Site icon Creative Electron – The X-ray People

Overcoming Parallax: The Magic Behind the TruView PC2

Firesidechat_03212021

Creativity is the name of the game in this week’s Chat with the Xperts.  Host David Kruidhof and his guest Griffin Lemaster provide a look under the hood regarding the creative process that culminated in the TruView™ PC2.

As Griffin shares, there are any number of parts counters on the market, but the TruView™ PC2 is unique among them for a number of reasons.  It is remarkably compact and as a result portable.  And because it’s portable, the team couldn’t resist adding battery power to its list of features.  He also explains the physics behind how Creative Electron succeeded in making our parts counter so small relative to the 10-foot tall behemoths it competes with.  The simple, yet elegant solution is so creative, in fact, that we were recently awarded a patent for a key design feature of the PC2.

Watch the video to get a look at the magic behind the TruView™ PC2.  Our Xperts also discuss a variety of other uses for this remarkable unite, as well as how its design made it ideally suited as a rental system.  We encourage you to reach out to us directly with any question.  Register for upcoming Fireside Chats with the Xperts and view our archives here.

_______

David Kruidhof:
All right. Time for another fireside chat with the experts. It’s me, David Kruidhof here again with Mr. Griffin Lemaster, our VP of manufacturing, and today I brought Griffin here because we want to talk about our TruView™ PC2 and the evolution of this system. We were recently granted a patent on the technology that we use here so just want to have a discussion about that, dive in, and get the whole story from you, Griffin.

Griffin Lemaster:
Of course, happy to share the experiences.

David Kruidhof:
Yeah, our parts counter has been a unique system on the market. Parts counting using X-rays has been around for a little while, but the cabinet itself is unique and gotten a lot of great comments, compliments from our customers but since you’re one who designed the machine, at least played an integral role in it if not the entire role, wanted to touch base with you, get the history on that parts counter.

Griffin Lemaster:
Yeah.

David Kruidhof:
How long ago did that start? And how did you guys go about doing that?

Griffin Lemaster:
Sure. Well, we first started seeing that parts counting using X-ray technology come out was at probably 2015, maybe, around there and we looked at it, explored the market potential, we’d liked it. We liked what we saw. So, one day, probably two weeks before a trade show, we said, “Hey, let’s get a parts counter at a trade show, the next trade show coming up.”

Griffin Lemaster:
And we definitely did, we made it happen and I will tell you that, that parts counter we had at that trade show was a piece of $h*t, without a doubt. It was an awful, awful machine that we put together in two weeks just to showcase we had the technology but I mean, the amount we learned from having that machine there was crazy.

Griffin Lemaster:
What we did, we took our current prime, got rid of the bottom, set on top of the cart and basically put a big sensor on the bottom, source on top. We put some elevator in there, raise it up and down. It was a garage door motor that lifted this stage up and down real quick, but that was our first attempt at a parts counter and it was ugly and it was really bad but it said, “Hey, let’s push the idea.” and that’s what it did.

Griffin Lemaster:
It pushed the idea so after that trade show, what we did was we literally that … As soon as we got back from that trade show, that week, right. We cleaned up, packed up, came home and that next day, probably five or six of us sat around just at the break room table and we said, “Hey, how can we make this sell right? What do we need to do to make this unique, make it sexy, make it impressive and make it work?” and so, the whole team we had, the CEO, myself, I think Glen was there, our VP of sales at the time, Brian was there, Carlos.

Griffin Lemaster:
I mean, we had all the minds there trying to put stuff together and, “How can we make the sellable?” and it was one of the most productive conversations I remember us having as a company is, “Hey, what do we need to make this work and when are we going to do?”, right. When can we do it? We were spit balling ideas, throwing ideas out, left and right. And some of the most unique ideas that we thought we could sell to our customers and will be useful, were the mobility of our system.

Griffin Lemaster:
So you put it on casters, right. You can roll it from inventory room to inventory room, we liked that but then we said, “Hey”, at that point we said, “Why does it just have to be mobile?”. We can also … If it’s mobile, shouldn’t you be able to move it anywhere at any time and use it, not just move it somewhere, plug it in?

Griffin Lemaster:
So we decided, hey, let’s put a battery pack in there and we ran it off battery and so it was like, no one else is doing that and we thought, this is really good idea and so we kind of… That was our key insights for that. We wanted to make it smaller too, right.

Griffin Lemaster:
The other ones on the market, they were 10 feet high, had to be bolted to the ground and we know why they were 10 feet high and I’ll get to that point here in a second but they were so heavy and so fixed and we thought we could really improve the customer’s experience by making it mobile and giving them some variability that they could use.

David Kruidhof:
I remember watching the intro video with that machine. I wasn’t at Creative Electron then, but I think you wheel, Bill’s in the front talking about it, and you like wheel it in from the back.

Griffin Lemaster:
Yeah. We staged that pretty well.

David Kruidhof:
But it was really impressive-

Griffin Lemaster:
That was part of the first of the new design.

David Kruidhof:
Outside of the company at the time, just seeing a mobile X-ray system, right and for your parts counting in your inventory. I’ve been in a lot of stock rooms in the electronic manufacturing world and most of them don’t center around one specific spot, there’s rows over and over again. Right. Corners you got to go to, and if you want to just count things right off the shelf, you can do that with that machine. So it was pretty neat idea. I was definitely impressed.

Griffin Lemaster:
That was the concept you’re right and that was definitely Bill’s marketing promos. Hey, let’s start the video and you just walk up to me, it’s already going to be on. Right, and so yeah, three seconds of fame for myself.

David Kruidhof:
You can add an extra 12 seconds.

Griffin Lemaster:
So yeah. So yeah, that was an initial design right and the solid cabinet, we actually had it manufactured by a guy who made kiosks and they made this steel solid cabinet for us. We designed the whole interior work with the battery, as I said, and it was just a push drawer and that drawer, push open and close yourself. I mean, it’s basically a easy way just to load loader real or multiple reels at the same time and so with that, I mean, once we actually had the counter in place, the physical cabinet that we could use then started actually the heavy work is what I call it is the software, the processing, the algorithms that were used to count the components on the reels because initially the idea when we made that awful first parts counter was…

Griffin Lemaster:
Bill’s over a weekend, created a MATLAB script that could count the parts and it did okay for a little bit, but then once we actually started counting lots and lots of different parts, different shapes wanting to get into four reels at the same time, all these other challenges presented themselves and so we start developing our AI before that. We thought this was a great application for it. Let’s put an AI engine to use, let’s develop machine… Using machine learning technology, let’s see what we can do with this application. So we started pulling out we got thousands of reels from various sources, all obsolete, use whatever from customers, collaborators and whatnot and we’ve got a ton of reels in house and we started taking pictures of everything right and started training the algorithm to count components.

Griffin Lemaster:
When the reel was right there in the center of the image, not a problem, it’s actually went pretty well and I mean, we count probably 90, 95% of components we saw when we put the reel in that center, there. The issues we started seeing and what the… Even the algorithm couldn’t get to work well, was when we basically started doing that quad count, right. We put four reels in on the stage, flat on the stage, and take images and the biggest issue that presented itself was parallax and we know what parallax is. It’s something that’s going to exist either way and basically what it was is when you have those components on the outside of your reel, overlapping with each other, as components, as opposed to the ones on the inside, you have a nice clean line.

Griffin Lemaster:
You can see around it and counting. When they started overlapping with each other, that causes problems with the counting and so we racked our heads. How can we count this, and we thought we could just train enough. We thought, what if we just trained enough images with the AI algorithm, it would… A solution will present itself right. Using AI, right? That’s the magic box, it’s the magic key word. A solution will come out if you just train enough and so that involved hundreds of hours of interns looking at images and counting the parts and saying, this is a part here, No, this is not a part there, when realistically we were trying to solve the wrong problem and we recognize this. How we can’t get rid of parallax. It’s always going to exist.

Griffin Lemaster:
There are certain ways to mitigate it, such as some… For example competitors, they might use different technologies as far as… They might use a line scan camera, as opposed to a flat panel, or you’re going to get less resolution and you’re still going to have parallax on the outside of the edges because you’re, still having that cone beam from your x-ray source coming out and so you might not have parallax in the move direction, that’s moving, but on those outside edges, you’ll still see it so it’s a little easier, but less resolution, slower technology. We’d like to… we want to do all of our four reels in one shot and so we were kind of moving along to what else can we do? And again, so now we had another… we’ve already sold a couple of these parts counters at that point.

Griffin Lemaster:
And we’re realizing it’s not performing as well as we want to. So we had to have another one of those design meetings where we all sat down and racked our brains again, how can we figure this out? You know, is this the end of the product? Are we just not going to invest any more time in this? Or do we keep pushing? What can we… What else can we do? And really what it came down to is that… Well we discovered, you dig down to the root cause of the problem and we found the root cause wasn’t the parallax itself, it was the fact that the parallax was uneven, so when you have four reels sitting around flatly on the stage, you’ll have basically more parallax on the outside corners of the four reels, than you will have the inside corners of the individual reels because each of these real sets of individual different components.

Griffin Lemaster:
It takes… each of those quadrants is treated as one image and so you don’t have a even parallax around the reel and so, hey, what do we… if we don’t have an even parallax around a reel, we just have to… let’s move the reel so that they… We can get… So they’re in line with the source so the… basically the direction of the source, the beam of the source, where it’s the center of the reel is normal to it so I think David, do you have that patent available? You could show my… Actually cleaning this up and make it easier to understand, rather than having me just explain everything,

David Kruidhof:
Yeah let me pull it up here. So this is, is our patent on US PTO. So this image is the explanation of parallax. So you can see here.

Griffin Lemaster:
Image on the outside will cause the… Shadow the image appear to grow right, as opposed to the center image.

David Kruidhof:
Center one you get a nice, perfect square, that’s when you get more of a rectangle.

Griffin Lemaster:
Yeah, and you get those… you get multiple images, there are multiple parts there towards the outside, they start overlapping, right. Center they might be separated, might stay separate but once you get into the outsides, those edges are going to start overlapping each other, which is exact… Which is parallax, we know about, it’s a known problem. It’s physics. Not much we can do about it.

David Kruidhof:
Yeah broken the laws of physics yet?

Griffin Lemaster:
We’ve been trying, we’ve been trying real hard. Still kicks our butt here and there.

David Kruidhof:
So here’s the original design, right?

Griffin Lemaster:
Yeah. Yep. So this is the original design. When you had the four reels flat on the stage, essentially what you get is you get those parallax to be greater on the outside edges of the reel than on the inside of the reel and so that was really what the root cause of problem was, is that it was tough. It was becoming extremely difficult to analyze one picture when it was essentially the parallax was skewed and it wasn’t even and so what we did is we tilting… Tilting the reels so that the reel itself is normal to the direction of the x-ray photons at the center,

David Kruidhof:
Right here in the middle of the reel.

Griffin Lemaster:
Yup, exactly so you keep that normal to keep that direction normal to it and that basically evens out the parallax on the four reels and so once we have that parallax, even around the center, then we have essentially four pictures that are centered that we can… It can be treated as if the part was on the middle or was either quadrant, either way now we can actually do an even measurement. Once we made that change, boom. I mean the algorithms started working beautifully and wonderful and that was, I mean ,it’s honestly, it’s simple, right? It’s really simple. Just tilt it so it’s normal to the cone and once we have that, then the algorithm can read that even distribution of the parallax around the reel and pick up the parts and identify the parts much easier. And so once we had that in place and gave that a couple of tests, we said, “Yeah, this is definitely the product we can keep selling.”

Griffin Lemaster:
And so we pushed that and that was the big breakthrough in that parts counter design for us. Now, of course us being us, we were never happy, right? We’re going to keep pushing to make it better and listen to the customer feedbacks. We learned some things and again, we hit the design round again and then we basically redesigned the cabinet for a second… third time and now, I mean, that’s what the PC2 is now so we have a beautiful, sexy cabinet where the form factor is much prettier. It still has the same, basically out, where the source and detector, how we stack the reels still exactly the same but it looks better. And so the reason, I guess I said earlier that, “Why are these other systems 10 feet tall, right, and why are they bolted to the ground?”

Griffin Lemaster:
Well, they are 10 feet tall, too… That’s their answer to the fixing the parallax. The farther you get that source and sensor away from each other, the less paradox you’re going to have and so we wanted to still do that, but keep it mobile and keep it maintained so we have a decent distance in there, but it’s not as far as those 10 feet tall systems and they put that source way up top and it gets really heavy. It’s top heavy and that’s why it gets pulled into the ground and I mean-

David Kruidhof:
The added distance also adds… You need more power, right, near your source?

Griffin Lemaster:
Correct. Yeah. You’ll need a higher power source or something.

David Kruidhof:
Yeah, you need that higher power to reach and still penetrate the samples enough to give you a good image so problem with that then is cost right.

Griffin Lemaster:
It’s all going to add your costs and so those are big, heavy systems. They cost that much and I mean, I’m not going to knock competitors as far as their performance. I think they… A lot of them have pretty good performance a lot of them count pretty well but I don’t think that was really… Everyone’s claiming 99% counting reliability. Right. I’m not going to go say others aren’t, but we definitely are at that 99%. It’s always going to count very accurately, within a 99% of your components and it works great but I think our benefit is really just the uniqueness of the design, the mobility of it, the size and it’s not going to take up a full room.

David Kruidhof:
Yeah. Yeah. That’s one of the advantages of the machine is that size, right. It’s about two feet by two feet upright.

Griffin Lemaster:
Mm-hmm (affirmative).

David Kruidhof:
And it rolls around so you move it out of the way for, for a little bit, you can do that really easily.

Griffin Lemaster:
Yeah and going back to AI for a second, that’s still actively being improved. We still have customers here and they’re saying, “Hey, this reel doesn’t really work too well.” so they’ll come back to us and say,” Okay, we’ll train it on there.” and they’re good to go and so that’s still active and still working really well, as far as I’m… I haven’t heard any negatives in that regard and I think everything’s moving pretty smoothly.

David Kruidhof:
Yeah. Yeah, and that’s also the advantage we have with the full software team we have in house. Right above you.

Griffin Lemaster:
Yes.

David Kruidhof:
There in the office. Having a team that’s actively developing this, I think they just released a new version with improved interface for the tubes and trays as well so there’s… We’re always getting feedback from our customers and developing new features to it. One of the other things that we have on there is attribute inspector so the idea here is you have an x-ray cabinet, you can put whatever you want in there really.

Griffin Lemaster:
Doesn’t have to just be parts counter, right.

David Kruidhof:
Yeah, doesn’t have to just count parts. We have some customers who put a full PCB, assembled PCB in there, and you can get a nice overview picture. You’re not going to get… You’re not going to need… you’re not going to be able to zoom in on a BGA, get real nice image of that, that’s what our other machines are for, but you can check if things are there, right. Does anything get placed down, is anything askew.

Griffin Lemaster:
Yeah we’ve seen it used quite a bit, actually, in the medical devices and putting those devices in there and checking for parts and placement, making sure everything’s there, it’s in the right position, orientation and it’s super easy for them to just open the door, load it and take a picture.

David Kruidhof:
Yeah. Yeah. So very easy to use. The other feedback we’ve gotten a lot over the years, as there is a wide varied, variance in stockroom situations. You know, some have a really nice… they’ve invested a lot to make a really nice storage situation. Some people have a little bit more of a mess-

Griffin Lemaster:
Boxes on shelves, boxes all over the place.

David Kruidhof:
But some people want a live, accurate count to the minute and that’s where this really excels in that as soon as your reels come off the pick and place machine, you can count them and put them back in the stock, you know exactly how many you have, but to do that, you actually need other systems in place. A good CRM, not CRM, get a ERP, some sort of tracking of where it will be, whether it’s a storage tower or just a smart shelving system, but that whole package is an investment and for the customers who aren’t there yet, they’ve asked to, “Hey, can we borrow the machine? I just need you to do an annual inventory. I need to do a quarterly inventory,” whatever it is and these are so easy to use that we just we ship it out to the customer. They plug it in and start counting pretty much out of the box.

Griffin Lemaster:
Yeah, and that wasn’t part of the design intention, right. We didn’t intend this to be a rental machine that we could just send out, but it’s actually worked very well in that regard in size and mobility. We have the… We even designed the crates for it right where they can just roll out of the crate on the ground no need to forklift. They would just rolled out the crate and go.

David Kruidhof:
Yeah, it was to rent, right?

Griffin Lemaster:
Yep. So it’s yeah… It was a happy learning curve there.

David Kruidhof:
Yeah, so we’ve had a number of customers who rented, roll it out of the crate, do their accounting for the week, or maybe just a couple of days, what used to be a week can be done in a couple of days, roll it back in the crate and seal it all up and ship it back. So it’s yeah like you said, it wasn’t really the intention making it a mobile system like that, but a really good, useful side effect so to speak in the design.

Griffin Lemaster:
Yeah. I mean you’re right. It works better as it’s… Works great as a… if you have your ERP system integrated with it, it can update that immediately. Lets you hit that as close to real-time inventory as you can and without having to suffer the loss of downtime because you ran out of a component lead that you didn’t realize you were that low on.

David Kruidhof:
Yeah. Yeah, exactly and we’ve had large facilities purchase two machines and the ROI for them was like three months. It paid for itself so fast for them but again, going back to the smaller facilities that may not have all the systems in place, it doesn’t make sense to buy one, right. Why drop that much money on a machine that… It’s going to save you an hour or two a year because you don’t do your counts too often. That’s where it just doesn’t quite make sense to purchase somethings. We’re happy to have had customers approach us about a rental situation so we’ve been able to build up that program pretty well to this point. Always in development right.

Griffin Lemaster:
All systems are always in development. That’s part of what we do and it’s part of who we are, right. We’re always looking to every system we build. We’re constantly having more ideas for it. They’re going to get implemented on the next go around. Every time we build it or every time a customer comes back with feedback that something we can change there, we will and if it we have to wait for a new design rev, okay, we’ll wait for a new design rev, but that’s constant evolution of our system. It doesn’t stay as it is for more than a year, to be honest, without some nice improvements.

David Kruidhof:
For sure, and the software to, right, both software and hardware.

Griffin Lemaster:
Oh absolutely more so the software, the hardware side. Yeah.

David Kruidhof:
It’s easier to push out software updates to customers than the other and put it in a new hardware. Yeah, it’s always exciting to be here, at Creative Electron with all the innovation and the changes that are happening and to be pulled into those design meetings of, Hey, we got a problem. Let’s just brainstorm. Right. Who’s here, let’s all go gather in the room, five of us.

Griffin Lemaster:
Now we have to do it in this form. We have to do it on zoom.

David Kruidhof:
Yeah. We’ll be back soon.

Griffin Lemaster:
Yeah. I mean that’s, honestly, one of my biggest challenges of… since this pandemic, is not having those quick… Everyone’s next to a machine, we just happened to be there and just have spontaneous conversations about how to improve it. Those aren’t happening as frequently and so we have to schedule those in now and so… Yeah, they’re more productive when you scheduled them but still it just make… You might’ve missed some things.

David Kruidhof:
Yeah. I mean we’ve had countless conversations, just in front of a machine of, hey, the customer said this and it wasn’t a big point, I wouldn’t have made a… Call the meeting to bring it up but if you’re standing there it’s easy to pass on the tidbit of information and next time you’re designing something you’re like, oh yeah, the customer requested this, wonder if I can tinker and add it. Yeah. Definitely looking back to… Looking forward to being back in the office more frequently.

Griffin Lemaster:
Yeah, I think things are starting to open up. More people are starting to come back in. So it’s nice. It’s nice having those bodies around.

David Kruidhof:
Yeah. All right, well thank you Griffin. I appreciate your time. This has been great. Very insightful. So the usual, everyone if you have any questions, feel free to leave them down below, if you’re watching on YouTube or go ahead and email us at info@creativeelectron.com. Thanks.

Griffin Lemaster:
Thanks guys.

Exit mobile version