As the demands on embedded systems continue to grow, system designers have been forced to move from simple microcontroller-based system designs to more complicated microprocessor designs. This move took designs from a single device with a single power rail to multiple devices requiring sophisticated high-speed routing and complex power sequencing. In order to cope with this increasing complexity many designers have turned to using modules instead of designing discrete microprocessor systems. Modules remove the complexity of working with a microprocessor system. The module will typically include the processor, power management, and all the highspeed signals, removing the need for the designer to deal with these tasks. Using a module means the designer can continue to focus on the parts of the system that truly differentiate their product while getting it to market quicker.
However, for all the benefits using a System-on-Module (SoM) brings to the design of a system, they do bring some added costs to manufacturing the system. This added cost has kept SoMs from being viable options for high volume applications where the design cost can be amortized over a larger number of devices. This means that if you were going to design a product that you expected to have significant volume you would most likely be forced to design a discrete system yourself, probably delaying your market launch, just so you can get the cost down.
System-in-Package (SiP), on the other hand changes this paradigm. System-in-Package is a new approach to modules, leveraging the many advantages of the semiconductor manufacturing process, SiP brings all the benefits of a module plus the quality and reliability of IC components to the design without adding extra costs to manufacturing. Our System-in-Package devices are a system by design and a standard component by construction, making them the perfect solution for any volume!
I will spend the rest of this article walking through why SoMs become too costly for high volume applications and how SiPs break free from these constraints allowing an application of any volume to reap the benefits of using a module.
When trying to decide if you should use a module or a discrete implementation for your design you will typically compare the Total Cost of Ownership for each option and choose the one that delivers the lowest cost. Total Cost of Ownership is very specific to each application, but it can typically be broken down into two main areas, the Cost of doing the Design and the Cost of Manufacturing/Sustaining the design. Some examples of the different costs that go into each of the areas are broken out below.
Cost of Design | Sustain/Manufacturing Cost |
---|---|
Cost of Engineering Time | Cost of Components/PCBs |
Cost of Prototypes | Cost of Assembly |
Number of Prototypes | Cost of End of Life Components |
Time to Market (Cost of being late/Benefit of being early) | Cost of failures |
Cost of Certification/Qualification | Cost of Recertification/Requalification |
Cost of Design is typically a single number that is independent of volume. It takes the same amount of time and effort to design a system whether you are building 100 or 100,000. However, the cost of Sustaining/Manufacturing will vary depending on the volume. The cost of components, PCBs, and assembly costs all go down as volume goes up, but the costs of failures and requalification go up with volume.
As stated in the previous section the main benefits with SoMs are seen on the design side, significantly lowering the Cost of Design. However, they do add cost to the manufacturing/sustaining side of the equation, so there is a point for every application where using a SoM no longer makes economic sense and you should use a discrete implementation.
For example, let’s assume you estimate using a SoM will save you $100,000 in the design and launch of your product. You plan to sell 1,000 units a year for 10 years, for a total of 10,000 units. You also estimate that using a SoM will cost you an additional $5 per unit in Manufacturing and Sustaining Costs, for a total of $50,000 over the life of the product. In this example using the SoM would lower the Total Cost of Ownership by $50,000 ($100,000 reduction in Design Costs – $50,000 increase in Manufacturing/Sustaining Costs) and is clearly the right choice for this product.
The previous example looks at a low volume application. Let’s look at another example for a high volume design that is planning on building 100,000 units per year for 5 years, for a total of 500,000 units. Let’s assume using a SoM provides the same $100,000 savings in the design cost. However, because of the volume you now calculate that using a SoM only adds $1 per unit in Manufacturing and Sustaining Costs. In this case you determine that using a SoM will cost you $400,000 ($100,000 reduction in Design Costs – $500,000 increase in Manufacturing/Sustaining Costs). In this example designing the system discretely would be the right economic choice.
As you can see from the examples above SoMs can offer large savings for applications with lower volumes but end up increasing the Total Cost of Ownership as the volumes get higher. This is typically because the unit cost of a SoM cannot scale at the same rate as discrete components can making it more expensive to manufacture a System with a SoM.
At the end of the day a SoM is a PCB that you attach to your PCB. SoM’s utilize the same supply chains, the same manufacturing flows, and the same test procedures as you would use in your final design. Because of this it will always cost the same amount to manufacture a SoM as it is to manufacture the discrete design yourself (for the sake of simplicity we will ignore that the SoM manufacturer will need to mark up their SoM above the cost of manufacture so they can make some money too). So when you are using a SoM in your system you are essentially paying for two PCBs instead of just your own.
On top of the cost of the SoM, most SoM’s require special connectors to attach to your PCB. These connectors can add another 5% to 25% on top of the cost of the SoM to the System. Also there is the additional costs associated with handling SoMs in an assembly line since they typically need to be hand installed.
No matter how hard you try a SoM is always going to have additional costs associated with it on a per unit basis when compared to a designing a discrete system yourself. This means at some volume it no longer makes sense to use one. In the graph below you can see in this hypothetical example using a SoM will add $3 per unit you build and save you $100,000 in design costs. This means that at around 30,000 units it no longer makes sense to use a SoM.
Now that we have looked at the economics behind the decision to use a module and why SoMs typically aren’t economical at high volumes, lets discuss how SiPs are different and can be an economical choice at all volumes. SiPs, like SoMs, provide significant cost savings in the Design and Launch of a new design. However, unlike SoMs, SiP’s offer additional savings on the Manufacturing and Sustaining Side reducing the total cost of ownership.
Contrary to SoMs that use standard PCB assembly processes, Octavo uses use high-volume semiconductor manufacturing processes to build our SiPs. This means the cost of our modules more closely resemble the cost of the actual IC components that are integrated inside of the SiP instead of a PCB with all of the components on it.
Also since our SiPs look and act like standard IC devices they can be used in a standard manufacturing flow. They are completely compatible with standard Pick and Place Machines and go through reflow just like any other IC on your design, so there is no additional manufacturing cost to using a SiP instead of discrete components.
System-in-Package also provides greater reliability, lower cost PCBs, and lower cost assembly giving you a Sustaining and Manufacturing Cost that is equivalent to Discrete while providing the same Design Savings as a SoM! The graph below shows an example where using the SiP saves $100,000 in Design costs while adding nothing to the Manufacturing and Sustaining cost, making it the perfect solution for any volume!
Modules provide significant savings in the Design and Launch of a product however have typically added cost to the Manufacturing and Sustaining Side. This made modules ideal for low volume applications but discrete implementations were still preferred in larger volume products. System-in-Package changes this calculus. It provides the same benefits as SoMs in the design side at the same (if not lower) cost in Manufacturing and design. System-in-Package allows every design no matter the volume to take advantage of modules. The chart below compares the Total Cost of Owner ship for a system designed with a SiP, SoM, and Discrete Components.
Cost | System-in-Package | System-on-Module | Discrete |
---|---|---|---|
Design Costs | |||
Cost of Engineering Time | $ | $ | $$$ |
Cost of Prototypes | $ | $ | $$$ |
Number of Prototypes | $ | $ | $$$ |
Time to Market (Cost of being late/Benefit of being early) | $ | $ | $$$ |
Cost of Certification/Qualification | $ | $ | $$$ |
Manufacturing/Sustaining Costs | |||
Cost of Components/PCBs | $$ | $$$ | $ |
Cost of Assembly | $ | $$ | $$ |
Cost of End of Life Components | $ | $ | $$ |
Cost of failures | $ | $$ | $$$ |
Cost of Recertification/Requalification | $ | $ | $$$ |
"*" indicates required fields
Octavo Systems LLC all rights reserved
OCTAVO is registered in the U.S. Patent and Trademark Office. OSD, C-SiP, and the Octavo Logo are trademarks of Octavo Systems LLC.
"*" indicates required fields
"*" indicates required fields
"*" indicates required fields
"*" indicates required fields
One thought on “System-in-Package: The module for all volumes”
Abdelghani Ouchabane says:
November 6, 2019 at 6:54 amGreat article, thanks
I agree that SOMs are unsuitable for ultrahigh volume productions but they have many advantages which make them very attractive to the designers:
– Scalability: SOMs achieve this by standardizing the footprints and the interface to the baseboard (no need for pin-compatible)
– Switching between processor generations and vendors is much simpler and always possible when the baseboard provides enough power.
– Base on the same design you can design different products with different performance and features is easy to serve different applications
– Extending the life of a product when the life cycle of SOM ends by switching to other SOM.
I will like to add that SIPs are great for small products where COM Express, Qseven, and SMARC form factor does not fit.
Comments are closed.