Reinventing stringers

The first attempts at soldering solar cells automatically were made in the late 1980s and early 1990s, when the first markets in the lower megawatt range came about and reproducible soldering processes were being sought for use in series production. Before then, photovoltaic systems had mainly been a niche application, limited to such extraterrestrial uses as supplying power for satellites, terrestrial uses in stand-alone systems such as solar home systems and radar stations, and minor applications like parking meters and electrical supplies for road warning signs.
The first noteworthy reference projects and market incentive programs, especially in Japan and the U.S., generated a demand for automating the time-consuming, and hence costly, soldering of solar cell connectors, and implementing it in reproducible processes. Firms in the medical and cable assembly fields, such as Spire and Ascor (both of the U.S.) and NPC (Japan), rose to the challenge and developed the first prototype automatic tabbers and stringers.
Tabbing means soldering copper conductive tracks, sheathed with a tin-lead alloy, to electrical bus bars on the front of solar cells. The cell connectors (tabs) are made extra long so that, in a second operation, the tabs of the first cell can be connected to the rear of the next, forming a string. After the first stumbling steps with pure tabbers, stringers soon gained acceptance. Earlier manual handling and soldering stations were now obsolete, while a new generation of stringing machines promised significant cost reductions.
Solar cells were typically connected with two cell busbars in four-inch and five-inch sizes, 270 to 350 microns thick. Among them were solar cells by Astro Power, which showed no damage when dropped on the floor and were still perfectly usable. Insiders liked to call them “concrete cells”. It was much more difficult to process the edge-defined film-fed growth (EFG) cells made by Angewandte Solarenergie – ASE GmbH in Alzenau, Germany (now Schott Solar) and the string ribbon cells made by Evergreen Solar of the U.S. and Sovello AG of Germany – quasi-monocrystalline cells that snapped like dry wood under strong internal crystal tensions. Handling, soldering and spontaneously occurring breakages were the order of the day, making up two to seven percent of rejects from string production.

A market is born

The increasing expansion of PV module production in Germany in the late 1990s prompted further companies to join in developing and constructing stringers, with the joint objectives of investing capital expenditures to create future benefits (“capex”), reducing breakage rates, and building up in-house manufacturing expertise. The company W+S Maschinenbau, from which Somont later emerged, built the first stringers for Solar-Fabrik AG, as did Solarwatt on its own account. Typical cycle times lay between 200 and 320 cells per hour. Cell singulation, plus quality control using optical inspection systems, was integrated into automated systems. Retooling from small cell connector spools to coils more than a kilometer long increased machine uptime in ever more closely interlinked automated production plants.
The introduction of the Renewable Energy Act in Germany boosted the production of solar modules, generating further demand for competent stringer manufacturers. Some unsatisfactory experiences with suppliers then active in the market, in terms of quality, service and uptime, created a situation that we in our project team habitually expressed in metaphorical fashion: “When it comes to the automated soldering of solar cells, we can only choose between plague and cholera. After some years of experience with the plague, ready to give a new supplier a try, you get cholera. In dire need, you then go back to the original supplier and get the plague again, but at least you know what you are getting.” Some things have changed, and today we in production engineering are struggling with this or that “cold,” to stick with the clinical imagery.
Along with a typical plant throughput of 600 to 1,200 cells per hour for each solder track, the leading manufacturers on the market give special attention to reproducible quality, stable uptime and stricter process control using temperature monitoring and/or energy flow measurements.
Almost all suppliers on the market follow the same basic principle. In a homogeneous, evenly controlled preheating phase, the solar cell is heated up with as little stress as possible to just below the melting point of the solder. Additional energy input from the soldering head raises the temperature of cell and cell connector to just above the melting point, before they are allowed to cool down again slowly to ambient temperature after the soldering process. This method prevents steep temperature gradients that can cause microcracks to open up, keeps cycle times short, and generates as homogeneous a temperature field within the cell as possible.
Infrared soldering can be looked on as a basic process here. Soldering temperature is reached by the absorption of infrared light in the cell. The primary benefit of such a soldering system is cost-effective implementation. Usually, only the lamps need be exchanged after some time in operation, keeping operational expenditure (“opex”) low.
The disadvantage of these systems is imprecision and variation in the heating rates in the cells. Darker cells absorb more light and thus heat up faster. Cell connectors mostly act as mirrors, reflecting incident energy. The use of infrared soldering was thus technologically limited, so the industry sought more reliable systems.

Current stringer designs

In their Rapid and Certus stringers, Somont today relies on its proven “Soft Touch Soldering” technology. The required heat is introduced into the busbar via a series of soldering pins. Individual soldering areas can be precisely targeted by a programmable logic controller (PLC) to produce an optimal result from preset temperatures and cycle times. The soldering pins gently press the ribbons onto the cells, where heat is then applied to produce the connection. The ribbon holder then moves into position to support the soldered joint, before the pins are retracted.
Komax, after initially experimenting with Hot Bar, infrared and laser soldering, now concentrates solely on inductive soldering in its Xcell systems. Cell connectors are held on the busbars by ceramic clamps. An inductive field heats up cell and connector to the melting point of the solder, enabling controlled soldering. Cell temperature is continuously monitored during the process. The actual processing field under the cell can be defined very precisely by means of a “walking beam” design. Greater homogeneity in the resulting soldering within a manufactured string can be expected, because this design eliminates conveyor belts, which involve mechanical differences that can cause variability in soldering in other designs.
With the NTS series, NPC focuses on a rather unconventional soldering technique using hot air. The necessary heat is applied to soldering points by a large number of hot-air jets. A clamping matrix in the soldering area holds everything in position until it has cooled down. In coming months, NPC will add further technologies such as inductive soldering to its portfolio, reflecting a tendency of stringer manufacturers to move towards customer-specific soldering technologies.
Teamtechnik goes a step further here by laying clamping frames on the string. The string is first laid out in an interaction between frame, cell connector and solar cell and is soldered cell by cell in a second operation. In addition to infrared soldering, the TT900 and TT1200 especially offer laser soldering. The main benefit is said to be the exact orientation and dosage of the required amount of energy for soldering. On this account, laser soldering has the greatest potential for getting stable results from varying materials and process windows.
Clients who are supplied from a broad portfolio of different cell manufacturers are precisely the ones who know the challenges of finding a suitable product-specific setup for the stringer. The broader expansion of laser technology on the market is unfortunately still held back by high acquisition and operating costs and the poor accessibility necessitated by safety engineering.

System costs

Typical purchase prices for a stringer are around 600 dollars per cell capacity per hour. All the manufacturers listed above have considerable experience on the market, and some have placed 100 to 500 systems since 1996. Omitted from this article IS any comparison of quoted uptimes, breakage rates, tolerances and expected throughput in megawatts per year, because such information is mainly supported by sales arguments and usually can’t be verified in a given case.
Opinions still differ today on the question of uptime. If a module producer would nevertheless like to maintain continuous production, many stringer manufacturers still calculate their uptime optimistically based on the Association of German Engineers (VDI) standard of more than 95 percent. Under VDI guidelines, the loading of a machine with material is not taken into account in the downtime calculation. Small faults during operation have even greater effects. The main offenders are software and operator errors, cell breakages and consequent faults because of delayed cells or defective production tailbacks in the stringer, or in subsequent processes such as the layup or cross-connection points. Various quality checks by electroluminescence methods, string flashers or dark current measurements can certainly be implemented, but they leave open the question of how a string can be unambiguously detected and evaluated as good or otherwise.
Visual checks on solar cells with camera systems can be guaranteed as 100 percent reproducible today. The visual characteristics of a soldered string can also be measured with sufficient precision. But how can flawless soldering be guaranteed without damaging solar cells that are being made ever thinner? Stringer manufacturers offer measurements made by electroluminescence methods on the finished string. In this process, the cell and/or string is subjected to current in a darkened room until it begins to glow. Faults such as cracks or fractures can be revealed by a camera system, as areas not carrying current emit no light.
Unfortunately, completely automatic analysis and interpretation of visible faults is not yet possible. For one thing, grain boundaries in multicrystalline cells cause misinterpretation; in addition, the required test duration can only be covered with the aid of very elaborate and therefore expensive camera technology. Even if fully automatic integration is implemented in the near future, the soldered joint itself can’t be checked inline. Sampling with “pull tests” (taking the connectors off the cell, interpreting the photographic print and measuring the adhesive forces) are still state of the art.
Technologies such as X-raying welding seams have only found limited acceptance in module makers’ quality assurance methods, and so far have been used exclusively offline, such as at SunPower.
Is everything okay so far, then? Is the technology now so mature that we need only carry on thinking about low capex, opex, long uptimes and good service? There are enough suppliers on the market. Besides the stringer manufacturers mentioned above, firms such as 2BG, Ecoprogetti, Mondragon, Gorosabel, Kuka, JVG Thoma, Schmid, IMA and a great many others are on the market, not to mention the great number of Chinese manufacturers who offer stringers at a fraction of the cost in Europe or the U.S.
However, this alone isn’t enough in my judgment. Many far-reaching changes are still needed in the module industry.
The existing pick and place and solder process continues to be time-limited by the actual soldering process. Processing times significantly below three seconds for heating up, soldering and cooling can’t be achieved with known technologies.
As manufacturing units, module production lines are at the same time becoming bigger and bigger, so that stringer batches are increasingly being set up at their front ends. From this point of view, what does a 500-megawatt or one-gigawatt line look like? Is it really worthwhile to devote high expenditures to logistics and human resources to transport hundreds of strings per minute to layups via portal axles or robots?
Future modules will furthermore have to be soldered without lead, implying higher temperatures in the cell. The solar industry still has an exemption from Directive 2002/96/EC on waste electrical and electronic equipment (WEEE), restricting the use of certain hazardous substances in electrical and electronic equipment. It will certainly lose this privilege in the next few years, as quantities of production and waste increase.
Cell manufacturers will keep on saving costs, making ever thinner cells with different pastes and different stages of production, while consuming less and less material. Module makers are not as a rule automatically informed of product changes, so faults often are not spotted until products are being certified or, worse still, being used in the field.
How are product failures in the field handled? To what extent can module makers rely on their cell suppliers if guarantees can’t be upheld and performance degradation has to be covered?
At this point, the jigsaw puzzle is shaken up again, and more thought has to be given to the processing of cells to form strings.

Evolution versus revolution

Can’t we do entirely without a stringer and build systems that eliminate this process step? Day4 Energy has been doing this for years with its far-reaching electrode technology. Manufacturers who make back contact (BC) cells, such as Solland and Photovoltech, are experimenting with “stringerless” production ideas that put cells into contact with each other using soldering pastes, contact foils or low-temperature methods.
Even if we do keep on using a proven stringer technology, why don’t we do so where there are manufacturing capacities of several hundred to 1,000 megawatts-peak, which is where they belong in my opinion – in cell production?
What would such a requirements profile look like?
We’d be considering not 600, 1,200 or 1,400 cells per hour, but systems with ten times as much throughput, meaning up to 10,000 cells per hour.
Qualitative changes in solar cells would thus be evaluated directly in the subsequent stringing process.
Solar cell manufacturers could also be held liable for their products, since fractures, cracks, solder losses and cell damage in the string could be detected in their own production facilities. The topics of cell flexure and cell stress occasioned by thermal effects on copper and silicon would also be taken much more seriously. A measured, proven and already classified string, rather than a tested cell, would be delivered to the module producer.
The “cell-module generator field” as a whole system would be considered not only in units of watts-peak per cell but also in terms of its performance in a module over time.
A development similar to the one at Bosch SE/Ersol could be anticipated. Increasing cell efficiency was pushed ahead there in the days of Ersol, with priority being given to cell output in an air environment. After going into module production, the success of the innovations is now assessed on the finished soldered module, taking account of its embedding in film and glass. These are the real benefits of an integrated production process: the optimal interaction of product features, going beyond the conventional division of tasks for wafer, cell and module makers.
What would such a tenfold stringer look like? To my mind, the following tasks should be resolved:

• Homogeneous cell quality, for example, with narrow color and thickness tolerances;
• Solder-friendly surfaces;
• Stress-free cell production without premature damage due to the formation of microcracks;
• Sequencing and decoupling of string assembly and connection processes;
• Stabilization of the finished string with embedding materials;
• The use of more cost-intensive testing techniques will make sense again and be easier to finance thanks to a high throughput.

A glance at the packaging and printing industry will help us see more clearly here. While simple packages for sweets and chocolates were made at one-second intervals until the early 1980s, production facilities need just a fraction of that processing time today. Sugar, dog food and high-quality sweets can be put into new forms of packaging such as tubular pouches and sealing strips by high-speed processes at far lower cost and with extreme precision.
These processes were driven by ever higher unit numbers and simultaneously falling market prices. On this point, the PV industry is no different from the foodstuffs and packaging industry.
To some extent, the courage of manufacturers in the early days appears to have now deserted them. Short-term economic success often counts for more than innovation for the market of the coming years. As one manufacturer put it, “I’m not going to ruin my own market with too much innovation.” At this point, suppliers seem to have arrived at the “old economy” already.
Markus Steinkotter, Managing Director, Sunnyside upP GmbH