The perils of figure of merit based circuit selection
The fact that you’re reading this has sparked interest and I am sure you are wondering what this means or you have picked a position to either defend your beliefs or want a different opinion. So, let me say that this topic and what I say applies to choices of circuit blocks and or architectures for many different applications from analog to digital. There are Figures Of Merit (FoM) for different blocks and architectures and each have distinct criteria. However, since I am an analog designer, I will focus specifically on the idea of figures of merit for analog blocks and more specifically FoM for ADC’s. There are many figures of merit out there for ADC’s, but most of the FOM involve the ratio of power, effective number of bits, and bandwidth. As shown in Equation 1 is a typical FoM used for ADC’s :
As shown in equation #1, the criteria for defining the quality of an ADC is only based on power consumed, number of bits, and effective bandwidth. These criteria do create a uniform way to measure the ADC but are insufficient for use in industry. Furthermore, if this is the only equation, a low FoM as shown in Equation #1 can mislead and cause the designer to make the wrong choice.
Let me explain. As I have said in other articles I have written in the past, when designing circuit blocks it is important the block or architecture works in the customer’s application or you cannot count your design a success. When designing on the newer nanometric technology nodes, the cost of fabrication including mask and wafer costs skyrocket on the order of millions of dollars. Making a wrong choice has a large impact to the design’s bottom line of financial success. So, what selection method is best and how does FOM fit in the selection process.
When choosing an ADC or ADC architecture for use in your design block, or if you are trying to measure success, FoM is one of the last criteria that should be used to make your choices.
Next: So what else should be considered?
The criteria for choosing should be:
1 Will the ADC work in the application? Do the specifications meet the customer’s criteria for a successful part?
2 Is the design of the ADC robust? In other words does the design follow your company’s and manufacturing site’s rules for robust and variation-tolerant designs? Did you implement DFM rules when designing the ADC?
3 Is the design cheap in terms of area and power? This is extremely important for the nanometer technologies where the cost of manufacturing is high. If the design works flawlessly but takes up too much area then the design will not have the margins needed for sustainable manufacturing.
4 Does the ADC meet the functionality requirements of the application where it is being used? If the ADC does not have needed features or meet specific specifications then the choice of the ADC is wrong.
5 Does the ADC have a decent FoM? It is important that the ADC is efficient and meets performance criteria.
Not shown in the list but equally important is the idea that the proper choice of the ADC architecture or circuit design can provide important performance features or reduces the requirements for other blocks. Such a criteria is a plus not comprehended in any FoM. For instance, the choice of a continuous-time ADC for the front end of a design provides built in anti-aliasing protection that removes the need for filtering in front of the ADC.
Therefore, the use of FoM in my opinion should change and include other criteria as part of the equation to judge performance.
Next: A new FoM
Might I suggest that a new FoM for ADC’s should include at least some criteria for design for manufacturability and or the inclusion of criteria for the area that the ADC uses.
In conclusion, let me restate as I have shown in the list, the use of FoM is an element of the selection process but clearly not the first on the list and in some cases not the most important. The use of FoM is great for use in academia for gauging one ADC’s performance relative to another but should not be the first element of a litmus test for selecting an architecture of design for producing products. Furthermore, if companies and designers continue to use FoM for selection criteria, I would suggest that the FoM change to include DFM and or area as part of a new FoM.
Does your company use FoM as criteria for selecting blocks for use in products?
Do you believe that FoM should be changed for use in selection?
Brandt Braswell is a Distinguished Member of the Technical Staff at Freescale Semiconductor and focuses on the development of data converters, with an emphasis on delta-sigma conversion.
This article first appeared on EE Times’ Planet Analog website.
Related links and articles:
Analog design and the age of digital correction
Is first-pass success possible for silicon IP? Part 1
Is first-pass success possible for silicon IP? Part 2
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
