The resist is supplied in solid form in 5KG canisters and then melted to operating temperature within the Lunaris system. When jetted, the resist remains in a gel phase during printing, allowing it to flow on the substrate and to interact with subsequent droplets jetted within the gel phase window. The resist enters the crystallisation phase, allowing the substrate to be handled and turned for double-sided printing.
Drop flight and position is controlled by mechatronics and the print head. Drop flow is managed by a mathematical print strategy in which we determine necessary droplet positions such that the resist will flow into the desired contour in the gel phase before crystallisation takes place.
However, even the best print heads fail and the consequence of such a failure is catastrophic in this application. To achieve production throughputs, Lunaris fires up to 50 million droplets of etch resist per second. Even with a one in one billion failure rate, this means a potential problem every 20 seconds. The single biggest cause of failure is air entrapment, which occurs on every piezo print head in existence. Initially, small air bubbles form around the nozzle and cause the resist to spray--spurious copper in this application. In time, the small air bubbles combine to form one large bubble that will block the nozzle completely--open circuits!
By listening to the acoustic profile of each print chamber, Lunaris is able to predict potential problems more than 10,000 droplets before they occur. When detected, the nozzle in question is switched off, allowing the air to dissolve before bringing the nozzle back on line. All panels are printed without down time, operator intervention and in the knowledge that all droplets are present.
Finally, the scanner used in fuducial inspection is once again employed, this time to inspect each core automatically and fully prior to output. We are therefore able to guarantee 100% yield into the etch line and zero copper waste at all times. All of this takes place within Lunaris’ own micro environment--a plug and print clean machine.
Primary resist imaged on Lunaris.
Specification of the release product is as follows:
- 100µm (4 mil) lines with 60µm spacing - 75µm/75µm (3mil/3mil) post etch;
- Maximum image size 21” x 24” (Extended version for 24” x 30” possible with current design);
- Maximum ine edge sharpness 7.5 µm wave;
- Positioning tolerance Front - Back: 25 µm over total 21”x24”image;
- Positioning tolerance same side: 10 µm over total 21” x 24” image; and
- Throughput of between 60 - 120 double sided cores/hour dependant on resolution.
Lunaris will be on a pay-per-usage basis in which users will only be charged for good panels entering the etch line. This pay-per -panel method includes the cost of all consumables (heads and resist), maintenance and even Lunaris itself.
Stuart Hayton is the Global Sales & Marketing Director for MuTracx. He has more than 25 years experience in the PCB industry in technical, commercial and executive management roles. In 2008, Stuart moved to the Netherlands to be involved with the Océ Technologies Lunaris project. For more information, contact Hayton at firstname.lastname@example.org.