top of page
Search
  • Writer's pictureDinesh Thogulua

Electronics design for startups

It is one thing to come up with an electronics design to solve a problem. It is an entirely different thing to make it friendly for mass production. If you are coming from the DIY world and have just stepped into the professional world of making electronic products, this blog is for you. Large companies have a dedicated team or teams to do what is called "Design for Manufacturing" (DFM). DFM is an entire field in itself. A startup cannot go too deeply into a subject like that - You need to get things done quickly and get your product into customer hands quickly. However spending just the right amount of thought, applying just the right amount of DFM principles, will actually reduce your time to market, and your time to next-iteration of your product. In this blog, I am going to go over just what is needed for a startup.


You know how real estate agents say that it is all about the location? When it comes to designing scalable electronics, it is (almost) all about picking the right ICs. So we will deal mostly with that in this blog. We will also talk about making scalable electronics that can also be made in small numbers when you are just starting off: After all, you never make anything in large quantities from the beginning itself - You always want to test the concept, find out if the market really needs your solution and likes it, and only then go to mass production. Disclaimer: DFM (Design for manufacturing) is an entire field in itself. Large companies Think of this as a starting point)


There is probably an IC for that

Some people rush into designing their electronics right after the problem statement is ready and hardware is spec'ed out. They may end up designing their circuits with many basic elements without asking if there is an IC that can take care of a whole lot of functionality of what you are want. For example, you may need to interface your device with several sensors , use PGAs to shift and scale the outputs of these transducers, and then use an ADC to get digital values from the PGA outputs. But remember that you aren't the first person in the world with such need! So don’t rush to use an analog multiplexer that you have used before, a PGA chip that you are familiar with, and an ADC that you have used time and again. Instead check if there is an IC that has all three functionalities in one package. In fact, I would say, there are probably, not one, by many such ICs, from different manufacturers, at different price points, spanning different temperature ranges etc. Similarly, if are you in need of multiple regulators, some linear, some switching, with the ability to turn them on/off using digital signals, don't design a circuit with that many regulators. You may be able find a PMIC that contains all those regulators and more.


If one IC can take up the place of three different ICs, a whole bunch of pull up, pull down resistors etc., it means your BoM will look a lot shorter. You will also have less things to test and debug: Let us face it: Texas instruments or NXP will be testing this one IC way more thoroughly than we can ever test our own solution of three different ICs and other components it replaces. Furthermore, lesser components means lesser components that can go wrong, less space occupied on the PCB (which translates to lesser PCB manufacturing, assembly cost) and less soldering during the PoC stage.


Choose Persians over Spartans. Every time.

You will often find yourself spoilt for choice when choosing ICs. There are just way too many manufacturers making the same damn IC. And each manufacturer makes them in way too many packages - SOT, QFN, VSSOP, BGA…. You need to some criteria to by which to choose just one among the many choices. One important criterion is their availability in large numbers. The one available in large numbers may not be the ideal one for you: You may not like the cost. Or you may hate the package - but you cannot choose an IC that is only available in small numbers even if the IC is just the perfect fit for your needs. Because, you really don’t want to find out, the day you get a large order from customer(s), that the IC is either not available or the lead time is too long.



Besides the simple aspect of availability, if something is popular, it is probably for a good reason: It may be that the temperature range is what is demanded by majority of the market. Or, it may be that the pin pitch is liked by most PCB assemblers. Whatever is the reason for its popularity, if something is good for so many designers out there, it is probably good for you too. So always go for items that are stocked in large numbers by suppliers like Mouser, Digikey, element14 etc.


No Ol’timers Please

Everyone dies. That is the hard truth of life. The ICs you love may not be around an year from now. If there is even slim chance that you will still be producing your electronics an year from now, you better choose ICs that are not at the end of their lives.

All IC manufacturers indicate the “product status” in the datasheet of the product. For the package that you are interested in is mentioned as “ACTIVE”, then you are good. But if it says “NRND” or “Last time buy” or “Limited”, then don’t go for it. Sometimes the suppliers themselves display “NRND” - i.e., “Not Recommended For New Designs” - prominently, right next to the item.



If you are making consumer products which are produced only for a short period of time, this may not be a big issue. However, for every IC that is at the end of its life, you can find an alternative that is active in a similar price range. So, why would you not just choose the alternative? It hardly takes more than 30 minutes to choose the right IC.




If the BoM is Long, You are Wrong

Try your best to consolidate components. If you already have many 4.7K resistors, then replace a 10K resistor elsewhere in the design with two 4.7K resistors in series. If you are using two different regulators, one that can supply 1A and the other 2A, because 1A is sufficient for part of your circuit, just go for two 2A regulators, even if the 2A regulator is slightly costlier than the 1A regulator. By shortening your BoM list, you are ensuring that you can make the most of volume discounts. Furthermore, the pick and place machine at your PCB assembler’s may allow for only a certain no. of reels to be loaded at a time. If you have more components, then they will be forced to go for two passes for assembling your PCB and that is going to cost you much higher.


Pay More, Get More?

Wherever you can, make sure there are higher spec’ed variants of the same footprint of the components you are choosing for your current design. “Higher spec” may mean wider supply voltage range, operating temperature range, higher switching frequency, lower quiescent current etc. That way, in case you come across a new customer who wants the same functionality of your product, but the ability to work at a lower voltage range or higher frequency, you can satisfy their requirements without having to redesign the whole thing (schematics, layout).


Miscellaneous

Your PCB manufacturer and assembler may have machines that have some limitations and some preferences. Talk to them and find out before you design something for mass production. Some may prefer components only in reels. Some don’t like tubes in general. These days, most manufacturers have machines that can work with all kinds of packages. However, it doesn’t hurt to talk to them first.

The first batch of your product is going to be in small numbers. So although this blog is about mass production, you have to keep in mind that no product goes directly to mass production stage. So it would be wise to choose components, that are friendly to any-volume production. It means that the component should be hand solderable, and at same time, available in mouse reels or digi reels for medium volume production and reels for large volume production. It also means that you give yourselves a few test points on your PCB for measuring current consumption, some load switches to enable you to debug issues by completely turning off some sections of your design, expose debug ports such as UART etc.


62 views0 comments

Recent Posts

See All

Works

Commentaires


bottom of page