On a previous poll, Tell Your IC Vendors What You Need, we asked about your selection process in general when you do a board redesign. That got us thinking about fine-tuning the poll to take a closer look at the needs that specifically drive your selection process.
If you do a board redesign — and if the opportunity is there to do some significant rework — how would you proceed? Some possible considerations relate to picking a device that is a lot newer so that obsolescence is not an issue. Perhaps shrinking PC board size, meaning figuring a way to put more analog functionality in a smaller area. Or maybe you just need improved performance due to market driven demands. Possibly it's just a matter of taking out some of the cost.
With that in mind, we asked, “When you do a board redesign and choose to use a new analog IC in place of one or multiple previous devices, what is the primary reason?” We got 53 responses to our “pick one” survey. The results look like this:
That last item in the list of choices, lower cost, drew a bit under one fifth of the respondents. Engineers like lower cost, but it's not foremost in mind or in their needs.
Of higher importance was the problem of obsolescence. Just over a quarter of the respondents said the reason was simply that the old (existing) device was going end of life (EOL), so they had no choice.
About one fifth of the respondents said they wanted to shrink the board size. This would imply that they want to gain a size advantage via increased integration of functions. Note that shrinking the board size generally will take some of the cost out, so even though that wasn't a priority, it's a pleasant result.
About 34 percent of the respondents said they were looking for improved performance. Sometimes, that just means a functionally equivalent part with better specs. But sometimes, as with shrinking board size, it means gaining a size advantage via increased integration of functions. If that's the case, over half of the respondents would be looking for a higher level of integrated analog.
So, while the survey didn't adhere to rigorous statistical guidelines, it does indicate continued interest in integrated analog parts. How does this match up with your experience?
Related posts:
>> So, while the survey didn't adhere to rigorous statistical guidelines, it does indicate continued interest in integrated analog parts. How does this match up with your experience?
That has been the pattern and trend. The more integrated is a solution, the cheaper is the overall cost budget. It saves space and cost. That is the way I see my demand curve in the industry.
Nearly a 2:1 preference for performance over price. This lines up with what the high performance analog semiconductor vendors claim. On the other hand, it begs the question, why aren't prices for analog semiconductors going UP? Gross margins, the amount a manufacturer nets after the sale, are definitely going down.
Engineering cost is high for building the integration. Also, when the high integrated IC chip delivers to clients, client might want to see any debug tool for each function block with more description. In this case, it is not easy to revise the integrated chip when part of block is required to modify.
@Daej you're right. The competition in today's industry demands that a company must learn how to design and build products in the most cost-effective and the fastest way. This has given rise to a whole new way- the use of reusable functional blocks, this could help us meet the product's requirements and shortened development cycles.
I have often to shrink the dimension of my board, this is normally due to the customer's space saving requirement. The number of integrated parts on customer applications increases progressively then the space for my boards decreases accordingly.
Usually the need for a redesign comes from a higher reliability need because the board is used for production, so parts with a better perfomance in terms of reliability are desirable.
It happened to me sometimes that I redesigned a board with cheaper components , but the board failed because the components on it had a worst reliability , at the end this redesign increased the costs !
I redesigned my board many times to add new features like for example trimming circuits to center a specific parameter , who was important for the application.
One of the main advantages in a board redesign activity is the experience that an engineer has got ,one knows what parts of the board are exposed to failure and what are the mechanical / thermal / electric stresses applied to the components of the board.
I agree with you, integration is the future but the process has to be accurately performed, because many effects can rise while integrating the components, due to many factors: heat dissipation, mechanical shrinking etc…
I had to redesign some testing board and requested some parts with specific characteristics but marketing people many times pushed for a different selection of the parts.
They selected parts coming from their preferred vendors.
I think that marketing people should have a technical background, sometimes it happens you choose a list of components with specific electric requirements and the purchasing people propose different parts that cannot guarantee the same perfomance of the board.
>> Also, when the high integrated IC chip delivers to clients, client might want to see any debug tool for each function block with more description.
I do not think any client cares what you have done in each of the blocks inside an IC. They always put inputs and check the outputs, if that is not right, they call another vendor!
>> Nearly a 2:1 preference for performance over price.
That makes sense – you need to have a product before you can think of pricing. If the product does not work, no one cares about the price.
>> I have often to shrink the dimension of my board, this is normally due to the customer's space saving requirement.
That has been the main driver of integration. Without monolithic integration of ICs, the consumer electronics industry will not have advanced at the level it is now. Expect more integration in any way that can be technical sustained.
>> so parts with a better perfomance in terms of reliability are desirable.
Understand that cost is largely small for some parts in the consumer market. We have parts in the range of $0.60. The real challenge is making that product with reliability. It is one reason why the margin in the industry is very low.
>> but the board failed because the components on it had a worst reliability , at the end this redesign increased the costs !
Electronics design is the weakest link in this industry. The worst part determines the overall performance of the product.
>> I had to redesign some testing board and requested some parts with specific characteristics but marketing people many times pushed for a different selection of the parts.
In our firm, the engineers make that call. The marketing team does not have the power to do that.
@ fasmicro: I agree with you, the customer requirement is oriented to integrate more functionalities on the same system so the shrinking is a trend that we will see in the near future
Yes Fasmicro, the cost saving is the goal in this industry, but the R&D team of a company have to guarantee that a device is cheap and reliable at the same time, I won't utilize anytime components coming from the supplier who made the failed parts.
This should be the procedure. The marketing people should at least propose a list of possible choices but the final decision has to be taken by the engineer .
This issue might be solved by utilizing only parts coming from a set of qualified products.
I have been doing repairs on equipment that was last built in 1980's. There are those few times where I have the option to do a complete board redesign. At this point I have to keep the board dimension and interconnect the same as the original. Next is locating parts that can handle the same input and output voltages.
When I get down to the component level, I definitely go for the part that will have a long life. Then cost.
As for performance, practically anything I get now will be better in performance, ESD protections, etc.
Currently I am looking at a system build from scratch. My biggest problem is finding and F8 processor, SRAM, UART, PIO and EPROM. I have the EPROM data in hex. Though not analog, it does present its own set of challenges to redesign.
>> The number of integrated parts on customer applications increases progressively then the space for my boards decreases accordingly.
That is the power of the Integration Nation. The value is created from that increased integration of parts. When you have smaller boards, you have an advantage to enter the consumer market.
>> Currently I am looking at a system build from scratch. My biggest problem is finding and F8 processor, SRAM, UART, PIO and EPROM. I have the EPROM data in he
That should be a hobby I must point out. There seems to be no business that can run on that model. Replacing parts in electronics with the cost of these parts becoming very low is a model that may not be sustainable.
@etna, electronics counterfeiting is the most difficult problem in the ecosystem. The only way to solve that problem is to order direct from the OEM. If you miss the supply chain, you have missed the opportunity.
@goafrit2, Yes I agree with you. This is the real advantage of integrating many functionalities in a single board, the key point is to guarantee the overall performance and reliability of the system and to avoid mutual interference between the blocks
RedDerek
My biggest problem is finding and F8 processor, SRAM, UART, PIO and EPROM
Can't help you with the F8 (I do have the databook), but I may be able to contribute the others- can you be more specific?
@antedeluvian – I have a redesign in process where I have full schematics, and even hex code that is to be burned into the EPROM. Worse case I am look at is to reverse the hex code into a program for the F8. Then use a more modern chip such as Atmel, Microchip type product. Oh, drawing last revision was in mid '80s.
@goafrit2 – sometimes they care – depends on whether the client needs an especially critical function. For example, very low leakage, or high speed with minimal distortion, or minimal skew between two adjacent channels.
@Scott – well, some prices are going up – like with the premium devices from certain well-known companies in Silicon Valley and Boston. But in broad terms, I suppose there is sufficient competion to cause a generally downward trend in prices.
>> the key point is to guarantee the overall performance and reliability of the system and to avoid mutual interference between the blocks
That is a big challenge especially as we go down to the nanometer CMOS process. When that happens, interconnect noise increases and the problems escalate.
>> But in broad terms, I suppose there is sufficient competion to cause a generally downward trend in prices.
Absolutely yes, The industry is experiencing that constant competitive space that is depressing earnings and making it harder for startups to grow and survive. One of the reasons we have fewer success for electronics firms in the industry when compared to the web and apps industry that seem to grow and exit at better rates.
@goafrit2 – good point regarding how difficult it is for startups to succeed. The overhead (costs) for a semiconductor company can be huge. For someone writing apps, overhead is quite low.