How Optical Inspection Protects Advanced PCBs

A crowded server board with ten thousand parts doesn’t forgive sloppy inspection—and neither do pricey GPUs and chiplets. From the floor of Productronica in Munich, we dig into how automated optical inspection keeps advanced packages honest once they hit the PCB line, where solder quality, coplanarity, and sheer component variety can make or break yield. Vidya Vijay from Nordson Test & Inspection joins us to unpack why AOI remains the fastest path to actionable insight, when X‑ray is the smarter choice, and how new sensor design changes the game for reflective, high‑mix assemblies.

We explore the real pain points engineers face today: shiny dies that confuse cameras, BGAs packed with I/O where hidden defects hide under the body, and miniature passives that crowd tight keep‑outs. Vidya explains how three‑phase profilometry creates true 3D height maps by projecting fringe patterns and reading them from multiple angles, enabling precise checks for corner fill, underfill, and coplanarity. We also get into multi‑reflection suppression, Nordson’s approach to filtering glare and ghost images so the system sees the joint, not the noise. With true RGB on side cameras and higher resolution, AOI can now pick out tiny solder balls and subtle surface issues at speed—fuel for stronger AI autoprogramming and more reliable defect classification.

If throughput is king, data is queen. We talk about closing the loop from inspection back to the line to prevent bad lots—flagging stencil drift, placement offsets, and paste issues before they explode into scrap. Then we spotlight Nordson's launched SQ5000 Pro: faster cycle times, a wider field of view, and configurable 7 µm or 10 µm sensors designed for modern PCBA demands. Whether you’re chasing yield on high‑value GPUs or balancing AOI with AXI on dense boards, this conversation offers a practical roadmap for choosing the right tool, tackling reflectivity, and using insight to drive predictable quality.

Nordson Test and Inspection
Delivering best-in-class test, inspection, and metrology solutions for semiconductor applications.

Disclaimer: This post contains affiliate links. If you make a purchase, I may receive a commission at no extra cost to you.

Chapters

• 0:00 Live From Productronica
• 1:01 Advanced Packages Hit The PCB Line
• 3:16 Complexity, Cost, And Solder Quality
• 5:21 The Problem With Shiny Surfaces
• 6:41 Choosing AOI, AXI, And Acoustic
• 8:15 AOI For Speed, Feedback, Repair
• 10:14 Launching The SQ5000 Pro
• 12:07 3D Profilometry And AI Defects
• 13:45 Multi‑Reflection Suppression Explained

Transcript

Francoise von Trapp: 0:22

Hi there! I'm Francoise von Trapp, and this is the 3D Insights Podcast. Hi everyone. This week we are recording live from Productronica in Munich, Germany, and we are talking about the importance of optical inspection in microelectronics manufacturing. Now, this year we've been talking with Nords Intestine Inspection in a series of episodes about the full spectrum of inspection and metrology technologies. We had an episode on acoustic microscopy, one on planar CT, and most recently on X-ray metrology. But today we're speaking with Vidya VJ on the role automated optical inspection plays across the microelectronics manufacturing spectrum and specifically around printed circuit board assembly. Welcome to the podcast, Vidya. Thank you. I'm glad to be here. Now we talked on the podcast a few years ago, I think, in Semicon West. For the front end sensors, for sure. Right. Yeah.

Vidya Vijay: 1:29

We talked about some vapor scent sensors that is also part of our solution.

Francoise von Trapp: 1:33

Right.

Vidya Vijay: 1:34

So how has Product Tronica been for you this year?

Francoise von Trapp: 1:36

Here we are, it's the first day. It's been quite good at Product ronica, very busy year.

Vidya Vijay: 1:42

So any difference this year than other years that you're noticing? I mean, uh it's the good quality of audience the very first day. I think that's a that's a big huge difference for us. So you're not exhibiting over at SEMICON Europa, which is happening concurrently. We don't have our own booth. We are with our distributor. We do have our products that are in the front end, WaferSense products uh being displayed, as well as our acoustic systems uh at the booth in Semicon Europa with our distributors. Okay.

Francoise von Trapp: 2:10

So here today now, can you tell me a little bit about your role at Nordson and then maybe um we can talk a little bit more about what Nordson does in the PCBA in general?

Vidya Vijay: 2:21

Yeah, definitely. I'm the director for growth and strategy. Also manage all the products for optical sensors and metrology division, which is our optical systems and wafer sense on the front end side, as well as our acoustic systems. I'm an electrical engineer by education and training and uh slowly moving on to managing products that actually inspect uh from the front-end products to the back end electronics assembly. Okay.

Francoise von Trapp: 2:49

We at 3D Insights generally focus on the advanced packaging, 3D integration side of things. But you know, we don't often talk about what happens after that device is manufacturing and and what it causes downstream when it goes to the printed circuit board assembly, because we're here we are delivering you higher density advanced packages. So, what sort of challenges does this bring to the PCBA environment? Oh, definitely.

Vidya Vijay: 3:17

With the first of all, with the advanced packaging, each component cost and the value of each component, how critically it's placed on the boat, is very, very high. Right. And uh these components now also have multiple IOs, meaning it could be a BGA with multiple bumps. Now, when you have a BGA, you need to make sure every bump is sorted in place and there is no shots between each. All three items the X, Y, the coplinarity, how it seats, all of that is important. Not just one, not just looking at the sorter stencil, but all of these multiple IOs on the board. And you also have components that are shorter and also even the passives that are much, much smaller, 02, 01, 08, 04. I mean, you know, things are getting a lot smaller in terms of passives with each of the I.O. And the number of components now you can have on a server board more than 4,000 to 10,000 components. And you have these um uh bigger uh GPUs, CPUs, a lot of memories, a different type of connectors, multiple variety of components very closely packaged.

Francoise von Trapp: 4:26

Right, right. So you're getting a lot more complexity in the packages themselves than used to be. Yes.

Vidya Vijay: 4:32

What is driving that? For advanced packaging, right? We have system on chip type solution, we have uh co-wast type packages that now it's packaged with multiple IOs, right? Multiple IOs. These types of GPUs, when they go on the board, now these are extremely expensive GPUs, still need a little bit of outside I.O. connectivity and so on. That is definitely creating a very expensive uh end product. Then the inspection, the soldering, the quality of solder, the placement of the component, all of that becomes extremely critical. And it used to be critical with a few micron offsets now that these are coming down, coming down slower, uh smaller and smaller.

Francoise von Trapp: 5:17

What other challenges are you finding with today's advanced packages?

Vidya Vijay: 5:21

So with the advanced packages, the other thing we are seeing is a lot of shiny components. Okay. So this is still like a camera. When you see something shiny, you know, you're reflecting and it is becoming part of your optical chain as well. So we are seeing a lot of requests for shiny components, shiny dies. That that's becoming a bigger challenge, actually. What's making them shiny? The substrate, the die itself. Okay, and we are also doing a lot of system on chip type inspection. Okay. So now that has exposures to a lot of shiny components. Okay. And a lot of the RF components are also shiny. Okay. And what about like chiplets? Chiplet packages. Yeah, chiplet packages can also go on to different types of boards. All types of components at the end, you know, can go on its own substrate or can go on an FR4 type assembly. And uh those can be inspected as well. And sometimes if it is just uh completely surface mode, no leaks, uh there we need to make sure there is corner fill and underfill properly to these components so they are making good contacts. Okay.

Francoise von Trapp: 6:28

So are there any challenges that you're faced with concerning different substrate materials like glass core substrates as opposed to the traditional laminates or even silicon?

Vidya Vijay: 6:40

So mostly on the PCBA side, we are still seeing a lot of FR4, but also a lot of other substrates within the IGBT can be a ceramic type uh substrate for those applications.

Francoise von Trapp: 6:54

Okay, so what I'm thinking about is the packaging, the packages that are coming in that have been built on a glass core substrate or a glass carrier or a glass wafer.

Vidya Vijay: 7:03

Yeah, it still goes on an FRPCBA type substrate, and we are okay with that.

Francoise von Trapp: 7:07

So finding any of the defects in the package itself that has the glass is not any different than any other kind of package?

Vidya Vijay: 7:15

Because those type of uh substrates, right? Yeah, that is not something we can directly inspect on optical. It gets packaged onto a lead frame or a something and then goes on the PCB. Okay. So that we can inspect. Okay. Or anybody. It's true for anything optical. Right. So all of that is like you're looking at wafer level, then it gets diced and stuff, right? And then that goes on to some type of molding or something, and it goes on like a PCBA. Right. So even like uh system on chip, we have the advanced packaging on an iPhone, right? So it's not completely on a PCBA, but it's all made on a different FR4 substrate. There is no more glass on that level, right? Because you can solder on to different components.

Francoise von Trapp: 7:54

Okay. In terms of inspection for these types of packages in the PCBA process, what are you finding is the best method of inspection for that?

Vidya Vijay: 8:04

So there are multiple methods of inspection, but optical is the easiest and fastest, right? Anything you can see with the system, you can automate that. And now we have the AI algorithm and stuff. So to make it faster, the cycle time is also important. Optical is way to go when you are able to see the defect, even a smaller defect. If the machine can see it, we definitely can automate it and it can be an AOI type inspection. And when you have voids or you know, you're looking at multiple layers, then we have to go to either acoustic or X-ray type uh inspection.

Francoise von Trapp: 8:40

Okay, so we've talked about this at the wafer level packaging level. Are you using all of these different types of inspection even into this PCBA? Is it just that the aspect ratios, the sizes change, get a little bit bigger?

Vidya Vijay: 8:53

In the PCBA market, it's generally usually AOI and X-ray. X-ray, okay. And you know, there is very few acoustic, but the FR4, it's not very good for acoustic wavelength to penetrate through. I see. So it's mainly uh AOI and AXI.

Francoise von Trapp: 9:08

Okay. Now let's talk specifically about AOI. You talked a little bit about the benefits because you of the the speed and finding things quickly. When else would you choose AOI as your method of inspection?

Vidya Vijay: 9:21

Anything you can see, cycle time, and also about uh a lot of automated defect classification and so on.

Francoise von Trapp: 9:29

And is this for sorting and taking the device out of the production line, or is it an option opportunity for repair and replace?

Vidya Vijay: 9:39

Yeah, it's option to repair, not just sorting good and bad. Okay, but it's a complete cycle. And uh you can feed the information back to, say, for example, at the stencil level, if something is off and you can do some predictive maintenance and catch a bad lot from being made. You know, being fast and being in line in the system can give us that benefit. And this is all non-destructive as well. It is everything is non-destructive. That is true with our acoustic, uh, optical as well. Okay.

Francoise von Trapp: 10:10

So let's talk about a new product that you introduced to the market. Can you tell me about that?

Vidya Vijay: 10:15

What we've been going towards is like we have our very high tiered approach for our AOI product portfolio. With being our five micron solution, uh 5 micron resolution with our SQ 7000 Plus that we've launched last year. And now we are launching our mid-tier SQ 5000 Pro that's being launched in Product Chronica. And uh this system now has a sensor with four cameras and a projector. It's basically a three-phase profilometry sensor. What is new here is we have true RGB on the side cameras. We can make auto-defect analysis a lot better. We can see as small as solder balls on a small solder balls on a solder pad. Very high-resolution system here, completely automated with uh both 7 micron and a 10 micron uh resolution, two different types of sensors for the same system and a much larger field of view. So we can go, it's almost uh two times the field of view. And when compared to our Lexi product, it is also 30% faster. Okay, and so how about the throughput? Uh so the cycle time, because it is 30% faster, based on the boat size, it's gonna give you your cycle time. Based on the boat size and the number of components you have, is gonna uh help with the throughput. Quite high throughput as well.

Francoise von Trapp: 11:34

So a little more specifically, can you explain what challenges the SQ5000 Pro targets and what the solutions are?

Vidya Vijay: 11:43

Yeah, SQ5000 Pro is our newest product that we are launching, and it has a sensor. This is our first product family with uh true RGB on the side uh for the site cameras. What that's helping us is provide higher uh quality uh images, and that's going to help us with a lot of AI defect analysis. So, right now we do the AI autoprogramming. A lot of AI autoprogramming, AI features are included in all of our AI systems. Having a better high-resolution image definitely helps with defect classification as well.

Francoise von Trapp: 12:18

So explain to me a little bit. You're doing optical inspection, and you so you're you're basically using cameras, and then you have AI algorithms within that that can identify what the defects are. Is that what you're saying? Yes.

Vidya Vijay: 12:32

Okay. So the type of sensor that we use in our most of our uh SQ systems right now are three-phase profilometry sensors. You project a fringe pattern from top. So based on the type of target and the height structure, uh it's going to reflect the light uh differently. And these these lights are captured by the four cameras on the side, and then we use these images now to figure out uh what the 3D height map looks like. Okay. Okay, so that is how general three-phase profilometry in a very quick version works. Okay. And uh with these images now, we are able to, because we have, you know, we are looking at the corners, we can do a lot of corner fill inspection we talked about. We also have our proprietary uh MRS, um, multi-reflection suppression algorithm. Uh that helps with, you know, say, for example, something is shiny, we are able to filter out those noise with our algorithms. So we do have uh we call our sensors MRS sensors, okay, very specifically for that.

Francoise von Trapp: 13:34

Okay. Okay, so um you're demonstrating here with this tool? Yes, we do. Is it all um available on the market yet or is it still in qualification?

Vidya Vijay: 13:45

It's fully launched and it's available for sale in the market now. And so where can people go to learn more of this? Uh please go to Lots and Test and Inspection. Our website has a lot of videos and uh brochures about all our AOI products and actually across test and inspection, all of the product information is available in our website. Okay. Well, I appreciate your time today. Thank you so much, Vidya. Thank you, Francois. It was great talking to you today.

Francoise von Trapp: 14:11

Thank you. Bye. Next week on the 3D Insights Podcast, we'll be celebrating 50 years of Semicon Europa, featuring conversations with Christian Koich of ESMC and Luke Vandenhoe of IMAC about global collaborations for European economic resilience. There's lots more to come, so tune in next time to the 3D Insights Podcast. The 3D Insights Podcast is a production of 3D Insights LLC.