I'm just coming off a bunch of research and writing about direct view LED, so it would be reasonable to think I know my stuff.
But this is technology that's evolving rapidly, and when you get into the weeds, there's still a whole bunch to learn and understand about LED.
Gary Feather is the CTO at the Atlanta-based LED display manufacturer NanoLumens, which has been an innovator for many years in the large format display space. We've gone back and forth through the years, by email, discussing advances. He offered to put his headset on and have a podcast chat about some of the emerging and changing technologies he's seeing.
We go into several things, most notably the rationale and use of displays that have engineered coatings that protect the screens from day to day abuse, whether that's accidental or intended.
Gary has an electrical engineering degree, so acronyms and technical terms roll off his tongue like snarky remarks do with me. The result is a discussion that's maybe a little more technical than normal. But if you are into direct view LED, you'll learn some good stuff over the 30 or so minutes.
So Gary, thanks for joining me. I just recently finished a big report on Direct View LEDs. So I think of myself as something of a Mr. Smarty Pants about this stuff, but you sent me an email while you're rattled through a whole bunch of things that are happening in the space, and then I thought to myself, oh I really don't know much about this industry at all. The more you learn the less, so one of the things you talked about it, and first of all, let me back up here and just explain who you are and what you do with NanoLumens.
Gary Feather: Sure. I'm Gary Feather, the Chief Technology Officer at NanoLumens. I've been with NanoLumens for seven years. I left Sharp corporation in the LCD business to get there, and we really had a great opportunity to see the evolution of LED display from discrete devices, SMT devices, and now the new evolutions we're seeing in the market.
So one of the areas when I was over in China, about three years ago, I saw the first iterations of LED Display modules that had some sort of an epoxy coating on them, which is since being described in ways like an adhesive onboard or glue onboard. You're suggesting, or at least your email was suggesting that we're going to be seeing a much more of a shift to that sort of thing.
Gary Feather: The industry is looking for wider application of LED displays and with that comes durability and reliability requirements. A surface that is coated is going to be dramatically more durable than one that has physical soldered devices. So generally the surfaces become an important aspect for both installations, as well as utilization of the display in active environments.
Now, the idea with coating these things is because they're there, the LED chips are soldered on that they can easily be bumped off and they can be extraordinarily difficult to repair. I've seen lots of LED displays wherein the corners a few of the LED chips have been flaked off and other ones have been scraped off.
So this certainly protects it. The concern that was being raised, at least in the early days of it was the image quality is not as good and there were worries about how the heat got out. Has all that stuff been resolved?
Gary Feather: Like any problem you're trying to solve, you mitigate certain aspects to make them viable.
Let's go way back to the LCD panels space. When LCDs first came out, the reflectivity of the screen was a problem. And so we used what was calLED the triacetate cellulose film on the surface so that it looked more anti-reflective. So surfaces have been an issue we've been addressing in the display industry really since the beginning of the industry.
Now we have a really exciting space to work in. We have a surface that we engineer with the materials we choose, Silicons through all varieties of epoxy materials of which then the processing allows a surface treatment to be customized to that, which optimizes the application of the display. So I would suggest while it's not part of the display, it has the capability to greatly enhance display performance now and dramatically improve over time in the future.
Is there any issue with the coating trapping the heat at all or does it go at the back?
Gary Feather: Certainly the coating is an insulator.
The management of heat and thermal calculations allow different approaches to get the heat out of the devices. Heat is a product of the efficacy of the LED, How many Candelas you get per watt, and then the brightness of the display is your Candelas per meter square.
So depending on how bright you want the display and what the efficacy of the device is and what the physical size of the device is, management of heat then use those three parameters. That's an interesting aspect as we look at smaller LED dye, going from a standard size to the mini-LED to the micro-LED, the challenges of getting the heat properly out of that device to keep the junction temperatures in the range to ensure the reliability of the dye itself under operation.
The other worry I've heard or at least had raised was the whole idea that because these are module tiles that you put on a kitchen or bathroom wall or whatever that if it's coated, you can't just replace an individual dead LED dye, you've got to replace the whole module. Is that genuinely an issue or a bit of a red herring?
Gary Feather: Most anything can be repaired. The question is trying to monetize the value of that. So do you have an LCD or an OLED television?
Gary Feather: And how often have you repaired that?
Let me count: zero.
Gary Feather: Okay, so we know where we want to go, and so the red herring maybe is to live in a world where we repair things as a starting point. So we design it to be repaired. We design it to be disassembLED and we design it to then be worked on. That generally adds dramatic cost to a product. So as the maturation of the systems reach the levels that we know they will hit, a philosophy of repair has to be disregarded and you have to look at the fact that solutions will last for the life of the product and meet the customer's requirements.
I don't want to downplay this as an issue. I just want to say we know the destination and we know we've moved magnitudes on that from the past where people were repairing things daily to monthly to yearly, and now sometimes never repaired at all. And these transitions we're talking about, they are critical to building a sustainable competitive market where LED if you will, Inorganic LED is able to rival any of the other display technologies that are out there.
So when I first started seeing these glue onboard or hardened LED modules and cabinets, I thought, okay, this is the way the industry is going to go, and I've been watching it for three years now and I've seen a number of smaller to midsize Chinese manufacturers come out with products, but I haven't seen any of the major manufacturers come out with anything with the arguable exception of the microLEDish products from the big guys like Sony and Samsung and LG that have some sort of coating on it, but there doesn't seem to be much in the middle, is that going to change?
Gary Feather: So you're saying other than the leaders in the market with regard to a vision for the future, putting coatings on their boards so that they meet these requirements you haven't seen the other smaller companies, which aren't major players doing it?
I think you've answered your question.
What I mean though, is I have seen four super-premium products, like Samsung's The Wall, Sony's Crystal LED, and LG’s Magnit, they have some sort of coating. I've seen from Cedar and CreateLEDs and companies like that, they have coated products, but I don't think I've seen stuff from Absen and Yuna Lumen and Layrd and some of those companies who are pretty big players.
Gary Feather: Well, YunaLumen showed at ISE a year and a half ago or so roughly, the coated boards. Everyone has initiated an effort. They have to decide why they are doing it, what purpose are they adding, what benefit do they add to the display and why is this better than the other solution?
So let's take a few other areas of why. If I want to wipe down a surface and in today's environment, wiping down surfaces may be an important aspect, I have to have a surface that's solid, that allows me to wipe it down. So now you see displays, LED displays for indoor that have IP5X and 6X ratings on the front. That would have been unheard of just a couple of years ago. So inherently we've increased the moisture capabilities of these displays. In addition to the fact that we've allowed you to have a surface that is cleanable, and that may be for dust and dirt, but it also may be for germicidal purposes and others related to our current environment with regard to the pandemic.
So I think you see a lot of emphases too, as to the durability, you can hit it with a hammer. Not hard, but you can hit it with a hammer. You can try to pick off a part, but you can't. But now, more importantly, you're able to wipe down the surface and moisture condensation. Somebody accidentally splashing something on the display isn't going to have a negative effect because there are no open electrical circuits on the face of the display anymore. So these number of forcing functions will drive to the right solution.
Let's talk a little bit about that surface though, so you talked about: It's an engineered surface and on the early CLEDis product or Sony micro-LED, in 2017, if you looked at it off, you could see dimples in the process they use to coat it. That was what they were using at that time. But the idea was you, if you realize you can have a shiny surface or a gloss surface, you can have a matte surface and these have been demonstrated at shows or you could have an engineered surface because within an epoxy material, over Silicon, which is much softer, but with an epoxy material, I can then go back and re-engineer that surface to accomplish a number of things with regard to the viewing and potentially optical effects of that surface to optimize the operation of the device. So I think the coding in general and the terminology used of glue onboard is probably not a good descriptor, but an engineered coated surface has a significant potential to change the way that adds value to the LED display.
Yeah, the whole description of glue on board just seems to cheapen the product in a way.
Gary Feather: I would agree with that, Yes.
It sounds like a hack and I know that's not really the case. So is the whole idea of an engineered coating to be table stakes moving forward, like if you're going to have a large format display you really should have that?
Gary Feather: Only in particular configurations. Outdoor displays, which still use coat. SMT devices and discrete LEDs, because they're adequate for and allow the performance and durability for the environments. There isn't a good reason to coat that display because the characteristics we're talking about achieving aren't necessarily even used in that application.
They may have louvers for coating, to cover the sun. They have their own maintenance approach that they take to those displays. So I think you focus on mostly indoor display applications and those in which are in close proximity to people and/or in atrium areas where you're going to have weather conditions resulting in condensation and others and you want to build a more robust indoor solution for an atrium class area.
NanoLumens did this gorgeous long LED video wall on a walkway at Charlotte North Carolina's International Airport, and that's going back two-three years now, I assume that doesn't have a coating on it, but that would be a good example of something that would benefit from that because of all the people walking along with the roller bags and everything?
Gary Feather: Absolutely and that falls into that category of durability, and you want to be integrated with the display as some people have put films on the surface of the displays to result in that. It's literally a peel and sticks either by the tile or by the display, and that tends not to be, when it's not integrated, not to be a good solution for the durability, reliability that we talked about.
Yeah. I've seen some of that. It didn't look very good. So tell me about Flip chips and SMT.
Gary Feather: As you know, we're probably in what I call the fourth generation for LED exploitation into digital signage and as you highlighted early with the Magnit projects and others with LG, we have commercially moved into a space where LED inorganic devices for displays actually will move into the classic space we see with LCD type solutions in OLED.
The transition though is, we started with all these monochromatic LED almost tubes at one point and in the first generation and all of us saw lots of signs that were either the yellowish colored signs or whitish colored signs that were monochromatic and then moved to the discreet LEDs in a triad position and that's your generation too. And it's a great solution mixing the colors with RGB and then that migrated to a more svelte designed with SMT, sticking them all in a flat package and soldering them down to the board.
In all those cases, you are taking a dye, putting it in a package, testing it and statistically picking out the good from the bad, throwing away the bad ones after it was finished and determining what is then good for the next level of assembly. So you can see we're integrating things a little bit more each step. Now from SMT, we've got a two-step we're going to do here. SMT parts, if I don't have high confidence in pre-testing my flip-chip parts, then I can mount the flip-chip devices into an SMT package and have an RGB LED in an array to make a pixel in a package, and then I can pre-test it.
Now, the reason somebody does that is if there are particular constraints with regard to the Chroma or the Luma, that is the exact wavelength and the exact brightness of the device, and you stick them down there without pretesting, the likelihood you'll get the performance you want is very small. So by pre-testing parts then you know that they're in the band you want, and what's typically calLED binning in our industry, and you can assure that when you put the whole display together, all the individual elements meet the requirements you have for a particular wavelength, a plus or minus so many nanometers and then a particular brightness or elimination from that device. So with that in mind, you can't jump all the way in. Now, if you can pretest in a flip-chip configuration the devices in what's classically now calLED a cartridge, then I'm able to check the devices before I mount them and then put them down.
So here's where we need to realize that magic just occurred in the system. When we talk about a chip much like your home phone going from wired to your cellular phone being wireless, the golden copper wire bonds are going to disappear in the flip-chip. Now, the reason I care about that is because the number one reliability problem I have is associated with the metalization and the wire bonding, so I lose the wire bonding. I lose the epoxy. I lost all the assembly issues that made SMT dye mount wire bonds may be less reliable. And I moved to effectively weld two-terminal devices down at a surface, with no wire bonds and no chance of breakage. So a dramatic shift in that area, it means I might be able to pre-test the part, I put the part down by welding it, put it in an SMT package and I build something that's pretty much going to endure any kind of environment.
So is the Flip chip and SMT, is this what's more commonly broadly known as a chip on board?
Gary Feather: I think as you wrote an excellent piece the terminology has been used differently by everybody and I just cannot claim that you and I have the same terminology, but let's take this slowly.
If the die is pre-packaged effectively into a format where it's bumped and ready to be mounted on a surface, that is flipped chip by definition, forget where it's going, but you're going to flip-chip the part. Now we've been flip chipping semiconductor dyes since the 80s. We bump them in a process, they have little bumps on them. Then we actually flip them upside down rather than wire bonding and reflow the whole device. I ran a facility in Texas instruments that did that very function. So the technology isn't new, but the issue being, you remove wire bonds, which is good. Now, how do I want to do that?
As I said before, put them all in one package, just in one package and test it or put it on the whole board. If I put it on the whole board and let's say a typical board size is something like 150x337 millimeters. So I might break that down into a couple of chunks but I'll have 5,000 pixels on any one board. If I can't pre-test stuff, it would be hard to put down 5,000 at one time. So I flipped a chip a package and I tested them, but if I can pre-test them now I put them directly on the board and these are the options that we have today in the world today that go onto a polyamide board material.
So you're mounting it right onto what you would classically call a PCB or a printed circuit board and that's good actually down to pixel pitches, probably around six-tenths to four-tenths where you can literally flip chip and then COB. So flip-chip, don't put them in an SMT package, flip chip put them right on the surface of the board. And when you do that, you get a C of RGB LEDs, besides C of RGB LEDs that are welded in place, not wire-bonded. Most of the early Sony solutions there, their particular displays were wire-bonded. So by removing that variable now, and you can pre-test the devices.
Now you can put down more than three at a time and get them right. You may be able to put 3000 down and get them right. So the shift from the flip-chip is a methodology. COB is an implementation. You can see OB dye or you can see OB flip-chip. I strongly suggest you see the flip-chip.
Okay. So what does all this mean in terms of manufacturing and for the end-users?
Gary Feather: As we look at the application of the move towards flip-chip and the move towards COB, let's talk about flip chip first. A packaging company that sells LEDs taped in the reel, so you might put 2,500 on a reel will have classically bought a package from a packaging maker, they would have purchased the dye on a wafer from a wafer manufacturer, and then they would have wire bond machines and they would have sealant stations to put an array of these down, put the dye in the package with epoxy, bond the wires out and fill it up with material. So you can see in the supply chain, you have wafer manufacturers, you have package manufacturers, you have packaging companies that put them on a reel and then they send it to a company that does the SMT process. So that would be basically the standard process today. So there's another two step process for this: the company that's selling the taped devices with the LEDs on them wants a better device at a lower cost and by putting a flip chip into the package, as opposed to die, he can increase the reliability, the durability, if he's able to pretest that he can improve his yield and subsequently, he can still sell a package, but it's a flip-chip package as opposed to dye mounted package. So he can win on that. Now, when he does that, what he realizes is he can vertically integrate backward and try to pick up some of what's going on in the dye, cause he needs to know more about that function.
And when he does that, he forms relationships with these LED manual factors and the companies building the flip chip devices. Then what immediately happens the company doing the flip-chip devices realizes why don't they just build the whole solution? Because why are they shipping things off to somebody to put it in this classic package?
So from that perspective, the company building the LED might get absorbed quite a bit because he's been taken over by the guy making the die. So that's one area.
The other area is your SMT company. So while they need to put parts on the back, if the front is COB, they really have no idea how to do any of that, they no longer can take flip-chip devices, put them on a board because it's not an SMT part anymore and build an array of those nd then reflow all those devices as appropriate and then coat those devices. So companies in the supply chain that classically did one side did the other side and then shipped it off to the company are now in a situation where they have to consider, they have to go somewhere else to have that work done. They don't have that tool and equipment.
So the supply chain is evolving and streamlining as well?
Gary Feather: Right and what you'd expect, and let's move all the way down to let's say LCD TVs, the glass is built with the switch, the optical light switch in it as part of making the solution by a company like Sharp or a company like Samsung or others, OLED fits in the same category. So as you can integrate more of these pieces together, certainly the overall output is improved in yield, the costs go down, the automation increases and subsequently it allows you to build a different model for that. So many people that were doing Parts of this job got absorbed into the totally more integrated solution.
In this case, eventually a CPB based solution with inorganic devices placed in a more effective way on a large area board with results that give you the display performance you're looking for, that will result in a large area displays built that a factory, completely not tiLED in the 110, 120, 130 inch range with inorganic LEDs that you buy much like you buy an LCD or an OLED TV today is just bigger than those tech technologies can support and are not tiLED anymore, but completely finished and a finished product at home. And that's what you're seeing with the solutions from the big people, as you highlighted before Samsung and LG.
Does all this make it easier for a company to spin itself up as an LED display manufacturer when they're not really manufacturing, they're just saying they're a manufacturer?
Gary Feather: I list in our internal strategic competitive list 30 key companies that I think are integrated manufacturers. There are about 140 that are out there. So indeed today in LED-based displays, there are many people that are brokering talent from other people to provide solutions to customers and adding very little value to the actual solution. So in one way, it does allow that. On the other way, these companies may decide they want to be vertically integrated all the way to the customers. And so they may not open that up. So we have yet to see how that actually works as they fail. There are factories that are doing many LEDs today and some microLED capabilities, but the new factories built, we're in the 50 to 70,000 square feet, and they allow processing of a lot of materials. So I assume any serious customers will be considered until those factories are full.
Let's talk about costs that have been coming down through the years. And I assume that's a function of buyer volume and manufacturing advances and everything else is this whole kind of shift going to also lower costs?
Gary Feather: Absolutely. The integration of these functions will lower costs. Part of the cost is yield. If the yield increases, that's immediate money into your pocket, but let's look at the whole solution. We talk about how we have the LEDs certainly, and we have drivers for the LEDs that support that and then the controlling system that goes with that. The automation of these systems from this, these are mostly now today, pulse width modulation non-persistent displays that are constant current devices. So that's what we build. So basically we build a light bulb that goes on and off fast enough so you see the mixing of the RGB the way you'd like.
As we look at the drivers, what's happened with the drivers is they used to be inaccurate and imprecise. And they were almost like analog, even though they're digital from device to device based on lead length drive, performance, voltage, and many other noise factors. So today, if you look at the newest release devices from the leaders in the industry, we have now integrated solutions in drivers that are so much more advanced. So what used to be in a single, it would take, say 64 packages are now in a single package associated with performance. So as you would guess, the cost has dropped as the level of integration has gone up much the same way with Silicon devices. So one of the big cost drivers in this is driver technology and driver technology continues to advance at a level that's quite phenomenal is the ability to control the LED and controlling the LED is absolutely the critical part of being able to create incredible images with LED technology. There's nothing like the speed, the performance, the color of an LED, and with the right driver in an amazing world opens up what people can do with these devices to give the user a perception that you otherwise never felt.
Yeah, in doing the recent report I did on LED, I got the sense that the marketplace is shifting from being fixated with pixel pitch, and who's got the finest pixel pitch displays and so on to a more mature market that understands visual quality is everything and you need to have a great control system, great drivers and everything else.
Gary Feather: That's absolutely true. Also, let's go back, this is the need, The need of the display initially was a sign and the sign basically is communicating fairly bold things in very large spaces, but the market shifted starting in 2014 to video displays. So what we're trying to do is replace and or meet the kind of performance with an OLED device.
They realize it's all about what the user perceives and having worked with creatives In the Hollywood structure with regard to images and within the whole physical sciences, the best stories are told in the dark. So the black are so important for you to feel the life-like nature of an image and we are just now addressing contrast ratios that begin to be a priority to realize very little reflected light of the display results in our blacks because we're off, obviously when we're black, we're not like an LCD trying to hold back the light, so more like an OLED. But once the black is attained and reflected light is mitigated and the contrast ratio. it's higher, the image comes alive.
So that's one factor. The other is a bit depth. As you go down for the brightness, the eye becomes so much more capable to see the black areas and you've got to make sure you don't stair step that, that it's a nice blend because that's what reality is. As you look at things and the result is it's the second area bit depth is accurate and precise. This resulted in images that do appear lifelike and there are about four other parameters, but those two I think is what really brings an image to life and allows for a story well told.
This is all pretty technical for a lot of people. If I'm a reseller or particularly if I'm an end-user, how valuable is it to understand and get into the technical weeds on this stuff? Or is this something they don't really need to know?
Gary Feather: I don't think they need to know it at all. In all honesty, not for any reason other than seeing is believing and you want simple messages to people about simple things. As I said, we're going wireless with the way we connect devices, that's a big deal. We're trying to emulate a lead contrast ratio, that's a big deal. We moved to the control set. You get the right color in the right spaces. No fake colors along the way in color matching across the whole CIE space that you've covered. That's a big deal. People can get that right color, reliable, durable, and looks real. That's probably the message.
The details of that are having the marketing spend to give that information to people. So they get it in a way that's valuable to their customers because when you're done, you want to look at it and be wowed with what and that's the only thing the customer sees. All the technology you and I are talking about is effectively what is behind the product to assure that's attainable
From my perspective, I think it's always good, particularly if you're making a six-figure decision to have at least a decent understanding of what's under the hood matters.
Gary Feather: Right and I think we can put together information for people to ask the right questions to basically audit what they're getting to make sure they're not buying last year, the year before last solutions, which will be limiting and they're moving into the solutions that are not limiting going forward.
All right, Gary, that was terrific. I think we should do this again sometime. Very insightful.
Gary Feather: It's an interesting market and you're going to see, I think about a hundred million in flip chips this year at retail and probably as much as 500 million in 2022. So this shift is occurring pretty fast in a $6 to $7 billion industry.
So I think that elements an important takeaway. Building durable products for long life, that's a great takeaway. And I think maybe the most important element is that LED now we'll begin to stand side by side in a larger format for what we've seen in the past with LCD and OLED.
Okay, Gary. Thank you.
Gary Feather: Thank you, Dave. Appreciate it.