Convergence of speed and flexibility in assembly

The concept of speed in SMT assembly needs little introduction. Most equipment suppliers specify speed in terms of thousands of Components per Hour or cph and most SMT assemblers think in terms of completed assemblies per hour. More sophisticated measurements account for process yield and thus take into account the effects of print and reflow process variables as well as pick and place precision. In summary, speed is easy to understand and easy to measure.

On the other hand, the term flexibility as used in our industry can be harder to pin down. For the purpose of this article we are talking about the ability to handle a wide range of end product assembly variations while maintaining high utilization and economic output. Thus, a flexible line can build three variants of cell phone for a profit today and just as easily and efficiently build a high-end server board tomorrow. While this may seem a strange proposition to the OEM, it is completely normal in the fluid world of EMS and ODM assembly where the loyalty of customers like Dell, Apple, Cisco, Motorola and others can be measured in pennies.

In the field of component placement, trying to achieve both speed and flexibility together on a reliable and consistent basis has proved to be a tough challenge for manufacturers and equipment designers alike. The result has typically been to opt for speed at the expense of flexibility. This is natural because speed is easy to understand, easy to engineer, and easy to market and sell. Building in flexibility requires thinking about what will come next. It isn’t easy to predict the future and so we tend to revert to what we know for sure. However, as more and more assemblers look to diversify their customer portfolios, this short-sighted approach has led to unintended consequences.

Does speed mean the same to everybody?
Pioneering vendors in the assembly equipment arena were quick to identify the ‘speed with flexibility’ issue, and all began developing various strategies to meet the changing needs of assemblers. These first logical steps centered on either improving the speed of an inherently flexible line configuration, or introducing flexibility to a line known for its fast throughput. But how exactly can you judge speed improvements related to placement?

While a basic measure of throughput per machine speed is Components per Hour (cph), it can also be viewed as the throughput of the placement section of the production line. Perhaps it’s better expressed by the ability to change from one product to another in an efficient manner, while maintaining high utilization of all the equipment in the line. In which case it should be measured across a range of products coming off the end of the line as that’s what modern manufacturers need to do. And that’s not simply throughput but productivity. With any interpretation, everybody’s perception is different, but some or all of these factors may affect the choices that an assembler must make in pursuit of the productivity sought.

Characterizing flexibility
We have stated that for this discussion, flexibility is the ability to handle a wide range of end product assembly variations while maintaining high utilization and economic output. However, there are other flexibility considerations too, for instance as a hedge against technology obsolescence. The current leading-edge but increasingly popular packageon- package (PoP) assembly (Figure 1) process demands sufficient flexibility from a placement head to position a high-density logic device, dip the top-mating memory package in flux, then return precisely to the original position to place it directly on top of the first device. Is your assembly solution flexible enough to accommodate these demands, and even more importantly, to cope with whatever the next generation of mobile technologies brings?

Flexibility can also be determined as the ability to handle a wide range of SMT component technology, be it leaded, bumped, chip scale package, pin-in-paste, etc., or to handle any lot size from small to several hundreds of thousands, maintaining high utilization of all equipment modules in the line. (Figure 2) And then there’s the type of flexibility that allows new products to be introduced into manufacturing at a small scale, defining and testing new components, boards, processes, etc. along the way, and then ramp rapidly to volume production for those products that catch the market’s imagination.

In summary, flexibility is crucial for short batch runs and frequent product changes, where downtime for changeover is typically more critical to overall productivity than the assembly lines’ throughput while running.

Changing needs
With the advent of mass customization, these needs converge: even stable products are not immune to the requirements of different geographic sectors, which dictate small but important changes to the base product that must be accommodated during manufacture. In addition, emerging consumer markets also demand better technologies more quickly, leading to shortened product lifecycles. OEMs and their manufacturing partners who can not rush a new product, or an enhanced variant of an existing product, to market fast will miss the boat as competitors carved into their share. Look at the cell phone handset sector as an example.

So now what assemblers need is maximum throughput speed at line level while they are running, plus the flexibility to rapidly implement frequent product changeovers. As well as intrinsic SMT assembly capability, the implications of this extend to process and logistics procedures with issues like broader component inventories, board size changes, line balancing between machines, software upload and variant control to accommodate.

Right-sizing your line
Selecting the right assembly equipment is only part of the answer. To deliver optimum gain, a line-wide or even factory-wide perspective of productivity proves more beneficial than simply considering individual machine throughput. ‘Right-sizing’ of lines can be advantageous. This is the manufacturing model where a small line is configured with sufficient capacity for each particular build – sometimes called a ‘fixed family’ setup. A multi cell approach comprising many small lines together can make sound sense for productivity by reducing the likelihood of product changeover impacting factory-wide productivity. As a manufacturing philosophy, this contrasts with a monolithic line that turns out product at massive speed for only a short time before enduring a comparatively long changeover routine.

Smaller cells (lines) can be used to produce lower volumes without changeover or used in parallel to produce higher volumes of a particular product. One key benefit is maintaining higher utilization by enabling changeover of a particular cell while keeping the other cells productive. This approach wastes less time setting up and tearing down products. From a capital expenditure perspective, perhaps the only disadvantage of a multi-cell approach is that you need extra infrastructure in terms of printers, conveyors and ovens to ensure that each line is kept supplied with products.

Falling somewhere between the two are modular platform lines expressly configured to accommodate a number of products and variations. These will feature two, three or four gantry-style platform placement machines (essentially identical but with potentially differing capabilities) with enough on-machine component inventory between them to deliver a diverse range of device types to eliminate the need to change feeders, reels or trays to supply components for a new product coming on line. This is a typical manifestation of a highly flexible line, known for its ability to handle a wide range of parts and balance the workload between machines. But while immensely flexible, accurate, reliable and productive, these kinds of lines have not been synonymous with high speed.

Another route to speed – avoiding slow-down What it is that causes a placement machine to slow down? Component size is probably the most obvious reason – the larger the component, the higher the inertial and momentum forces when traversing the board on the spindle of a placement head, and the more likely it is to move on the nozzle due to excessive acceleration and deceleration, or directional changes. On placement machines that employ a moving table, such as conventional turret chipshooters, the same principles apply even after the component has been placed on the board – even a tiny movement of a fine pitch multi-leaded device can cause lead to pad misalignment which could easily render the board useless after reflow.

The latest generation of gantry-style platform machines that feature fixed tables offer significantly higher confidence in this regard, routinely achieving full-speed operation, even with components up to 30mm x 30mm. Here, equipment design plays its part. For instance, the possible cantilever effects caused by driving gantries from only one end can also lead to de-rating of placement speed, since time must be allowed for oscillation to settle between motion actuations before the component can be accurately placed. The answer? Drive it from both ends simultaneously. The message? Check out the design before you buy.

Similarly, nozzle design can assist: a large QFP part benefits from a larger nozzle to hold it securely during the motion phase between the pick and place operations. But you wouldn’t want to use that nozzle for an 0201 device. So how many different nozzle types do you need? Or more importantly, what is the capability of the placement head in terms of component range and how much flexibility exists with the nozzle selection? Something else to consider when specifying equipment.

Ultimate speed gain
So how can speed be improved? There is no doubt that linear motors are the key to achieving the highest speed ratings while ensuring precision placement. They offer levels of motion performance that cannot be matched by ballscrew technology. In fact, linear drive actuation design seems to have gone full circle – literally – as with the example of one manufacturer using a linear motor in a circular configuration as a the rotary placement head. (Figure 3)This head delivers a pick to pick or place to place cycle time of just 55ms without shaking the component off any of its thirty spindles.

Most of the world’s installed placement equipment base is established on earlier ballscrew technology that presently limits maximum throughput. However, all placement equipment manufacturers accept that linear motors are the way forward and it seems all are now incorporating them into their latest generation or high-end equipment designs.

Maximizing the number of spindles in the rotary placement head is another sure way to improve flat-out speed. Remembering that every time you have to move the placement head, you’re not actually placing parts so the goal is to reduce the number of long motion arcs to improve throughput. The more spindles (nozzles) you have, the more components you can load up during the pick cycle over the feeders and the more you can place on the board before having to return to the feeder bank. The current maximum available on one head is 30 spindles with resulting placement rates firmly in the domain of turret chipshooters.

However the number of spindles is not the only factor that can impact throughput. Mentioned earlier was the range of sizes of component that can be picked and taken to the board by any given placement head. Many placement systems using mini-turret style heads will need to bypass individual spindles once a part exceeds dimensions of 3 to 5mm.sq. The result is more trips to the board based depending on part size or, in some cases, even a requirement for a different heads. In such instances, having more spindles can actually limit speed over a broad mix of parts where the handling range is not extensive enough for the board. Be sure to understand the part range of the head technology a supplier provides.

Head technologies
As we have seen, central to the high speed, fully flexible concept is be the placement head itself. Inline spindles rely on gang-picking, but the loss of this ability where small components are involved can reduce throughput by as much as 25 percent. Rotary heads overcome this drawback and also reduce the potential for mis-picking, combining the adaptability of a turret head to pick small components reliably at high speed with the flexible placement accuracy of an inline head. A multi-nozzle radial head equipped with a single Z-drive mechanism can deliver a cost-effective, fast and accurate solution capable of placing components ranging from the smallest (0201, 01005) up to large, finepitch QFP semiconductor devices.

A decision must also be made regarding rotary head type – if changing the component mix also means changing the radial head to suit, the user must be prepared for a considerable investment of time in recalibration of the head, even if this is performed off the machine prior to it being needed.

In practice, this means that many users will choose not to change the head and will instead run the machine or line in a non-ideal configuration, possibly having to reduce its speed as a result. The consequence is a bottleneck in the line which leads to poor asset utilization. An alternative is to specify a system that allows all heads to handle all component types, controlling the changeover of component mix via software instead of the hardware.

Designing the ideal solution
Having established what can slow down the placement operation, identified how higher speeds can be achieved, and attempted to define the major facets of flexibility, how do we now relax the boundaries between traditional high speed placement and flexible fine pitch to create the ideal equipment solution?

As a basis, the modular platform concept now adopted by most manufacturers and endorsed by major OEMs, ODMs, CEMs and EMS providers is a sound starting point to ensure scalability and future-proofing. Module functionality is typically designed to overlap to assist line balancing; new functionality can be added easily; and new machines added in line as required, boosting productivity. And key among the differentiators in adding speed while maintaining flexibility is placement head technology.

A fixed table system will eliminate one of the reasons for slowing a machine down from its potential maximum. Linear motors will deliver sustainable speed, accuracy and repeatability, and employing dual motors driving each end of the gantry will eliminate possible cantilever effects and with it any associated settle delay or reduction in motor speed and acceleration. (Figure 4) They also require less maintenance, so reduced scheduled downtime is an extra bonus to factory-wide productivity.

In defining what flexibility means, it becomes apparent that an equipment designer’s remit must also be flexible if we are to fully recognize and serve the requirements of the end-user. It’s also clear that in the quest to deliver speed AND flexibility the end-users need, those equipment designers and vendors coming at this challenge from a background of inherent flexibility are ahead of the curve. They and are succeeding in finding decisive ways to add speed – but not simply as throughput at machine level, but as much-more-meaningful productivity at the end of the line and across the factory.

Scott Gerhart is Director, Genesis Platform at Universal Instruments and can be reached at gerhart@uic.com