With the continued growth in advanced portable (and not so portable) electronic devices, there is a demand for increased circuit density to serve both cosmetic, haptic, and form-factor issues as well as greater functionality integration. High levels of system integration are needed to address this burgeoning need for more stuff in a smaller space.
System-in-Package (SiP) solutions are one means of attaining this goal. An SIP can be the core of a small device, or a building block in a larger and/or more complex solution. Developers like Octavo Systems create SIPs based on standardized systems that can be applied to a wide variety of different applications. We recently had the opportunity to talk to Octavo Systems’ CTO Gene Frantz and Marketing Manager Greg Sheridan, going over the company’s past, and what they see as their future.
Alix Paultre: Tell me about the origins of the company, Gene. What got you started on this path?
Gene Frantz: Many years ago, I was at Texas Instruments looking for new innovations out in the industry and a friend and I found this little company called Tivo, you may have heard of them. We went to visit them, to see what we could do with TI technology to help them. When they shared their bill of materials, I found it interesting that there were no two components in the system from the same vendor. I then realized that this system, like others I had seen recently were designed with a new methodology called Design by Google. “Google” the keyword; find the best component there, and then stick it on the board.
Gene Frantz: I began to talk to people about how this integrates down to a single chip solution, so when Tivo becomes single chip, how many vendors will that be? Well, the answer's one. And so, I saw this more as an a-ha moment for Texas Instruments than I did for the customer. If I wanted my customer to stay with me, I needed to figure out how every time I did integration that all the components would be mine; and when it got to one device that was sold to them, it was my device. So, that was my a-ha moment on system integration. What we're talking about here with the C-SiP or the Complete SiP, we're talking about everything you need to run a computer in a single package. It is not a complete system. And the reason for that is that I have to define what a system is; and a system could be my cellphone, the system could be the wireless network in our building or in your building, the system could be the cloud as a whole, so how do I say I'm doing system in package if I can't put the whole cloud computing system into a package.
Alix Paultre: Right. Actually Gene, to me, when I think of a complete system in a chipscale package, I think from the engineering point of view, all I need to do to it is add a power supply, a case, and a touchscreen display, and I'd have a product. For example, some of the more highly integrated FPGA systems out there currently, you know what I mean? Or, if I'm staring at one of the top of the line, 32 bit microcontrollers from Microchip, we're getting really close with some of those systems. Now of course, those are actually multi-chip integrated devices, but you know what I'm talking about in that sense, that we're getting to a point where we will have a product where all you have to do is put an antenna, display and there will be external peripherals. So to me, at the engineering level, the system is, I only need to add peripherals and I've got a device.
Gene Frantz: So, we're almost there but we're a little bit different in that our end goal is computers that no longer need to talk to humans, only talk to other computers, what is a system then? So, that's our ultimate system.
Alix Paultre: Out of curiosity, are you software agnostic or are you gunning for anyone?
Greg Sheridan: I'd say we're software agnostic. We'll run anything that you want to run, or that can run, we'll be happy to do it. There are two sides to that question. One is, yeah we're agnostic. You can put whatever software you want on our products but there's a second side and that is, as you begin to go for the more innovative customers in the world, the more innovative designers in the world, you actually can't give them that much latitude or they won't be able to do their thing.
Gene Frantz: I have talked about the concept of sticky platforms.
Alix Paultre: Please elaborate.
Gene Frantz: OK. Let's just assume that you're my customer and you are wanting to put electronics into clothing.
Alix Paultre: OK.
Gene Frantz: How good are you at writing software? Probably not at all. And so, if I gave you a platform and say, “you can use whatever software you want on it and whatever operating system you want on it,” I will probably not work with you very long because you will spend a lot of time and may never get to the point where you can be creative. And so, sticky in my mind is, how do I add more capability to my platform so that more innovative people, who are not as technically savvy, can use it to their advantage. And sticky says that if I make it easier for you to use than anybody else, you will tend to stick to my platform just because it's the easy path to your solution.
Alix Paultre: I'm trying to think of the best way to phrase it, because I don't want to be impolite. Because I'm a big open standards guy. And whenever I hear about limiting development choices, and I recognize some of this may be well intentioned, I have the fears of a proprietary system that may trap users into a product field.
Greg Sheridan: Just elaborating and going down your path. When Gene's talking about sticky here, he's not saying that its proprietary code or a proprietary system. If we can figure out how to package our device with an open source Linux distribution that is just super easy to run and get people up and going on as quickly as possible they're going to want to use our system because, hey, everything just works. You can still take our device and do whatever you want on it but sticky doesn't necessarily mean proprietary, it just means giving the user an experience. Kind of that out-of-box experience that allows them to just quickly start creating.
Alix Paultre: Like I said, that's why I was trying to think of how to best ask. It's not so much that I don't like proprietary systems even, my philosophy though is that you can have angels dancing on pins on the inside of the box, but the outside of the box has to be able to communicate with the rest of the world in standard protocol.
Gene Frantz: That is why I used the term sticky. I can assure you for many, many small, very innovative companies, open source is very sticky.
Greg Sheridan: And what we're seeing is that in the past people that were innovating in this space were focused on electronics. They were experts in electronics, electronic design, software. The trend today is more people are experts in an application or experts in a field, but not necessarily in electronics and they want to take advantage of these technologies to apply them to their field. So, how do we lower the barrier to entry in the electronics space as much as possible that allow these experts in their certain fields to take advantage of this technology and innovate in their fields without having to know all the ins and outs of electronic design, like DDR routing, power sequencing or how to build a Linux kernel, right?
Alix Paultre: Right, I hear you completely. I'm totally there. It's just, the other side of that coin is I've also heard a very similar pitch from a lot of companies and my worry is that there may wind up being a monkey knife fight amongst competing protocols.
Gene Frantz: Yeah, and this is all good. And once again, if I go back and tell this story as I was from the position of the clothing person who wants to put electronics in the clothing, for some unique reason, you've just described a great moment for them and their path because there are four or five different companies who are doing their best to make it easy for them to take their platform into their clothing line. Which means what we have done, as an industry, if we put ourselves in the semiconductor world, we have put ourselves into where we are competing with each other on what we do well for the good of our customer, giving them choices. So, it's good that you're hearing the same story from multiple sides, multiple companies. Now the one thing we have going for us as a SiP manufacturer is, we're silicon agnostic. We could put any one of those systems down on our substrate and build a system with it. So, we're not tied to a silicon position, we're tied to a system integration position.
Alix Paultre: Right. Well, my position is that with this high level of integration and technology convergence, the next generation of engineers must be better systems integrators because the only people who are going to be designing power supplies 10 years from now, are people who work for power supply companies.
Gene Frantz: You're exactly right and, you know, one of my other jobs in life is I'm a professor at Rice University, and what I'm having the most interesting time with is trying to help the university and the students come to the conclusion that we need to learn to integrate at the system level and not at the component level.
Alix Paultre: Yes, agreed. But then again, going forward, I think we're going to see more and more modular solutions, surely because of the fact that, all of the solutions are going to be modular and integrated to that effect. My decision will be between which chip I should select and the engineering comes in when I'm matching my antenna to the circuit board.
Gene Frantz: And we see it a little bit different from that in that, and the reason I'm saying that there are things like power dissipation, performance, reliability, if I can tie the PMIC and the memory and the processor all together, I can impact all of those aspects in, what might be, non-obvious ways but they can be impacted.
Greg Sheridan: So, I'm curious Alix. You're talking about these modules and stuff, I can't tell. Do you think that being able to pick from these different modules - is that a good thing in your mind or is that a bad thing?
Alix Paultre: I think it's a good thing as long as they don't lose track of the core engineering. That's why I was saying that it's all well and good to be able to select from various subsystems solutions but then you still have to be able to match the antenna to the board, because if you don't, you'll waste RF energy. And, that skill is not something you can pick out of a parts bin.
Gene Frantz: You're right and one of the comments I've made about SOC, system on chip technology, is it really should be called SSOC or subsystem on a chip, because each time we accomplish what we call a system on chip, it becomes a subsystem on a larger system.
Alix Paultre: Yes. Once upon a time you could buy a DSP as a single chip.
Gene Frantz: Yeah. And, that is what we're after with SiP technology: I can buy a DSP with its power managed and, with its associated memory and its associated sensors and actuators all in a single, and I'll use it in quotes, "chip." I look at it from the module approach here. I look at it as a parallel to what's happened in the software realm with code and code reuse and how code is modularized and re-used. I look at us and see this module approach as what's really happening in hardware. That is getting these basic building blocks like the C-SiP, it's a core functional block that can be reused in a bunch of different areas.
Greg Sheridan: Now you still need to have, like you said, some knowledge of electronic design to be able to integrate this and make it unique for your system. What these modules, re-usable modules, do is, allow you to not have to go and repeat and reinvent the wheel every time. Hooking up DDRs to a processor isn't something that's going to differentiate your product. It's the thing that is just needed for you to do your work so ...
Gene Frantz: And will continue to get harder to do.
Greg Sheridan: Yeah, so by integrating it into a module, you can just put it down, you can focus on what makes your system unique, doing those tweaks, doing that RF design a little bit better, that way your system gets better performance or something that sets you apart in the market, right, without having to spend time on some of this table stakes type stuff.
Alix Paultre: Well, I agree with you completely and I do see it as a good thing. Like if it’s just as long as they don't look at it as a development crutch.
Greg Sheridan: The idea of a developing crutch is interesting, right? We've had some feedback from customers that are using this technology now that haven’t been able to before because we've integrated it in a SiP.
Alix Paultre: Yeah, but that's enabling. I don't consider that a crutch at all. I think calling the FAE and saying, “Hey I’m having trouble getting this to talk.” Or I'm having trouble ... something that they should know, like making sure they have the right capacitor or that their antenna isn’t terminated properly is one of the things that pops at the top of my mind because no matter how modular and how integrated the system is you're still going to have to make sure the antenna fits.
Gene Frantz: I agree. As an example, what we've found by integrating into a SiP is we don't need to use as many bypass capacitors as the vendors say we need. And so, we've actually lowered the cost and, in essence, increased the reliability because the path lengths between the various components is so short, and from the power supply to the components is so short we don't need as much bypass capacitance to make sure the power stays regulated.
Alix Paultre: There are all kinds of benefits when you have that level of integration, aren't there? I mean, does it also affect your thermals, it'll affect parts count, it will obviously affect parts count. The footprint savings will cascade through ergonomic ability in that you can maybe do things with the packaging you couldn't do before.
Gene Frantz: And we've done all those, yes. That's all the fun of it. One of the things I used to yell and scream at our circuit designers that... should I say semiconductors designers is they would pin out the device to their convenience not to the convenience of the customer. And, they never did understand how important it was to be thinking like a designer using your part rather than the designer of your part. If you look at one of the things we've done with the C-SiP is we have painfully laid out the ball maps so that it's easy to layout in the fewest layers of a printed circuit board.
Alix Paultre: What I was about to say is, for the audience, why don't you spell out exactly what's on the C-SiP so they know exactly what we're talking about?
Gene Frantz: I'll let Greg handle this part.
Greg Sheridan: Sure. So, in the C-SiP, so we call it a C-SiP because it is a complete system in a package. It has your Cortex A8 processor from Texas Instruments, an AM3358 Sitara processor from Texas Instruments; it has the PMIC that goes along with it, a TPS65217C PMIC; it has an LDO that we threw in there as well so people can get a 3.3 volt power rail out to power the rest of their system, it's a TL5209. Then we include up to a gigabyte of DDR3 memory to go along with the processor, up to 16 gigabytes of eMMC non-volatile flash, an EEPROM for configuration data, and then we also include a low jitter, low power oscillator. So really, when we look at this, all you need to do, back to your example earlier of system design, is hook up power, your peripherals and you are up and running with a system now.
Gene Frantz: Effectively you can dead bug this System-in-Package and have it up and running.
Alix Paultre: Very nice. Having said all that about not needing design crutches and all, sometimes you really do need assistance. What kind of handholding do you offer? I mean, you’ve got to be easy to talk to them, you've got reference designs, what do you have that can help them get their system integration moving along?
Greg Sheridan: Our mission, from the beginning has been to make it as easy as possible for people to develop electronics. So, we look at that, not just from a system in package approach but the entire ecosystem and how we engage with customers. So, like everybody these days, we have all of our information on the web. We have reference boards that we've created that we make available, and we make all of their schematics, layouts, and all the design files open source, readily available on our website. We don't even make you give your email address or anything like that to get them. We try to make it as easy as possible for everybody to get their hands on them.
Greg Sheridan: We also have a whole collection of application notes. We actually have a tutorial series that will start you off from “OK, hey, this is my first time I've used Eagle, I'm opening it up, how do I go ahead and build a system using this SiP,” and then will basically walk you through every step of the design process and, then through the manufacturing process. So, really trying to help people out.
Greg Sheridan: And then we do have the forums with an active community and we do have a team of FAEs to support our customers as well.
Alix Paultre: Excellent. First off, how can they get in touch with you? What's your website, are there contact phone numbers some place they can reach in the front door and talk to you?
Greg Sheridan: Yeah, OctavoSystems.com is the website. Obviously, that's available 24 hours a day so, whenever anybody wants to stop by, it's great. As I mentioned earlier it has design documents and application notes there. It also has a place for people to contact us and of course the forums. We do have a phone number.
Greg Sheridan: It is 512-861-3400 but I'd say the website is probably the best way to get in touch with us and we really encourage people to use the forums. We've got quite an active community out there right now.
Alix Paultre: Excellent. So what's next? And before you say what's next, what are some of the applications spaces that you see that could best benefit from this?
Gene Frantz: A lot of the areas this works well for that meet a lot of the criteria are industrial opportunities where you need a computer system in your product and so it works nicely in those areas. If you look at any kind of robotics, it works nicely there because you can manage the system relatively easy. So, those are the main areas. Did I miss any?
Greg Sheridan: In addition, it's the industrial Internet of Things, IIoT or Industry 4.0, that kind of application where you are adding smarts to sensors. This current device is really strong as an aggregator node. If you look at the whole smart sensor, Internet of Things structure, you have the sensors and then they talk to that first level aggregator that then communicates up with the larger cloud. The first level aggregator is really the sweet spot where this thing fits because of the flexibility of the interfaces that are on it.
Greg Sheridan: It can talk to a wide variety of industrial protocols so it works really well in industrial applications. It can easily interface to different Wi-Fi, Bluetooth, LoRa, and other wireless interfaces, so it's really that first level aggregator is the sweet spot.
Alix Paultre: OK, so now then, Gene, what's next?
Gene Frantz: When we started this company, I had a couple of premises that I thought were important. One is, I have a mantra that says, "Innovation is inversely proportional to size of the company." So, large companies can afford to innovate but don't really know how. Small companies are innovative but can't afford it. And we chose to aim our product at that innovative part of the industry and work with them to go off and create brand new things and then what we're giving them is an integration path that they really have never had before. We know how to take, at this point, the significant parts of their problem and integrated them into a system component of which now they can use and focus more on their world of innovation than our world of integration.
That's the present thing. In the C-SiP, something we're looking to as the next step, and it's kind of one of those "am I stepping over the bounds of speaking out of turn or am I giving you a little look in the future?" and I'll say a little look in the future, Greg will say I'm stepping out of bounds.
And that is, how do we give the customer some flexibility to do a semi-custom version of the C-SiP that adds some more of their components into it. And remember, we're not committed to only one vendor. If you look at what we have, we have multiple vendors in our C-SiP when you count the passives, the eMMC, the DDR, the ARM and the power management, so we already know how to go to multiple different vendors, to pull those in and so, if you have a significant part of your design that you can see pulling in, then we're going to go down that path.
Now the question you have of going down that path is what is the minimum volume to make it interesting? We believe we have ...
Alix Paultre: It's got to have to sell to get the volumes to be able to afford it.
Gene Frantz: Well, that's right and understand there's this little phenomenon: the more your component looks like the system, the lower the volume it has. And so, how do we resolve the issue of lower volume ... multiple lower volumes system components without the costs of lower volume? We think we know how to do that.
Alix Paultre: Excellent, excellent. I didn't notice the price point in here. Is there a price point that we can quote? And availabilities?
Greg Sheridan: So, availability for this device should be early next year. The price point we're quoting for the first device which will be in the commercial temperature range with 512MB of DDR3 and 4GB of eMMC is $35 for 1K volumes. And again, our goal when we create these, is to be right in line with a discreet equivalent solution, so if you look at the cost of the components plus the board cost plus the assembly cost and all that kind of thing together, we believe our pricing strategy puts us at a cost advantage when you compare it to total cost of ownership of a discrete solution.
Gene Frantz: And then you add a couple of system level cost advantages. One is the re-lay out of the ball map, so you use fewer levels of interconnect on your printed circuit board, that's one. Another is a little bit better reliability because you have everything in one component all working together, all captured in a package. And then the manufacturing cost is a little bit lower because we are using semiconductor manufacturing companies to assemble our parts rather than a PC board shop.
Alix Paultre: Actually, I've almost literally run out of steam cause I can't think of anything else that I don't think we've touched upon. Do you have a final word for our audience?
Gene Frantz: Yeah, just real quickly. We believe that we're at the beginning of the revolution of a semiconductor technology that will allow for system integration rather than component integration. We're at the beginning. When I talk about Moore's law, I begin to not talk about Moore's law but Moore's original vision. And, we believe we're still on Moore's original vision, on that path, because if you've read his article, he really was more focused on we're going to be able to integrate more and more as we go into the future to make things better and we almost see the SiP world as Moore's law 2.0. If you'd want to use that terminology.
Alix Paultre: Right, the higher level of system integration.
Gene Frantz: Yes, but still integration and still putting more transistors on the same platform.
Alix Paultre: But now in co-habitating packages?
Gene Frantz: Yes. And, at the end of the day, the problem is not silicon. The problem is packaging. One of our founders was one of TI's leading packaging people and what we're innovating on is how do we package these things both cost effectively and reliably. Without changing...
Alix Paultre: I've always been a packaging nut.
Gene Frantz: Yeah, but it used to be the thing you thought about when you were done with your IC design and we're trying to put it up front as the first thing we worry about.
Alix Paultre: Right, I've got you completely. And actually, when I think packaging, the biggest thing in my mind right now is parasitics with the current generation of new semiconductors.
Gene Frantz: Oh yeah. That's why I say once we can control. I should put it a different way. The more of the system we can control, the less we worry about parasitics.
Alix Paultre: Yes. And the more highly you can integrate the package, the fewer opportunities for parasitic situations to occur.
Gene Frantz: Oh yeah.
Alix Paultre: Alright then, thank you both so much for your time.