1dB Designs | Cypress Semiconductor
April 08, 2011
I hate them! These are designs that are hacked together and made to work on the bench without a fundamental understanding of the real problem. They don’t have much margin built in and will tend to fail if any parameter is changed. I find that just because someone has a schematic it doesn’t mean he really knows how it works. He may only know how to get it to work. This will become apparent after the first two or three questions. (A good clue is when I ask a designer how they determined some particular value. If they answer “its tricky.” this means he tinkered with it on the bench until it worked.) Someone originally knew how it worked when he designed it. After that it went from designer to designer each using some fuzzy rules of thumb and a lot of bench work. Each time the design became less robust. Eventually a time will come to develop a whole new architecture. Since 1dB designers don’t know how it works they cannot do analysis on it and do not know how well they have fixed it or even if they really have fixed it at all. Just because someone can run the tools, it doesn’t mean they have done anything useful with the data. I was asked to look at an error analysis that had 38 variables in it. 18 were second order and 19 were third order at best. Taking time to understand the error analysis allowed me to find a simple solution. 1 dB designs depend on fixing the symptom, not fixing the problem. I had one customer that was having problem with their ADC, I took a look and determined that a weak power supply injected noise into the ADC. Their Engineer had all kinds of code to attempt to compensate for this problem. I suggested that they spend an extra 3 cents on their power supply.
When I am out showing off PSoC I will have customers draw out a schematic and ask if PSoC can do that. A schematic is a solution, not a definition. I end up asking just what are they trying to do. When they tell us We then figure out how to do it the PSoC way. Chances the 1dB designer is going to resist these new ideas until they actually see them work. Six months later they will think is was their idea.