From Concept To Reality: A Quick Look At The Journey of Chip Designing

There is no dearth of new chip ideas — ones that are faster, cheaper, more powerful, have newer features, are temperature resistant, consume lesser power, etc.

As the global demand for “smart” appliances and devices grows, so does the need for new chips with different sets of requirements.

Source : Getty

By Shashwath TR

Every device around us today has microchips — computing devices like computers and laptops; communication devices like mobile phones; home appliances like refrigerators and washing machines; smart devices like smartwatches and smart speakers; and more. Some of these devices have more than one microchip - a modern car with driver assistance, for example, can have more than 200 chips in it! These chips control the engine, run the infotainment and air conditioning, help in navigation, etc.

Have you ever wondered how engineers create these extremely powerful chips? If a business has a specific need or comes up with a new idea, how do they build that chip? 

Let’s take two examples. In the first example, a car company feels that their existing car models have an infotainment system that uses an older-generation chip. The company’s engineers are worried that modern hackers can hack them! They want to upgrade their electronics to a new generation of chips. In the second example, a popular kitchen utensil company wants to build smart kitchen products - smart pressure cookers, mixers, grinders and juicers.

In both these examples, the engineers working in these companies will write down the specifications of the chips that they need for these applications. 

At Semicon India 2023, Vinod Dham, who is known as “the father of the Pentium chip,” explained that specifications are written in terms of speed, power, cost and features. Often these are in conflict with each other. For example, the car company’s new chip might need to be high-speed and have safety-related features that need more power. Smart kitchen appliances might need to be very cheap, which will mean a compromise on features or performance. 

Once the car or kitchen appliance company’s engineers have narrowed down the specifications, they will look in the market for an existing chip that matches these needs.

On the other side, semiconductor chip design companies will be looking for broad industry requirements to design new chips. For example, they may say, “The new generation of cars will require a chip with security features,” or “Home appliance companies will need a very cheap but durable chip.”

Chip design teams in these companies will reduce these product ideas into another set of specifications. They will also specify the “PPA” - power, performance and area - of the chip. The last item - area - is important because the higher the area of the chip, the costlier it is.

From the requirements, engineers would either make a logic design (for digital chips), a circuit design (for analogue chips), or a mixed signal design (when both analogue and digital are involved, e.g., a WiFi transceiver). These designs are simulated on CAD tools and prototyped on FPGA boards. This part is called the “front-end” design. 

In the next step, “back-end” chip designers will translate the logic/circuit schematic - typically specified using a programming language like Verilog, into the actual physical circuit. They will plan the layout of the physical Silicon chip, like how we set up furniture in our house/room etc. Front-end and back-end designs may take a year or two for a new chip.

Once a chip is designed, it will be sent to a Fab (fabrication plant) for “tape-out.” In this, the fab will produce a small run of the chip (usually 100). This is the prototype stage in chip design.  After the prototype is proven to work as per specifications, the company will place a manufacturing order for the chips. For chips to be made cheaply, manufacturing orders for chips have to be in lakhs or millions of units. So any small error in the design can cost losses in crores of rupees.

Silicon chips are manufactured in Fabs. These extremely expensive and technologically advanced factories are available only in a few locations worldwide. After the silicon chips are made, these Fabs send the chips to ATMP (Assembly, Testing, Marking and Packaging) plants (which may be in other locations). It is from these factories that the final chip comes out.

As the global demand for “smart” appliances and devices grows, so does the need for new chips with different sets of requirements. Advancements in technologies also give rise to new requirements. For example, some cities plan to implement camera systems to identify traffic violators in real time. These systems require high-speed cameras, fast and secure video feed transmission and AI tools running on high-performance servers, all at an affordable cost.

So, there is no dearth of new chip ideas — ones that are faster, cheaper, more powerful, have newer features, are temperature resistant, consume lesser power, etc. And each time chip designers want to get to a final product, they go through the same journey: idea → specifications → logic/circuit design (front-end) → simulation → FPGA → physical design (back-end) → Tape-out → testing → production.

(The author is the CEO and Cofounder of Mindgrove, a Sequoia-backed chip designing company which is making sure that chips as per design use case are available across industries)

Disclaimer: The opinions, beliefs, and views expressed by the various authors and forum participants on this website are personal and do not reflect the opinions, beliefs, and views of ABP Network Pvt. Ltd.