System-In-Package Platforms

Welcome to macro fabs engineering podcast. We are your guests, Jean France and

Eric Welsh.

And we are your hosts,

Steven Craig and Parker Coleman.

This is episode 143. Gene is one of the founders and the visionary behind Octavio systems. He currently serves as Chief Technology Officer. Previously, Gene was the principal technology Fellow at Texas Instruments where he built a career finding new opportunities and building new businesses to leverage T eyes DSP technology. Eric is the applications and systems manager for octava systems. With over 16 years of industry experience designing hardware and software systems, including 11 years at Texas Instruments. Eric has supported hundreds of developers bringing embedded systems quickly to market.

And for more background on October systems, check out the Mac fab engineering podcast episode number 17. System and package platforms with Greg Sheridan. And that was almost two years ago.

And that was one of our that was one of our first guests

on the podcast. Yeah, that was guesstimate two, Time sure flies. It's crazy. So a year and a half later, has Octavio systems mission change? Or how's it evolved? Or, I guess, actually and also reiterate, what does Octavia systems do?

What we do is builds System in Package devices to solve a significant problem in the industry. And that is how do I get at this point, the microprocessor, the DDR and the power management all to work together in a small area on in my system. So that's generally what we're after.

Right? And if I remember correctly, from a year and a half ago, the one of the main solutions that you were or problems that you were solving, was the fact that having all of that on the same silicon die isn't necessarily the best solution. Because there's, there's specific reasons why this subsection would be preferable on this type of silicone versus that.

Yeah, yeah. Let me give you a little bit of background because I find it interesting that in the semiconductor world for the last five decades, the value creator in the industry was on the silicon, the wafer fab the ability to integrate more and more transistors on an integrated circuit. We were driven by Moore's law, and I think you're beginning to hear people reluctantly, once again, say, Are we near the end of Moore's law? What I've said to the industry is, we are still on Moore's vision, which was to double the number of transistors you can integrate, I believe what we're finding is that we are moving from innovation occurring on the front end of the process to the back end of the process. And so what we're finding is the best way to integrate is with the assembly and test part of the of the semiconductor manufacturing. And so we will continue to integrate using that method. And that kind of makes it interesting for us, because we're kind of at that forefront of how do we integrate on the system's level using still integrated circuit process technology. And that means the SOC is no longer the solution. But the SOC is part of the solution all in a package.

Yeah, and it's really interesting, too, because you see announcements from major companies like global Global Foundries not going to do the seven nanometer node. And then there's, there's still some question about whether Intel is going to do their 10 nanometer node. And so you definitely can see that there is a consciousness within the industry that it might not actually be worth it to move down to the next process technology node. And so you do see this move to how do we continue the innovation? How do we continue the integration, and part of what we believe is that through system and package technology, and through the use of packaging, you're able to now still realize those integration gains. But now it's at the package level and not necessarily at the silicon wafer level.

And what's interesting is over that five decade period, the back end of the process, and if you're not a semiconductor guy, you you have trouble understanding front end and back end, but we split the world of manufacturing into the front end or the wafer manufacturer and the back end the assembly and test what we You've seen in the last 50 years is the back end the assembly and test their goal in life, their job in life was to not screw it up. So how do I get the yields up and the cost of manufacture down? And that's all I do. And so very little innovation and, and I say that and I have a lot of friends in the packaging, were going to look at me and say I'm, I've been done a disservice to them, but very little innovation has occurred in the world of assembly and test to handle what can happen in the future by doing system and package.

So by assembly and test, you mean basically taking the dye from the silicon wafer, and putting it into the packaging, gold and gold, bonding it to wherever it's going. Correct.

That's right, and then put it putting plastic around it, and testing it. Okay.

Yeah, and it's not just, it's not just wire bonding it down. It's also I mean, they're they're doing, basically, they can, they can put little bumps on a on a wafer, so basically little solder balls on the backside of a wafer. And so you can actually do things like flip chip technology, where it basically the dye acts as a mini BGA, they actually get soldered down to the substrate. And so you have a lot of technologies like that, that also allow you to get a lot higher densities within the package, and a lot. And so that's, that's where you're seeing a lot of the package today is taking these really high dense connectors and connections from the the wafer, or the actual silicon die, and spreading it out such that you can actually use it on a printed circuit board, I mean, you're not going to be able to deal with the spacings and the tray sizes and stuff like that, on a standard printed circuit board, it'll just cost too much. And so the with with most package technology, today, you're seeing, it's mostly just kind of a fan out type of technology. But really as as packaging progresses, you're going to see a lot more innovation, being able to go through and put a lot more silicone actually in a single package.

But to give you an example of the ball, the ball spacing that you can put on an integrated circuit is somewhere below 200 microns bold, the ball spacing, and that's kind of hard to use in the world of systems. But in the world of system and package that's very capable and very Oh with a very much within the capability of the technology.

I'm just looking at Abby, point two millimeter. That's correct. Our listeners that are not in the United States. Sorry.

Correct me if I'm wrong, but but you sort of get to distribute the cost a little bit better in that format, because the the the substrate of a small system on on system on chip, or system onboard, you if that requires, I don't know, 1620 layers or something, you're paying for that in a small area as opposed to across your entire system. So you end up distributing the cost a little bit in a nicer way?

Well, one of the aha moments we've had recently is, there is a possibility with some interesting SIP in our system in package technology to take a processor device that maybe is in a 15 millimeter square packet, you know, 15 millimeters on the side package, and put in the SIP that device, the power management that goes with it, and the memory that goes with it in that same size package, or maybe even a little bit smaller,

for sure. And is that possible? Because basically, you're taking your dye, which has a 200 micron, or I actually looked up as well as it's eight mils. If you're you don't think in micron, because, you know, I don't either. So did you go on from like a 200 micron pitch to a bigger pitch would like, you know, point four millimeters like on a QFN. And so you have all this empty space inside that package? Is that how you how you fit in those parts? Okay, yes.

We also think 3d. Yeah, you

stack on top?

Yeah, I mean, if you think about most packages, I mean, you can go a lot thicker than than what you typically see today. I mean, you can have packages that are one and a half, two millimeters thick. And that way, you can actually stick a lot more components within that particular package by kind of doing that kind of stacking type technology. And we've been looking at a couple of different things that you can do in order to allow you to do some of those kinds of stacking things and that way you utilize that that third dimension, a lot more for how you're going to actually pull all these components in and put them Put them together.

Right? And if you if you have a larger, I guess, z direction to play with, correct me if I'm wrong, but that sounds like it would affords you the ability to have larger value components at the same time, right?

Yeah. But the one problem, there are a couple of problems that exist by going 3d One is this thing called heat. Somehow, you still have to dissipate all that heat. And when you stack things up it, it becomes a problem. So what we've done is one of the partners in the company are one of the founders, it was a top level packaging person at Texas Instruments before he retired, and we stole him. And so what we decided multiple years ago is System in Package is a packaging problem, not a integrated circuit problem. And so he's right at the forefront making things happen.

Eventually, you just have integrated heat sinks to right. Yeah.

Oh, yeah.

We know how to do that.

I mean, yeah, you got it, you got to get to get the heat out somehow.

Right? I'm just actually imagining how you would reflow such a part, if it had an integrated heatsink on it.

If I were at my desk, I could show you one.

So what is our table systems method of packaging,

standard semiconductor packaging, we run it down the same production line, every other integrated circuit goes down. We're our goal is never heard abate the integrated circuit assembly process, but to do leave it alone and let it go forward. And we're just going to design around it.

But we are we are doing a lot of r&d with with partners in order to try to see what what are the changes that might need to go into some of these lines. So while we want to make sure that everything's standard, we still are working with with partners to help push some of this technology. So as we start looking at more of this three dimensional technology, that way we can we can make sure that that is a standard part of the packaging process, and came to run on the same standard line.

Gotcha. Sure, sure. And I remember from when Greg came on the podcast a year and a half ago, or whatever, I believe he showed us some prototypes that they basically look like a board that was encapsulated in a clear epoxy. And so it sounds like things have kind of evolved from from them.

Yeah, and so that that's actually, we've actually got a couple of different products out now. So I mean, we I believe when Greg came on, we had just released kind of our first product to market. Since then, we've released one more last year. It's basically a 21 millimeter part that that has processor, power management, IC, ldeo, DDR, and all the integrative passives and stuff that are necessary to make that a system. And then we actually we just announced that we have a new product called the OSD 335 X CSIP. And so that's a complete system in package, we actually beyond the processor, DDR, P MC and everything like that, we actually integrate the nonvolatile storage as well. And so basically it is a computer in a package. And so really, all you'd need to do is hook up power to it and then hook up whatever peripherals you might need. And then it actually is a one gigahertz Linux computer. And it's it's, it's pretty cool.

Effectively, you can hook it up that bug and it'll work. Oh, that's cool.

Yeah, yeah. So that that's, that's one of the as we're so we've announced it, or we haven't we haven't gotten everything shipped yet. But one of the one of the things we want to do is, uh, is in terms of videos, is to actually go through and get it a dead bug. one gigahertz Linux computer working.

That sounds awesome. You know, one of the things I'm curious about is the fact that you guys are taking, I guess, I guess, various versions, various IP and compiling it together into one system. When you provide a datasheet, or information on this, I'm sure that you have tested it all as a system correct. And and so when someone wants to see the thermal performance, they're not seeing the thermal performance of say, one item on the system on chip, they're seeing the thermal performance of the entire thing. Correct?

Correct. Correct. And so it's, it's actually a very, very interesting thing. And we've been talking with a lot of a lot of customers about thermal because it's such a big care about due to the fact that we have a form factor that allows you to get into Just small, has a such a small size, it enables a lot of new applications which are small in size, and therefore you have all these concerns about heat. But what we kind of see is that that the package provides a little bit of a thermal average. So you'll get kind of some some hotter spots, like where you where you'd normally think like over the processor dye over the P MC dye and things like that. But the the hots aren't quite as hot and the colds aren't quite as cold because the plastic X is a little bit of a thermal average or, and so it's, I mean, it's not much, it's kind of like a one C degree difference. But it the package does help in terms of helping mitigate a little bit of the peak of the heat. Also, when when we look at doing designing for thermal, we make sure that that all the components are rated for the the case temperatures that that we're we're we spec out as part of our device. And so you have to make sure that all your passives meet all the temperature specs, all the DDR is and all the different processor elements and everything. And so that way, you can guarantee a case temperature, not necessarily some of the other I mean, a lot of a lot of silicon die uses theta j and some other things in order to denote heat, heat dissipation and thermal performance. And so really, we look at it more from a case temperature because in many respects, it acts more like a printed circuit board than then a particular silicon die. And so we look at trying to think spec things for more more of this case perspective. And is

the case fully encapsulated? Or is it hollow in some way?

No, no, no, it's fully encapsulated? Yeah, no, it's actually it's very interesting, because basically, there's this plastic mold compound that gets heated up, and they have forms that it gets injected into. And it flows around all the different components and basically kind of locks them in place. And so it's, it's, it's fully, I mean, the plastic mold compound has flowed everywhere, under, if you use a package device, it will actually flow underneath the package device between the balls, that it, it could be seated on the substrate. And so you actually, there's actually a bunch of interesting physics problems about how mold compound flows and things like that. So you get to see that it is it is all locked in place. Because it's basically, everything's encapsulated by plastic.

So the one of the interesting side bars on that is once you lock all those passes in place, bending, it doesn't pop them off. And so they're they're stuck there. And as I've told Eric, in the past, even if the passive kind of cracks away from the substrate, you just added capacitance.

Got a couple of weeks ago about flexing terminal, on on ceramic capacitors. So I guess this would kind of mitigate that, since you're kind of locking everything into place.

Yeah, they're stuck there.

Have you had any issues with high speed signaling and the plastic acting as dielectric in any way?

No, I mean, it really it's, I mean, every I mean, if you think about the signals coming off off the processor, I mean, it's, it's the the plastic has, has a I mean, the high speed signals are far enough apart and everything like that, that we have, we haven't actually seen any, any issues with any high speed routing, because most of your most of your capacitance and everything like that is going to be seen either at the PCB level, or potentially at the substrate level, depending on where your connection is. And so we haven't actually seen any, any issues that the mold compound causes in order to get set high speed signals routed.

There's a lot of years of experience in the mold compound to make sure it doesn't do those kinds of things. And so we're not screwing up that part of the recipe.

Yeah, I imagine since most microcontrollers are encased in this plastic, then it's probably fine.

That's our story, and we're sticking to it.

And so this system of packages is basically you're taking an entire computer and putting it onto a substrate and then encapsulate into a part. So what does this mean for product designers and hardware engineers?

So really, it just makes your life a heck of a lot easier. I mean, so when when you when you look at how much effort goes into routing a DDR I mean, there's there's a whole host of issues you run into with the high speed signaling, the the differential and all the impedance matching all and So there's there's just a whole host of problems to route, the DD the signals between a processor and a DDR memory. And really, we see that that stuff like that doesn't add any value to your system. I mean, you're not going to go put on on your box that DDR works. That's not That's not a bullet point that goes on on the box. So, I mean, I'm gonna put

that on my next product. Yeah, exactly, exactly. I'm, I'm so proud that the DDR works.

But, I mean, you just think about how many man hours have gone into people routing DDR for different projects and everything like that. And so when you when you look at something like a system and package device, I mean, it's that same kind of value proposition that you get with a system on module. So I mean, you're, you're integrating a bunch of different components. And so from a system perspective, your design just got a heck of a lot easier. Now, instead of having to worry about how do I how do I route seven or eight different power planes or power power pours in order to be able to get all the power management, all the different voltage rails and stuff like that. Because I mean, you got a processor running at 1.1 volts, you got core voltage at 1.2 volts, you got DDR 1.8. And so I mean, and then the rest of the systems running at 3.3, and five, and so you have to manage all these different power rails having to manage all of the kind of these high speed DDR signals, it's a lot of effort, that doesn't really add a lot of value to your system. And so you can come at it from much more of the kind of microcontroller perspective where I just need to put something down, I need to hook power up to it, I need to hook my peripherals up to it. And that's my system. And so really, that's that's where, from a system point of view, you get the power of a of a microprocessor. But really, from a design effort point of view, you're looking more at a microcontroller level PCB design. And so

if I, if I take, I want to interrupt you, because if you take that a step further, we're routing things in the matter of a centimeter rather than four or five centimeters. And so everything's closer together, that means I can get a little bit higher speed and get a little bit better signal, my power management becomes a little bit simpler, because I'm not having this long line between the power and the, and the sink. The other thing is when you get to this, why this RF problem, because all of our traces are much smaller than normal. Your and the probability of any one of those traces being an antenna is reduced quite a bit.

Yeah, I was gonna bring that up is one of the big things with like, modules, like Bluetooth modules and stuff is they're usually, you know, pre certified from the F. FCC. And so is using, like a system package device. Like I don't assume it minimizes your risk as a designer for passing FCC CE.

No. Well, so I mean, it's, there's, there's kind of two tasks you can take, I mean, really, the Bluetooth modules that you got that you use, in some respects our system and package devices, they might have a metal shield over them. But in essence, they are a System in Package device, you treat them as just a single device. And yeah, single part, if you pop open the deck deck metal can, I mean, there's a whole host of passives on there, there's a crystal and I mean, and so in essence, they are already system and packaged devices. And so when when we're integrating a processor, a DDR a P MC, and and some other things together, that's more of an IC level thing. And so you don't necessarily see ICS going and doing FCC certification because the route scales like Gene said, or an on a scale that doesn't really it's not going to unintentionally radiate radiate in the bands that you concerned about. And so from from kind of our system and package device, we haven't gone through and done the FCC, sort of like pre verification and stuff like that. Because it just we know that it's not going to radiate. But at the end of the day, you're still going to have to go put your system through the paces, because that's where you're going to get the radiation you're going to be, you're going to be routing, the the clock lines, you're going to be routing, the different things to do our sensors and stuff. But

Eric, you you've you've actually taken a few board level devices through FCC with our CIP on it, and the CIP never radiated, right?

No, no, it was it was always the problems were always other clocks in the on the boards and that's that's always a fun thing to try to track down as to why that is and then then fixing A design when you're at FCC is always always fun, because you're, you're already late. And any any fix you have to make makes you later

and you paid for the whole facility for that?

Yes, yes, it's and I mean, and one of the things we we try to try to as part of part of when we help a customer through some of the design process and stuff, we always want to make sure that you put down extra footprints for caps and inductors and resistors, in order to help yourself out so that that way, if you if you do have a clock line that you need to add an extra filter cap or something like that. You have the footprint on your board, and you can do it while you're doing FCC, as opposed to not having the footprint in which case it becomes either some crazy mod or a board spin.

Yeah, scratch boards pockmarked with either populated or unpopulated, zero ohm resistors. Right. Yeah. Been there done that. So I'm curious, from a hardware engineers perspective, let's say I wanted to use one of your parts, what kind of design resources are available? Because I'm sure it's not as simple as just like, oh, slap it on the board and give it some voltage and it just, it then becomes a computer?

Oh, no. And, but in a lot, a lot of respects. So so we do have a lot of design resources. Because, yes, there is a learning curve for everyone using a microprocessor. Because if you haven't got it, I mean, it's a step up from a microcontroller, both from a hardware level and from a software level. And so you have to have resources out there. And so what we've done is we've, we've done a lot, a lot of application notes, as well as a tutorial series. So on our website, www dot octavos. Systems comm we actually have a whole app Notes section, where there's there's a ref tutorial reference series, which walks you through how you hook up clocks, how you hook up power, how you route, USB, how do you then go bring up the board? Once you've actually done done all this stuff? What are what are the software things? How do you configure the device tree? How do you go through and make sure Linux gets booted. And so we've actually gone through and done a whole tutorial series on that, along with a number of other more specific hardware related app notes about how do you design to make sure that you can source eMMC so are part supports eMMC, version 4.3. And that's an older standard. But the newer eMMC is are backward compatible. But there are some design things that you need to do and to make sure that you don't run into problems when you're actually trying to route this, these newer EMCS to some of the older standards. And so we provide a bunch of different AP notes like that. And so we try to dive in, technically, and make sure that we provide all of the necessary steps that you'd need in order to actually get your system up and running and get to production quicker.

So let me let me jump in on that. Because one of the things we've learned, I said it for years inside of TI, that innovation is inversely proportional to the size of the company. Large companies can't innovate and little ones can. And and by the way, to management, we continue to explain to me that TI was a big company. And my answer is, yes, I know that. But what you find is, as you get more and more to the system level of integration, the volumes per integration get lower and lower, they get the end. So you have to figure out how do we create these devices that are going to be lower in volume, and meet the needs of these innovative small companies? And so that's a lot of what we're working through on how do I go after this new market when the more I integrate at the system level, the lower the volume of my product seems

like a fine line to walk. Yes, it is.

Yes, it is. The good news, the good news, we don't have to play, we don't have to play with the big verticals of which believe you to be their servants rather than your vendor. And so we talked to people who actually are happy that we're talking to them and so we have this innovative crowd that what we've done is put all of the material together to help them do their job where

can people find all that that knowledge that we're just talking about the design docs and stuff,

all of our all of our design resources and stuff like that are all up on on our website. And so you can always go there to see see all the different information, all the different different information, things that we that we publish And then also, we have a couple of different reference designs. So we've posted some open hardware reference designs on our website, we also are a vendor to beagleboard.org and and so you can actually find Octavio systems devices on a number of the BeagleBoard devices BeagleBone a series of open hardware reference design notes and and boards to get get started with your development. So like the, for example, the, the original OSD 335 x part is on the BeagleBone, black wireless as well as the BeagleBone blue, which the BeagleBone blue is a really interesting board to get started with for robotics applications. So you can actually drive DC motors, you can drive PWM for servos, it has a lot of a lot of really cool robotics applications. And then you can actually find the 21 millimeter part, the OSD 335 X SM on the latest offering, which is the pocket Beagle. And so this development board is roughly the size of a mini Altoids. Tin. And so really, it's a very stripped down bare bones kind of design. But it's really in an awesome form factor. And so it has a USB port. So you can plug that in, plug the the other side into your your computer, it boots up, runs. Yeah, and you can run CLOUD NINE IDE from it in order to actually access your Linux terminals, run code on it and do development. And so it's a really, really cool development platform in order to help get started with some of our devices.

So let me interrupt again, if I remembered and Eric, you can embellish this with the the 21 millimeter device, you can actually do a layout with none of the components on the backside.

That is correct. Actually, if you look at the pocket Beagle, you can, you'll find that it is actually a single sided board, and has a very, very few components. Really, it has a micro USB connector, and a micro SD card for storage, a couple of pull up resistors for setting the boot configuration clock and some reset circuitry and a few LEDs to tell you what's going on. And then it has these really amazing expansion headers that allow you to interface with different ecosystems like the micro electronic clipboards. And so there's all sorts of device device tree overlay drivers to allow you to interface with these boards. And so it makes it really nice from a prototyping developing your own embedded system. Point of View.

And that's, that's a result of, since we are System in Package, we actually lay the ball map out to be system friendly, rather than

have all your power grounds and your, your, your eye, your GPIO or for like registers and stuff or all in line. Yeah, like, oh, like you'd have, like gate A, A through one through seven or in order. Yeah, all over the map.

Exactly. And what's cool is, so with the SM device, we actually did, so the pitch is 50 mil, so one 1.27 millimeter. And so it's wide, wide pitch. And so actually with six mil trace space routing rules, you can get to traces between balls. And and so what we what that enabled us to do is actually be able to route out the entire package in a single PCB layer. So we we structure, the power and ground pores and everything like that, such that you could actually do all of your signal escape on on one layer, because everything's in the outer, all the signals are in the outer three rows and columns of the BGA. And then you actually have your power inputs as well on those outer three rows and columns. Then there's ground in the center, and then a couple of of your output power rails in the center so that those can go down.

And it comes down to that dream we all had as systems guys, and that is when will the icy circuit guys actually lay out

in a way without just adding extra layers to your board? Right? Of course. Yeah, yeah. That's right. It sounds like you have it you have it set up where like I said earlier, the expense can go into purchasing your part as opposed to just making an expensive board just to access your part,

correct? Yep.

Is there any other reasons why octaver SIP is an advantage over other SIP offerings.

Yeah, let me jump in and give you one I see I beat you to at this time. One of the things we're looking at as you get to the system level, you more and more have to be more customized to a specific implementation. And so we have, we have a couple of things on the drawing board that will allow our customers to have a portion of the SIP to customize to their taste. And so they can suck in. I guess that's a technical term. Yeah, they can suck in some components in the system that they would like to have integrated into the CIP, which is kind of interesting. In fact, one of the implementations has a concept of just putting a breadboard over in one corner of the CIP. And

yeah, actually, I found a article that Eric wrote on your website about that topic.

Yeah, no, we we actually, we presented a paper at the one of the packaging international packaging conferences about two weeks ago. It's the eye maps conference, if you go to imax.org, you can find information about that, that packaging society. But we actually have done some some layout looking at how, how can you take the concept of a like your I mean, everyone's familiar with a solderless breadboard and how you can go prototype your system design on a solderless breadboard. And so what we wanted to do was take that same type of concept and miniaturize it so that it was inside a package. So that way, basically, through the software of the pick and place machine and the wire bonder, we can actually go through and and basically do a mini breadboard inside a package. And so that allows us to take the burden of in basically allows us to have infinite flexibility with our designs, be by mainly through software configuration of your pick and place machine for the substrate as well as the wire bonder to hook them all up.

It's like a very manual FPGA kind of on there. Yeah, yeah.

Not not not reconfigurable on the fly, but but yes.

So we had a, we had a question from one of our Slack members. And they asked about built in LEDs.

Ooh, interesting.

So give me I can give you the simple answer, we could we can integrate LEDs into the system, there might be a little bit of a problem getting any kind of light out through the plastic.

I figured that would be the problem.

But we're, but no, but but let me finish that now that you're saying, Oh, that was a dumb thing to ask. We're looking at how we could create that. So those LEDs would kind of float to the surface of the package. And be visible. You don't want to

integrate light pipes into your, your your chip design.

That that would, yeah, we could at the end of the day, they have to grind off the top anyway, they inject all the plastic and then they they make sure it's planar on the top. So you just have to have to get all the light pipes lined up correctly. That's that that would be the tough part. Mate. Why and

you're kind of alluding to one of the things that's interesting, once you do system and package, you're no longer limited to only silicone based devices. And so you can you can go Gan or organic, there's a lot of different directions, you can go and stick things on there, that you're that gives you kind of a limitless

opportunity I could see doing like a pipe would be like an integrated, like high end MOSFET style with a smart controller, you know, set up using kind of the SIP technology.

Yeah. And there was an interesting presentation, that packaging conference about how folks at like Global Foundries are looking at how do they integrate Silicon Photonics. So basically, integrating little lasers on to silicon devices, because the the process to create a laser is basically quite different than than it is to create a silicon wafer. And actually, the components inside the laser are contaminants for a silica piece of silicon. And so how do you integrate the two together? Well, you do it with packaging. And so there there's all sorts of interesting stuff from from that side of things in terms of how do you integrate more of these exotic technologies together in order to be able to reduce heat and and increase performance?

So you mentioned earlier about your new product, which is the OSD 335 X CSIP. Yes, that's si hyphen. Si P

Yes. Yep.

That's what It's a computer computer and a set

which one's winning in the office right now? You can tell it's either winning or is complete, I think

complete is winning in the office right now, though there's I think there's still some argument.

I'm against that it's, it's not complete.

I like this guy.

Yo, yeah, just needed. It. Come on. The processor itself has some ATD converters Come on. That's it. That's analog. So

that's, that's a halfbreed. It's only half analog.

I'm can can you share with us the secret of how you create your part numbers?

No, unfortunately, you're gonna have to invite Greg back because he is he is the one who comes up with with all the part names. And so he has the master decoder ring. And that that is the the octavo part naming scheme.

We spoke about that last time. Yeah. Well,

I'll give you a little bit. I'll give you I'll give you a little bit background on the OSD. The other option was OCD. And it just didn't seem to be correct. So we left it with OSD.

So yeah, back to the CSIP. What's special about it now.

So I mean, really, it'll like, like we were talking about a little bit before it takes the integration within the system and package to kind of the next level. So with the SM device, we had the processor, the DDR, the power management, on ldeo, the an EEPROM and the associated passives. And so then, the next major thing in order to kind of complete your system is to integrate eMMC. And so eMMC is a big component of the new CSIP device. So it integrates a four, four gigabytes, or actually up to 16, gigabytes, four gigabytes for the initial offering of eMMC, non volatile storage for your operating system. It also integrates the main oscillators, so you don't have to worry about actually putting an oscillator on your system. So from from the point of view of the system designer, really it is hook power up to it, you don't even have to hook up a clock to it, you just hook up power, it will power up, the clock will power up itself. And then you can you can start using it as a one gigahertz Linux computer. And so really, it's taking that that integration step to kind of that next level, such that you don't have to have I mean, Gene gene, and I'll paraphrase for gene, but genes goal is eventually there should be a package with zero pins. And that should be your system. And and so this is kind of that next integration step of what are the things that we need in a system in order to? And how do we get those inside the package. And so that's where this is kind of the next step in integration for for our devices. And so and then, as as Jeanne can attest, we're looking at what else we can do to kind of move forward and realize that that goal of a zero pin package,

you only need pins to talk to you.

So is this the product that you want to show dead bugged?

Yes, this is this is the one that we can we can dead bug. And so because really, you just have to provide it five volts of power, and and then we just hook up probably probably hook up a USB connector so that you can tell it to live maybe a few LEDs, that'd be that that's kind of kind of what we're looking for from a dead bug point of view. That way, that way, you can actually see that it actually works as opposed to just having to take our word for it.

Yeah, unfortunately, it doesn't radiate. So you can't really see any signals.

I'm thinking I can probably do with I don't know what peripherals it has, but you know, power ground and then like TX RX, so you can get a terminal. Yeah,

no, you you are terminal. You can do Yeah, you do your terminal. I mean, USB is a little bit more functional, though. Because then then you get then you get all the fun stuff that comes with some of the cool Linux software packages because the you can do like rnds over USB, and then you can pop open your web browser, use cloud nine, and be able to actually go through and do all sorts of cool stuff like do file transfers and actually have multiple terminals as opposed to just the single U R DX RX.

Oh, got the in the future when y'all get that dead bug. Y'all have to come back on with Greg. Okay, explain the partner Before, yes. And is there anything else you want to talk about? Oh, I guess not.

I'll talk about one more thing. Eric mentioned the idea of no pins. Oh, in my other life, I am also a professor in the practice at Rice University. And so I have a senior project running, where the end goal is to create that zero pin device.

So how's that gonna look? Like? How do you interface with that? wirelessly? Oh, okay. Okay. So so I mean, the goal of that makes them actually makes a lot of sense. Okay, it generates

is all our has its own sensors and communicates wirelessly. Yeah, so

it would be the whole system now and you think about energy harvesting. So I mean, there's a lot of different energy harvesting ways and a lot of different ways to store energy locally, such that you could do some types of sensor applications. I mean, you think about what would be a cool little sensor to have out there that would wake up every so often take some readings transmit some data back to you. There's a lot a lot of neat interesting applications that that are enabled by something like that.

And our goal is it'd become the size of a piece of Chiclet gum

Wow sounds like the beginning of Skynet to me

yeah, I'll yeah, I don't don't want to don't want to take over the the weapon systems. Just just

Well, I think I think that was supposed to happen in 1997. So we're a little bit behind here.

Just a little Yeah, well you want to sign this out? Oh yes. All right. So I guess from a sign off point of view, that was the macro fab engineering podcast where your guests were Eric Welsh

and Jean France

and we're your hosts spark Dohmen and Steven Craig don't get sky netted Thank you, yes, you our listener for downloading our show. If you have a cool idea project topic, or a way to stop Skynet? Let us know. Tweet us at Mac fab or email us at podcasts at macro comm also check out our Slack channel. That's where all the T 100 Hangouts. If you're not subscribed to the podcast yet, click that subscribe button. That way you get the latest episode right when it releases and please review us wherever you listen, as it helps us show stay visible and helps new listeners find us