The Peak-to-Average Ratio

Here’s something I noticed a while back: how often our problems boil down to optimizing the peak-to-average ratio. This might be flirting with madness, but now I see this dilemma popping up in other aspects of life.

I should explain. Let’s say we’re selecting components for a power supply design. You have nominal input conditions and output requirements. For argument’s sake, let’s say the design typically requires 10 watts. That’s what you want to design for, and if you hit it exactly, then you’re not wasting money by providing capacity that is rarely used. Here’s the problem. Suppose every Thursday when the moon is full, the load demands 14 W. If you design for 10 W and are asked to deliver 14 W, then the power supply is under stress and might fail. An engineer developing a reputation for creating reliable designs would fight the battle and pay the extra price for circuitry that supports 14 W and might even go further. If you know you might sometimes need 14W, you might design to support 15 W to anticipate cases beyond what is expected. Good engineer. Reliable design. The boss says, “You spent $0.15 when you should have spent $0.10. You’re a bad engineer.”

Then you have to account for all the other variables you can capture. What are the environmental extremes? What are the input extremes? What else can go wrong?

To me, the tradeoffs boil down to how much it’s worth to achieve excellence. Every company will tell you it wants to be excellent, but is it mature enough to pay for it? To carry on with the power supply thought experiment, I’ve worked with the biggest semiconductor company in the world, and it drew an interesting conclusion from studying the peak-to-average ratio. It wanted the safety of the 15 W rating but then said: Look at the capability between 10 W and 15 W. Why aren’t we taking advantage of the excess capability? What design strategy gives us more access to capacity we’re paying for but not using? Maybe those guys are smarter than they look.

How does this analysis play out in other areas? For example, in the engineering lab, you might have a piece of equipment that gets used once a year. From a customer perspective, a problem or test requirement pops up: Having the tool on hand and readily available means you don’t have to hunt for it, or beg someone to loan it to you, or rent it, or whatever. There’s an even worse scenario, and that’s when the engineer doesn’t want to face the hassle and skates around the issue. A company should never make it hard to do a good job.

Imagine taking your car to the shop. If it needs a specialized tool to do something, you won’t notice how rarely the tool is used. You don’t care. You’ll notice the quick response to solving your car’s problem. That’s excellence.

I expand this to managing people. You might have an adequate employee who is perfectly capable of handling the normal day-to-day job, but what happens when you’re faced with an extraordinary challenge? To be excellent, underutilize the talent on a day-to-day basis so, when the crazy thing happens once a year, you’re able to expertly deal with it. The MBAs won’t get this, because it looks bad on paper. You could have an employee for $50,000 a year and you’re paying $100,000. Bad manager.

I’m not saying we should throw money away. There is an art to striking a practical balance in the peak-to-average ratio.

Once you notice grappling with the peak-to-average ratio, it pops up everywhere.

How excellent do you want to be? Optimize the peak-to-average ratio to get there.

What are your thoughts? Set me straight in the comment section.

8 comments on “The Peak-to-Average Ratio

  1. aklompe
    January 13, 2015

    I think in the past there was plenty of peak capability to go around in the workplace. Not everybody was able to get the highest formal education, so there where people in the workplace with the brains of a professor doing 'just' simple tasks. If needed they could handle a lot more complicated tasks and didn't need the sometimes ridigulous management structures of today. Also, having a greater capacity than continiously required lowers the stress related fatique. So it improves quality of life in general to not having to stand on your toes all the time.

  2. D Feucht
    January 13, 2015

    Ken, you are expressing an outlkook that appears to be growing among engineers and users of technology alike – an attitude that was just “natural” during the Golden Age of American engineering, when most products were designed to last a long time. Now, planned obsolescence is designed in. That is  horrid for an engineer to have to do, yet many engineering enclaves seem to be caught in the junk spin cycle, quickly turning out new generations of productrs that offer nothing different than before except that the plastic is thinner, the electromechanics corrode faster, the reliability margins are shrunk. It is time to slow down and do designs right.

    In power electronics, one of the important performance factors for circuits is the form factor, the rms to average ratio, which is minimized when optimum. In inverter design, the crest factor, peak to rms, is minimized. Utilization can be defined as the average to peak ratio, and ideally, is made as large as possible, to achieve the most use from a part. In other words, it is a measure of how little the part is oversized. In recent years, utilization has been too zealously maximized at the expense of reliability margins.

  3. kencoffman
    January 14, 2015

    How many corporate mission statements mention excellence as if it happens just becuase we say it without it being necessary to create it or invest in it. From a customer point of view, what is your bandwidth? How ready are you for the next challenge? This is one of the many ways MBAs are destroying the state of our art by mothballing under-utilized equipment, streamlining the workforce or moving workfunctions to regions where the hourly cost of labor is less. I feel like I'm William Wallace and we all know how the Braveheart movie ends. 

  4. cpeguy
    January 14, 2015

    Thank you for this important post.  All I can say is “Amen, brother!”  A significant amount of my time is spent fighting the forces atevery turn that want to cut corners (i.e., cost).  I also noted recently some problems with some Agilent equipment.  They advertised some options that have not yet been fully implemented in software.  IMHO, that would have never happened with Bill and Dave around.  Sigh…

  5. chirshadblog
    January 15, 2015

    @cpeguy: Well I do not think it would have changed as such. Anyway now we have to make things up for the loss that we are facing at present. 

  6. fasmicro
    January 15, 2015

    >> This is one of the many ways MBAs are destroying the state of our art by mothballing under-utilized equipment, streamlining the workforce or moving workfunctions to regions where the hourly cost of labor is less.

    The deep problem is that they do the bad work and yet they get rewarded. The highest paid in the world and the most destructive. You cannot make it up. Boards always look for MBA degree before they move people up to top leadership.

  7. goafrit2
    January 15, 2015

    They advertised some options that have not yet been fully implemented in software.

    Those are called marketing feaures. Some are features you may never use but some have asked about them. The best strategy is to bundle them in the design. If you ask hard, you can get it but at extra pay. But that does not mean they cannot advertise the features.

  8. Vishal Prajapati
    January 20, 2015

    You noticed it right. Almost all the companies have mission statements as being excellent in their respective field and giv e the best possible customer service. But when it comes to providing budget to be ready to be excellent, all the excellent minds goes down the drain. They can never think such a samll thing that you need to be preplan to be excellent.

Leave a Reply

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