Intelligent SPI feedback moves the process closer to true “lights out” printing.

Printer know-how? People have knowledge and intuition, but machines? You may be skeptical after reading the first 10 words of this column, but bear with me. I’m confident that, by the end, you’ll agree that printers can, indeed, learn.

Know-how, or learned knowledge, comes with experience, and that is precisely what some innovative printing technology is designed to deliver: a “smart” printer that can make informed decisions about required adjustments. Working in tandem with solder paste inspection (SPI) technology, printer software has been engineered to assess SPI data with a high level of process reference and, using that input, auto-correct when required. This certainly isn’t the first time such a system has been attempted, but, in my view, it’s the first time the software and the SPI technology have been robust enough to enable meaningful and yield-enhancing results.

What’s changed to make this technology now viable? First, the demand brought on by miniaturization. In the world of 0.3mm CSPs and 01005s, there is little room for error; paste pretty much has to be spot on the pad, or you are going to have process issues. When pitches were a bit more forgiving, closed loop – and SPI systems, for that matter – were a “nice to have.” But, now, alignment correction is a near necessity for a good process.

The second factor that has made closed-loop printing more attainable is the advance of SPI capability. In the past, technically superior SPI systems were also very costly. Today, sophisticated SPI technology that provides accurate data for paste height, area, volume and alignment is available at a price point acceptable to the market.

The convergence of these two factors now makes closed-loop printing an effective reality. But, of course, the simple task of a printing platform reading SPI data is not beneficial to the process engineer. The value is the know-how and the process expertise that’s built into the printer software. This allows the print platform to make adjustments only if warranted by process trend analysis. What you don’t want – and what older-generation systems may foster – is oscillation. If a machine is constantly overcorrecting – 50µm one way for one print and then 50µm the other way for the next – then the printer is never settled and the process somewhat out of control.

If, however, the printer software can act just like a responsible production engineer – that is to say, not respond to the first misaligned deposit with a knee-jerk reaction – the process will become far more stable, reliable and self-controlled. This is precisely the type of logic and intelligence that is coming online today.
Here’s how it works. The printer takes multiple data points from the SPI feedback, analyzes them and, if required, adjusts the offset (alignment) accordingly. Communication is on-the-fly and seamless so that no additional throughput overhead is introduced. Data are constantly fed to the printer, which is trending the data against programmed process expertise and making corrections when required. Just to be clear, though, the SPI system is still the gatekeeper on the process; the printer isn’t controlling the SPI. What the technology aims to do is ensure that the SPI doesn’t have to flag a board and stop the line. Manufacturers will set their own tolerances and upper limits for each device on the SPI system. The printer offsets, logically, are then set at less than that. So, there is a tighter limit on the printer and a wider limit on the SPI.

As an engineer, I have to admit it is difficult to stand back, be totally hands-off and observe the system do its thing. I’ve witnessed many a production engineer become anxious while watching closed loop in action, and I know they are wondering, “Is it doing anything?” And then, voilà, they see the results. Printers can, indeed, learn and use that knowledge to improve the process. Offset correction is just the beginning; there is a lot more that print platforms can learn, and future assembly operations will be all the more efficient for it.

Clive Ashmore is global applied process engineering manager at DEK International (dek.com); cashmore@dek.com. His column appears bimonthly.