Advertisement

Blog

Analog 1 Percenters: Stop Reinventing the Wheel

Everything in life happens on the order of 1 percent, give or take. Pick anything you can think of and I'll bet that at least 1 percent of the world is experiencing it at this moment. Even in the case of analog design: Does an analog designer really believe their circuit design requirements are so unique that no acceptable alternative exists?

The truth is no one (or no design) on the planet is so unique that they are alone in their uniqueness. The six particle physicists — including the lucky one named Higgs — who in 1964 simultaneously postulated how particles acquire mass, know this all too well.

Having participated in the design of analog circuits for over 30 years with design teams on every continent except Antarctica, I have a hard time accepting a claim that only one acceptable design exists for any analog circuit application. There are a lot of smart analog engineers, all working with the same analog ICs and all solving pretty much the same set of problems. Maybe it's time to start knowledge sharing at the analog system level, so that analog IC companies can provide solutions everyone needs — rather than having them guess what they think everyone will buy.

Let's consider the example of low noise amplifier design. For the moment, forget whether it is RF low noise or sensor low noise.

To apply my 1 percent theory, I used Google Ad Words to check the number of times anyone searched the phrase “low noise amplifier design”. The answer is 880 times per month. With less than 10 percent of the world speaking English as a native language, the real number of people searching is probably at least another factor of two. And then the expert analog types probably don't search for things they already know how to do, so we need to add yet another factor. Summarily, I'm going to go out on a limb here and claim that there are probably 20,000 or more people per year contemplating the design of a low noise amplifier circuit.

Further applying my 1 percent theory means that about 200 of the 20,000 engineers are researching exactly the same type of low noise amplifier design — say, sensor low noise — within six months of each other. And another 100 are perhaps researching 5GHz RF low noise amplifier design. And so on.

We could even go 1 percent further yet and say that two of those 200 people are feverishly drafting a patent application right now to protect the same circuit design they independently believe embodies a unique level of genius not seen since Einstein — who, interestingly, was searched 73 million times over the last year; about 1 percent of the 7 billion people on the planet.

Okay. So what's my point?

Simply that it makes no sense for the analog industry to continue down a path of reinventing the same, new, analog “wheel” 200 times per year using different subsets of the same 250,000+ unique analog ICs available. Especially considering that the combined volumes of any 200 applications most certainly render a cheaper and better-integrated solution as viable.

Now, before anyone starts accusing me of leading a chorus of “Kumbaya”, let me share an interesting transition I see going on in the world of integrated analog using a couple of examples.

For awhile now, Apple Computer has been actively recruiting analog IC designers to join their company, who, upon starting, will never design another IC again. Rather, they are hired to architect the IC system design that is then handed down to what most would call an analog IC standard product company, but in effect they are an IC service bureau: hired transistor-pushers.

And then there is China, once the bastion of assembly for all things 2 cents or less. Not true anymore. They don't want to fool around anymore with external 1 cent transistors and 2 cent op amps. It's too much trouble trying to chase down manufacturers competing for volume with other OEMs while waving a measly $10,000 purchase order.

The future of analog design is not an FPAA or a metal-programmed component array or some reconfigurable analog IC. Those ideas have been languishing around for decades. Manufacturers are moving from amplifiers and LDOs to systems. And the race is to find who has the best system level-defined ICs in their portfolio.

The time has come for PCB analog designers to start thinking in terms of an IC system. Architect the optimum IC system solution, patent that system, and then go get the hired transistor-pushers to execute your vision. Even if your volume is low, put together the business case to show how the worldwide usage is much larger. Offer to share your system knowledge for a price break and first adopter strategy as leverage against the other adopters in the world. You get the best all around: a fully integrated and optimized analog IC system with a significant price advantage and you're out there first. And you still had lots of fun.

Related posts:

17 comments on “Analog 1 Percenters: Stop Reinventing the Wheel

  1. Steve Taranovich
    May 29, 2013

    Scott, you are spot on about the world of analog designers working on the same solution to an existing problem. This is especially true in 2013 with communications pervading our entire being. We not only search the Internet to see what we can find to help solve our design problems, but much information comes to us in an unsolicited manner and the analog world finds out about a problem in nanoseconds. Of the many solutions, some will converge on a common design and some will remain quite unique, but the problem is solve faster than it could have been even ten years ago

  2. Scott Elder
    May 29, 2013

    Steve, I think it would be great if there were a Google schematic search option.  Or schematic compare rather.

    One could submit their schematic as the “search terms” and get back all of the existing schematics that are close matches.  Think of the learning.

    The search results would either return things that look substantially like yours or very few that are not even close.  In the later case, you'd better be able to explain what the rest of the world doesn't understand before you build your circuit and see why your design is in the minority!

    Just seems like there would be so much more learning possible.  It is always important to have metrics to measure performance.  Google schematic search would be a great feedback.

     

  3. Steve Taranovich
    May 29, 2013

    I love the idea! Perhaps I can pitch the thought to some schematic capture people I know. With modern software, I would not think that a compare function would be impossible

  4. Brad Albing
    May 29, 2013

    The tough part (and I know you actually know this already) would be to develop a way to describe the ckt/schematic so that the search engine would be able to find highly similar, somewhat similar, or not similar – as delineated by your search parameters. Way beyond my ken. And my barbie.

  5. Scott Elder
    May 29, 2013

    @Brad – There are already standardized interchange format languages for schematics (i.e. EDIF) that could be used.  And then like google does for words…you queue off the distance, in this case electrically, from component type a to type b.  Easy to write in words, but I don't have enough passion in software to do the coding.  But I'd sure love to play with a tool like that.

     

  6. DEREK.KOONCE
    May 30, 2013

    Great idea on a circuit search. My only concern would be from an education point of view as to how many students would end up plagurizing the circuit to satisfy an electronic class problem. In a way, wish stuff like this was available 25 years ago to help me solve some of my engineering problems.

  7. Scott Elder
    May 30, 2013

    @Derek — Isn't your concern the same as with any intellectual property sharing on the internet?  Students copy thesis.  Book reports.  Movies.  Music….  I think privacy went out the door on Dec 31, 1999.

  8. Brad Albing
    May 30, 2013

    @Scott – I agree – students boosting IP is an ongoing problem – but that's a separate issue from making a searchable schematic database. That's another blog for another time.

  9. Davidled
    May 30, 2013

    I get more concern for engineering education. Student might think and analyze each circuit based on circuit theory. If student copy from the exiting circuit, in the end, there is no creation from any type system. One of beauty for engineer is to create their own pieces, not copy it.

  10. Scott Elder
    May 30, 2013

    @DaeJ — I don't think one can force creativity.  Either one is creative or not.  And every student/inventor in the world (me, you, Brad, everybody) has their inventions based upon those who came before them.  Thank goodness they all were willing to share their knowledge so that I could learn.

    “If I have seen further, it is because I stood on the shoulders of giants.”  Isaac Newton.  Openly sharing knowledge is a good thing.  And I accept the risks that come along with that.

     

  11. Netcrawl
    May 30, 2013

    @Scott I agree with you, brainstorming that exactly the keyword here-sharing ideas, the reality is it takes many heads to design a better product, and sometime these team of designers (often with different skills and methods used) tackle the complexity of today's electronic products, often without the right resources and tools.

  12. Netcrawl
    May 30, 2013

    @Scott you're right IP is a huge issue, and need to be address, we're not making enough innovation here, not just individual but companies also are copying works from other well-established players.

  13. BrianBailey
    May 30, 2013

    Well, 25 years ago, I had to memorize my multiplication tables, or buy a book of logarithms and other trig functions. Thank heavens we don't need to do that anymore because we have better ways to do it. There are also many things that I was taught as a digital designer, such as solving Karnaugh maps that I dont have to do anymore. Should every design engineer be able to derive a circuit from first principles – should every digital designer know how to construct an AND gate?

    If the IP no longer hold real value, then I say plagurize away and solve the tough problems.

  14. Scott Elder
    May 30, 2013

    Hi Brian — “Should every design engineer be able to derive a circuit from first principles – should every digital designer know how to construct an AND gate?”

    The answer is NO.  I think every generation of engineers needs the majority to move one more level higher in the food chain.  One level higher in abstraction.  Logic gates becomes RTL becomes System Verilog becomes System C.  Next stop–completely GUI based design.  Simulink becomes Chip.

    If electrical engineering requires that one person have a grasp of every topic from an IV curve to Ultrasound image processing, then the whole innovative process will eventually come to an end–information overload.  Jack of all trades–Master of none.

    Schools today don't teach board level analog design.  They teach IC design.  And that's what the students want to study.  In the digital domain they teach via an HDL–as you well know.

    I think it is actually cheaper to study analog IC design in school rather than analog board design.  The latter requires spending money for parts, test equipment, and boards since simulation models for high level analog ICs are not very accurate.  But for transistors on Cadence SPICE [which is a free license at school], they are darn near perfect.  A student will see very very precisely how the op amp will respond to a capacitive load.

     

     

  15. Netcrawl
    May 31, 2013

    @Scott I agree with you, you a point here, today's fierce competition demand that every company must develop complex prodcut or innovate in the most cost-effective way, this means using resources in part of the world, different skills and methods, this also means a whole new generation of engineers well-trained and well-equipped, placing the right person in the job.

  16. Netcrawl
    May 31, 2013

    @derek people tend to solve problem by dividing it into manageable sizes ( this is the most effective way and the fastest) each engineer is concerned with different parts of the problems, and they work and communicate to create a solution. This is exactly how we should solve the problem. This is a good pattern for a group of student trying to learn more.  

  17. RedDerek
    May 31, 2013

    Teaching people HOW to break down problems is the key to many challenges. Thus, I believe in teaching people the building blocks. Sort of like a mechanic has his screwdrivers, wrenches, etc, or a carpenter has his saws, hammers, screws… As electronic engineers, we have our tool box as well: resistors, caps, opamps, etc. The key is to know them well and how to put the puzzle pieces together. This is the forward approach.

    When one has a design problem, they need to think backwards and break it down until they are at a level in which the tools can be applied.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.