[Electronic Hardware and Software Design Solutions]

The Design Process - Choosing Components

Randy Nuss
Copyright 1999 Idea Consulting

OK, so your circuit design is sketched out, and you have a pretty clear picture of what components will be necessary to implement it. Now you must choose actual components for your design.

Component choice has the single largest impact on product cost and therefore profit, so extra care should be taken at this step. There are several conflicting factors to balance when choosing a particular component. These factors include performance, cost, quality, lead-time & availability of alternate sources.

Design requires balancing tradeoffs to get the most performance and reliability for the least cost. Described below are several general guidelines for wisely choosing components. I hope they prove useful to you.

#1 - Minimize the number of unique component types!

Depending upon the application, component cost can play the major role in which particular part to select. Keep in mind that component cost is highly dependent upon volume, so it’s good to have an idea of your estimated annual usage for the component when asking for budgetary pricing from the component vendor. It is for this reason, that one should strive to minimize the number of unique component types. By doing so, the volumes of those parts actually used will be larger and greater overall savings will result.

For instance, suppose your microprocessor based design required 2 ea. 74ALS245 octal transceivers and 4 ea. 74ALS244 octal buffers. Note that an octal transceiver will work perfectly well as an octal buffer so why not modify the design to use 6 ea. 74ALS245 octal transceivers? The volume increases on the 74ALS245 by 50% and there is no need to buy or stock the 74ALS244.

This cost reducing strategy works well when the components are similar in price. Obviously, it would be unwise to substitute a much higher cost component for a low cost component to eliminate the lower cost component, unless perhaps there are only one or two of the lower cost component in the design. The engineer will have to make these tradeoffs to optimize the overall cost to produce the product.

It may not be apparent, but each unique component type in a design carries a price tag throughout the manufacturing process. Some of these costs are:

  • Cost of procurement - A buyer must take the time to purchase each component type.
  • Inventory and storage - Each unique component requires space in the stockroom and ties up cash in inventory.
  • Assembly - Each component type must be handled by assembly personnel and machines. Each unique type handled increases the odds for a mistake to be made. Quoting algorithms for robotic assembly systems typically include a fixed charge per component type.
  • Field Service and Repair - Field service personnel and repair technicians must familiarize themselves with each component on the assembly. This represents an incremental cost per unique component. Spare parts inventory also adds to long term cost.

For all of the above reasons, one should take the extra time to analyze the finished design and look for ways to minimize the number of unique component types before going to PCB layout.

#2 - Choose components with alternate sources

Always attempt to design with components available from more than one supplier. Inability to procure ANY component on an assembly effectively stops the manufacturing process for that assembly. Entire production lines can be stopped because of the lack of a single component. This is unacceptable for any sized company and is guaranteed to get the attention of senior management faster than a speeding bullet.

Second sources give the buyer leverage and choice in deciding which vendor to buy from. By designing a part with at least 2 and preferably 3 sources, money can be saved at least 50% of the time. By having more choices, the buyer can always create an atmosphere more favorable to your company.

Naturally, the part from the second source must fulfill the design requirement for that particular component. Most corporate MRP systems do not allow for a preferred component and an alternate component. Many times I wished that I could specify a preferred vendor and a second choice if the preferred component was not available. Inevitably, whenever I got an assembly back from manufacturing, ready for debug, I would see the off brand component soldered in instead of the name brand one. Purchasing simply buys the least expensive or easiest to obtain component from the sources specified during the design. For this reason, make sure that all sources specified for a particular component will work robustly.

In some cases, alternate sources are simply not available. Some examples of electronic components for which there are either limited on non-existent alternate sources include: microprocessors, custom logic including FPGAs and Gate Arrays, sensors, switches and relays, some inductors and some connectors.

If the design calls for a component with no alternate source, choose a component from a larger, well established

#3 - Use components that are already designed in to your company’s products

After you have a pretty good idea of what components will be required in your circuit, spend the time to look through your company’s component database (e.g. MRP) for specific components that you can also use. This will save the company money by increasing the volume for that component. You will save the company more money since there won’t be a new part to stock etc. (See Minimize the number of unique component types above).

Its great to save the company money, but you’ll also save yourself a lot of time too. Researching alternate sources, gathering data sheets and filling out part number request forms are time-consuming tasks and really not much fun. Most component database systems are not easy to search, but persevere and you shall be rewarded.

#4 - Choose components which are used in very high volumes by others

Because components made in large volumes tend to be less expensive, one should try to use components that are purchased in large quantities by others. For example, the primary use of DRAM memories is in the PC industry where these devices are used in the millions. Using simple DRAM instead of more exotic types like dual-port RAM will significantly reduce your overall product cost no matter the application.

This strategy does have its downside. Choosing a component that has one specific high volume application and is nearing the end of its demand life cycle could be a big problem for purchasing 4 or 5 years down the road. Microprocessor chip sets and memories seem to have the undesirable property of rapid obsolescence as semiconductor manufacturers shift capacity toward newer (and more profitable) products. Nevertheless, significant savings on component costs can be achieved by making a conscious attempt to find and design in popular devices.

#5 - Prefer components that do not degrade with time or use

Whenever possible attempt to use components which will not age or degrade with use. I would like my designs to last 30 or more years of normal use. For this reason, I try to avoid the use of fuses (use resettable polyswitch devices or circuit breakers), relays (use MOSFETs or other semiconductor switches whenever possible), electrolytic capacitors (tantalums cost more but are somewhat smaller per microfarad), manual switches (try to use software control) and any other component which has the potential to degrade.

Make sure that you use appropriate derating to ensure that semiconductor junctions will never be exposed to voltages anywhere near their absolute maximums. Dissipative devices such as semiconductors, resistors and inductors must be sized properly to ensure that internal temperatures stay within reasonable limits, even when operated at high ambient temperatures. Heat and voltage are an electronic components worst enemy.

#6 - Choose components that exceed the required performance, but don’t overkill

Clearly, designing in a component that does not allow the product to meet its performance target is not wise. It is also unwise to overdesign a circuit. For example, designing in an op-amp which has 10 times the bandwidth actually needed, can lead to unstable operation including oscillations and ringing.

Choosing an oversized programmable logic device with many pins unconnected because "we might need to add something in the future" wastes money needlessly. The designer should know enough about the thing being designed to put the needed circuitry in, up front using the smallest (and least expensive) device that meets the design criteria.

Unfortunately, one cannot always be sure of the software requirement. How fast should the processor be? How much RAM and ROM will be required. Software tends to get larger and slower as development proceeds. For this reason, an exception to the "avoid overkill" mantra is, if possible, allow for memory expansion and faster processor clock rates. This way you may be able to avoid redesigning the board when, inevitably, the software becomes too large or slow.

Summary

Hopefully, you will find the preceding guidelines useful in choosing components for your next design. It really is one of the most important tasks to achieve a low cost, high performance product design. Good luck!

HOME

Copyright 1999 Idea Consulting