No Loose Screws: How Automated Screwdriving System Assembly Can Create Consistent Product Quality

August 26, 2020 | 6 min read

Stevan Dobrasevic, Director of Product Marketing, Bright Machines

Screwdriving is an incredibly common skill on many product assembly lines. Traditionally, we think of this process as a human operator using a power screwdriver to fasten a screw. In electronics, that can mean screwing together sub-assemblies, screwing the sub-assembly to the product housing, and screwing the product housing together.

Three Options for Screwdriving System Assembly

And indeed, there are three options when considering how best to complete screwdriving assembly for a product, and the first is the standard, human-led process. While human operators bring the benefit of being fast and flexible when it comes to deployments and changes (a simple explanation of the task at hand and you’re off to the races), this approach to screwdriving assembly lacks scalability. If a line needs to add 50 percent capacity, it’s hard to achieve when that means adding 50 percent more human operators. Relying on human operators also means that a line is inherently not resilient to disruptions – as we’ve seen more acutely than ever during the pandemic, human operators can lead to unpredictable output when folks are unable to come to work. Perhaps most importantly, when considering consistent quality, this human-led approach is also one that doesn’t lead to improvement. Screwdriving is challenging work on hands and wrists, and it’s difficult for a human to stay focused on this task for a long time, which can lead to quality issues. Ultimately, an approach that relies exclusively on human operators can be challenging and inconsistent.

A second approach is what we call “basic automation” and is composed of custom hardware and software developed for a specific purpose, to do an assembly for a particular product. This approach makes sense in high volume scenarios, where the product has a long product life. It can indeed solve the problem of being more resilient to disruptions – robotics can work through disruptions that the workforce might not be able to – but because of its custom nature, it’s neither fast nor flexible. Custom solutions are not particularly scalable either; because they’re designed for a specific line capacity, that capacity is what you’ll always get. Custom lines also typically lack feedback loops that lead to continuous improvement and adjustments. Simply put, you can expect custom automation to do precisely what it was initially designed to do – nothing more, nothing less. 

The third approach is what we call next-generation automation, a software-driven approach, instead of the hardware-driven method, like what’s found in custom automation solutions. Think of this approach as more like Tesla, which creates a software platform and then builds a car around it. Next-generation automation includes a software platform and uses modular hardware to deploy assembly lines quickly. These building blocks of equipment, configured through software, enable fast, flexible lines. Instead of a year to set up, a line can take just 3-6 months from design to deployment. This approach is designed from the onset for scalability, with the ability to add cells or adjust the software to solve bottlenecks and increase throughput. Like any automation, next-generation automation is resilient to disruption, but it goes beyond by offering flexibility and by continuously improving, thanks to software that allows for feedback and adjustments. 

How Next-Generation Automation Transforms Assembly Lines

Specifically, when it comes to a next-generation automation approach in screwdriving, there are considerable advantages in flexibility. For example, one can swap in different screwdrivers to support different torque ranges, and the modular hardware units can accommodate various scenarios of bringing in screws. Multiple types of screws can be supported, different sizes, different lengths. 

For bringing in screws, automation can also do the heavy lifting – and do away with the less common “hand start” in which a human operator would hand start a screw, followed by an automated screwdriver to finish the job (an approach that leads to high cycle time). What’s becoming increasingly popular are pick and place feeding and blow feeding scenarios. With pick and place feeding, a screw presenter and robotic arms work together to insert a screw and begin the screwdriving process. But, in some cases, even that doesn’t have a fast enough cycle time – that’s when manufacturers can use blow feeding, which has a much quicker cycle time, as screws are being taken directly to the head of the screwdriver. 

If we look at how the software comes into play, it can be used to set up different profiles for different phases of the screwdriving approach. Most typical is a three-step tightening strategy: first, a slow driver speed is used as the screw finds the thread; second, the driver speed increases so the screw can quickly drive down until it touches the workpiece; and third, when torque is increased to ensure the screw is fully tightened. These phases are set up and managed through software, which translates to a wholly controlled screwdriving process. Other controlled elements include the angle of insertion and closed-loop inspections that allow manufacturers to set up pass/fail bounds based on quantifiable metrics like the number of revolutions or max torque. 

The software also allows manufacturers to have full traceability into what’s happened, so they have confidence in the results. If they have a product that’s being assembled with four screws, that all need to be screwed to a certain depth at a precise torque, they have the data to know that was achieved on any given unit. This is very different from a human-driven approach since it removes any issues of human variance which can cause unpredictability. 

Coffee Maker Assembly Gets A Jolt Through Automated Screwdriving 

One coffee machine maker sought to improve the consistency of a tricky component of their single-serving coffee machine assembly. The “showerhead” that essentially pushes water over coffee and creates a perfectly brewed cup (and no mess). The showerhead sub-assembly requires correctly aligning sub-components and then fastening them together with screws. With humans, they struggled to get the alignment done correctly, resulting in yields as low as 60% at times, which is just not acceptable for any manufacturer. We worked with them to implement a Bright Machines Microfactory that provides dramatically improved consistency. The three software-defined robotic cells making up their microfactory, replaced five human operators for screwdriving sub-assembly.

In the first station, machine vision guides a robotic arm to pick and place components and align them and screws them together. In the next two stations, additional screwdriving operations are needed to fully assemble the showerhead and use advanced techniques like software-driven control of the angle of insertion and driver speed, to ensure the result is done with high yield. 

The line went from producing 960 units per eight-hour shift with humans screwdriving by hand, to 1440 units per shift when they implemented the microfactory. While more units per shift wasn’t the primary goal (improved quality was), this increased output was undoubtedly an excellent side-benefit for a company working to reach increasing consumer demand. 

Production cost also dropped dramatically: with five human operators, the cost to assemble is $.83 per unit, compared to $.25 per coffee machine with automation, for a one shift scenario. For a two-shift situation, the cost to assemble with human operators stays the same, but with automation, it’s $.13 per coffee machine – and at three shifts, it drops to $.08 per unit. 

Software-defined automation has proven to deliver numerous benefits for screwdriving assembly: it can help improve product quality, create lines that can be deployed fast and can respond to changing demand, improve scalability, produce high yield thanks to closed-loop feedback systems, and ultimately, lower cost per unit. 

To learn more about our capabilities in building the backbone of AI, visit Bright Machines.