Smart Manufacturing
M2M Communication is Here.
Do We Still Need Judgment Calls?
Data-driven processes require IP coordination among vendors – and that means humans. by Chelsey Drysdale
We hear a lot these days about smart manufacturing, but is there a broad consensus on what it means, and more specifically, its application in electronics assembly?

Brian Morrison, vice president of engineering for Vexos, a mid-tier multinational EMS with manufacturing facilities in the US, Canada, China and Vietnam and more than 900 employees worldwide, explains his views on smart manufacturing to PCD&F/CIRCUITS ASSEMBLY in July.

Chelsey Drysdale: Let’s get this out of the way first: How do you define “smart manufacturing?”

Brian Morrison Headshot
Brian Morrison
Brian Morrison: To me, smart manufacturing is essentially a methodology that leverages equipment software and integration protocols that allow continuous feedback to the process. Basically, it’s the ability for us to use the equipment and data to have the owners act, react, execute and adjust non-value-added activities and optimized production. For me, in order for something to be smart, the process must be divided with targets. That means to monitor, report, detect nonconformities and be able to make adjustments based on those data.

CD: If one views smart manufacturing as “automation and computer systems to detect deviations from the norm,” what are assemblers doing in this regard beyond what is available off the shelf?

BM: I think what you hear is a lot of AI. Everyone is coming out of the cloud, and they’re saying “data management” – and there’s more data. I think as we get more intelligent with the equipment we have, the more data are available. You’d be able to use that data to actually make a judgment based on what is actually critical.

With normalization of the data, I know the manufacturers are coming up with CAMX and IPC standards to be able to communicate between equipment and software to make that available. I think the real differentiations between what’s considered a smart manufacturing facility and something else is the ability to use that data and make adjustments to your manufacturing, and I think that’s what a lot of people are doing nowadays: integration; software; decision-making; upstream and downstream feedback.

Mike Buetow: Brian, to follow up on that, that really becomes part of your IP, right? The ability of an individual company to not just collect that data, which everyone’s doing, but then how you process that data and put it into action would really become your IP.

BM: That’s absolutely correct. With equipment and software becoming smarter, it’s becoming a lot easier now. I think the more challenging part is that manufacturers nowadays have a mix of older and newer technology, so being able to use all those different inputs to basically go in – you’re absolutely right – requires IP coordination with the manufacturers and software providers to do that. I’ve been in a number of calls where I’ve had competitors on the line to basically work together to create that IP and make a manufacturing solution that makes us competitive, and the results are outstanding to the point where we have equipment talking to each other, knowing what’s going on and able to react whenever something’s needed.

MB: Does that conversation tend to get initiated by the assembler, or does it tend to come from the equipment manufacturer or the software supplier that recognizes a similar problem across multiple or maybe several of their customers?

BM: I’ve never seen where an equipment manufacturer reached out to another one. Although it is rare, some manufacturers are moving toward adopting a standardized format to facilitate communication, but typically [these requests] come from the manufacturer: “I have a need. I need equipment A to talk to equipment B. They don’t have an interface available. This one has output D that doesn’t work with C. How do we make those things work?” I think a lot of it is collaboration, from the assembler bringing the parties together, coming to a mutual agreement, and a common place. What will happen is with the IP we generate, the manufacturer and software vendor will see the value in it because other assemblers had asked the same thing, and we’ll work together to provide that level of smart integration, and they’ll actually sell it to the next customer, and it just becomes a great working relationship.

CD: How do you balance the cost of implementation of additional software or machines with the cost of operators and the annual volume of product being built? For smaller run production, is smart manufacturing even feasible?

BM: It all starts with a value-stream mapping. You have to take a look at your current process: How are you doing it? Where are the areas of improvement and productivity as they relate to quality, cost, waste, etc.? Those are lean elements that drive the opportunities.

From there, what you need to do is take a look and say, “Okay, what are the solutions?” Usually that is software or equipment, which could be capital; it could be someone’s time, or it could be operators. How do I eliminate high-cost operators in a high-cost region? Do I put robots in place to do that?

So, I think from a cost of implementation, it’s a cost versus benefit, and what we typically do is we look at decisions like what’s the cost? But in addition to that, what is the risk? What’s the timeline? What are the resources, and what’s our benefit?

Do you want on the smaller run versus larger? Whenever someone thinks automation, they think of automotive: single SKU, millions and millions of these things optimized all day. But actually, we are in a small manufacturer’s world, high-mix, low-volume in some cases. We do a lot of prototypes. If you don’t do automation and smart manufacturing, you’re not competitive. You need to make sure you are leveraging all your changes, looking at your changeover [times], and being able to collect all the data where you may not run a product for six months to be able to find out what you did here and what the next one was six months ago to determine [whether] you [made] an improvement or not to make a decision for the next corrective action to go from there. Having that data and being able to look historically to make decisions in the future are really important. That could be low-cost. It could be simply software. It could be MES. It could be data collection. It could be connecting to equipment, so it doesn’t have to be costly to get the benefits.

‘I’m very encouraged
with what
equipment manufacturers
and
software manufacturers
are doing
to bridge the gap.’
MB: We’ve heard a lot over the past few years of the so-called digital twin, which is basically a virtual model designed to accurately reflect the physical version of that same object. How much of smart manufacturing is tied to the use of the digital twin?

BM: One of the things that we embrace from our perspective is what we call the essence of DFX, or what we used to call virtual prototyping or using models to make decisions, the ability to transform virtual data into a physical model that we can actually perform analysis on.

We usually look at it for assembly, test, fit; we use ECAD and MCAD to take a look at those models and find the opportunities as they relate to design rule checks and then make decisions on new product introductions. A key factor to using this digital twin is what helps us make good decisions [for our customers] of changes they should make before we even order a single part or release a single PCB or place a part on the board. It’s really a differentiation. It costs no money, and it allows us to integrate at the time when they can make changes.

‘Leveraging CFX-QPL to Integrate
Equipment and Create a Smart Factory’
Leveraging IPC-CFX, companies can use AI-powered technology to help manufacturers realize a smart factory. These tools collect factory data on defects, optimization, traceability, and more to improve metrics, increase quality, and lower costs. Yet, successful CFX implementation on the shopfloor requires confidence that equipment has been qualified to IPC-CFX using the QPL certification platform.

At PCB West in October, Ivan Aduna of Koh Young will explore how the inspection equipment OEM successfully applies real-time data to improve the production process by converting data into process knowledge using CFX and other software tools. Combined with IPC communication standards, the gates to a smart factory are open to anyone.

Aduna’s presentation is part of “Free Wednesday,” a series of nine free technical presentations on Oct. 5 at the Santa Clara (CA) Convention Center. See pcbwest.com for details.

Futuristic manufacturing line render
Smart manufacturing relies on data-driven machine-to-machine bidirectional feedback.
CD: It sounds like smart manufacturing and Lean manufacturing really intersect.

BM: I think the elements of Lean are what drive us to smart. I think a lot of people look to smart because the elements of waste within Lean drive that. You look at your process. You look at where your elements of waste are. You apply elements of smart to address that as an optimal solution, either through equipment or software. In my mind they’re analogous in terms of one leads to the other.

CD: How do we use the tools available today to reduce defects that are inadvertently designed-in? For instance, tombstoning can be the result of surface tension imbalance due to unequal lands. Or perhaps it can be caused by mounting passive parts over a via, whereby the pad with the via heats faster due to the lower thermal mass. Are these issues best addressed in the DfM rules? Or does smart manufacturing have a role to play?

BM: One of the things we do is identify opportunities and risks in manufacturing. We talked a little bit about the digital twin, so being able to predict the risks of the product, identifying where potential problems may occur. DfM is an element, but also there’s a supply chain risk: What are discontinued, end-of-life, not recommended? There are elements of the land pattern, as you’ve alluded to – vias and pads clearance, as well as the test access and other strategies open to improvement. By going through that risk and determining where we are and running it through manufacturing using the smart information to determine whether our predicted units will have a problem, and then validating that, making corrective actions, and integrating that, is part of our continuous improvement.

MB: If you think of smart manufacturing as something of a spectrum, where we are somewhere between the embryonic stage and fully mature, where is the electronics assembly industry on that spectrum right now?

BM: I’m pretty optimistic about where we’re going. I’m very encouraged with what equipment manufacturers and software manufacturers are doing to bridge the gap, which wasn’t there before. Equipment manufacturers didn’t want to talk to one another. I think nowadays they realize talking to one another is the essence of us being competitive and moving the industry forward. I would say we are closer to sustainability, almost mature, on the upper echelon, at least from where we were from about 10 to 20 years ago. I think we’re almost there. It’s going to take a while because people are still a little bit hesitant to jump on board, but I think we’re almost there.

Chelsey Drysdale is chief content officer of PCEA (pcea.net); chelsey@pcea.net.