Embedded Muse 225 Copyright 2012 TGG May 7, 2012
You may redistribute this newsletter for noncommercial purposes. For commercial use contact email@example.com. To subscribe or unsubscribe go to http://www.ganssle.com/tem-subunsub.html or drop Jack an email at firstname.lastname@example.org.
EDITOR: Jack Ganssle, email@example.com
- Editor's Notes
- Quotes and Thoughts
- How Low Can You Go?
- The End Isn't Near
- Responses to How Hard Are You Working?
- More on The Demise of Print
- Tools and Tips
- Joke for the Week
- About The Embedded Muse
UBM's 2012 Embedded Market Study was recently released. It shows that the number one concern of engineers and managers is hitting schedules. Building a higher quality development process and building a profitable development process are also concerns that are not far behind that of schedules.
In fact it IS possible to accurately schedule a project, meet the deadline, and drastically reduce bugs. Learn how at my Better Firmware Faster class, presented at your facility. See http://www.ganssle.com/onsite.htm .
Quotes and Thoughts
Our developers never release code. Rather, it tends to escape, pillaging the countryside all around. - The Enlightenment Project
How Low Can You Go?
When I was a young feller few ordinary souls had any access to computers. Costing millions and often the size of a building, only governments, corporations and schools could own one.
The microprocessor changed all of that, of course, and no one is surprised to find a typical household with a fair number of PCs, Macs, and the like. Most of us have a pretty-high-performance machine in our pocketed phones.
The embedded revolution put processors in a lot of applications, some of which are so cost sensitive I get vertigo thinking about the big iron model of computing of yore. Reputedly there's a beer glass in Japan that wirelessly signals the wait staff when it starts to get low. Anything to avert a beer emergency.
At the latest Embedded Systems Conference Microchip showed another application: an old-fashioned non-electric razor that's, well, electric and powered by a PIC10F222. Schick's Hydro 5 Power Select has five blades, a power switch, and a PIC that causes the blades to vibrate at one of three user-selected speeds. The dissected one at their booth showed a PCB with just a handful of components, one of which was a six-pin PIC10.
The razor is $7.49 from Amazon, so Microchip must be selling the processors for a dime or two. It's listed at $0.36 on Microchip's web site, but razor-level volumes must drop that more.
A brochure that lists some of the tech details is here: http://ganssle.com/misc/hydrorazor.pdf .
What about your toothbrush? Does it have Bluetooth? This one (http://beamtoothbrush.com/) does. At $50 it's a lot more expensive than the razor, but, really, connectivity in toothbrush? The idea is that it will beam brushing habits to your cell phone. That way, when the phone is inevitably hacked, your dental records will all go on-line. The company feels managing the data will result in better brushing habits.
I wonder if Intel's engineers ever, in their wildest dreams back in the heady 4004 days, imagined processors in such mundane low-cost applications.
Between the toothbrush and razor, better have a bunch of spare AAs around for your morning ablutions.
The nice folks at Microchip provided three of the above-mentioned smart razors for Muse subscribers. So, answer this puzzler for a chance to win a free Hydro 5 Smart Razor: What is your favorite piece of test equipment and why?
I'll present my Better Firmware Faster seminar in Melbourne and Perth, Australia February 20 and 26th. All are invited. More info here. The early registration discount ends January 20.
The End Isn't Near
An article in Network World makes the tired argument that Moore's Law is running out of steam. One cited professor believes that around 5 nm "quantum mechanics suggests that the Heisenberg Uncertainty Principle will result in electron leakage from the chip." That's a nonsensical statement. Electron leakage has always been a characteristic of electronic circuits.
(It's worth reading Moore's 1965 paper about this - see ftp://download.intel.com/museum/Moores_Law/Articles-Press_Releases/Gordon_Moore_1965_Article.pdf. At the time Moore says an IC could profitably employ only 50 transistors, but speculates that that number will grow to the "tens of thousands.")
Ever since we hit the 1 micron node (1000 nm) I've heard people sagely noting that Moore's Law is doomed. At every process geometry it's felt that we've pushed the limits nearly to their breaking point. Yet next month Intel is expected to release Ivy Bridge, which debuts at 22 nm. That's some 50 times smaller - in length - and 2500 times smaller - in area - than the parts that first made pundits shake their heads about the future of Moore's Law.
True, Dennard scaling has failed. Way back in 1974 Dennard predicted that all of the goodies we get with shrinks (like faster transistors, better clock rates, etc.) improve with the scaling factor. Steve Leibson (whose blog is here: http://eda360insider.wordpress.com) has noted that those benefits tapered off at the 90 nm node. But Moore's constant gift of cheap transistors continues unabated, and is now being used to dump ever more processing cores on a die.
Process engineers always surprise us. Remember the gate oxide problems that made 45 nm a former end of Moore's Law? High-k dielectrics rode to the rescue. Or the capacitance issues of tiny gates? Intel's tri-gate technology licked that and made 22 nm possible.
Sure, there are some limits, and nothing can go on exponentially forever. A silicon atom is about 0.2 nm in diameter; gate oxides are sometimes only 1.2 nm, or an unfathomable six atoms thick, and have been stalled at that dimension for nearly a decade.
I see hope ahead. Though two-dimensional shrinks will bang into quantum limits, process engineers are already going 3D. Xilinx has Virtex-7 FPGAs that are considered "2.5D." That is, the parts are composed of four die that all connect to a silicon "interposer," an element that provides 10,000 interconnects between the die at chip-level speeds. Generally, off-chip connections require high-power (and slow) drivers, but the interposer removes that limitation. The entire part has 6.4 billion transistors and 1925 BGA connections to the PCB. And it only needs 20 watts.
Intel has some of the brightest engineers in the industry. I'm dubious at times about some of their embedded-related strategies, but would never underestimate their process genius. The company's "tick-tock" strategy (a two year cycle which nets smaller geometries and better architectures) has been routinely successful. They anticipate hitting 10 nm around 2015. Who knows? Maybe tick-tock will morph to "thick-tock" as they, and the other leaders in semiconductor manufacturing, move towards true three-dimensional designs. Sure, there will be plenty of challenges, and it'll be fun to marvel at the staggering cost of the fabs. But since we have yet to see any flattening of Moore's Law curve, I expect the "more transistors cheaper" mantra will be with us for a long, long time.
Responses to How Hard Are You Working?
Greg Holbrook sent this link (http://www.cnn.com/2012/04/16/tech/web/cashmore-facebook-sandberg/index.html?hpt=hp_c2 ) which is an article about how Facebook's COO Sheryl Sandberg leaves work at 5:30 every day. However, this reasonable-sounding balance is disrupted as she logs on at home to stay in touch.
Niels Malotaux wrote: "It's not about how hard you are working but how smart: focusing on what you really have to do and not doing what later proves to be not needed.
"In the project I'm currently coaching (earth observation satellite) people felt they were overloaded with work and never met the deadlines. Slipping schedules indicated that they would probably not make the launch deadline. Within a few weeks, however, they now feel no overload anymore and meet their deadlines. The stress is gone and all the stress-energy can now be used productively rather than destructively. Because about half of what people do in projects will later prove to have been unneeded, there is a lot of room to do a lot more (the part that IS needed) in a lot less time. And it takes only a few weeks to make this happen, because it's not difficult but rather contra-intuitive, which is why it doesn't it happen automatically. See for the options that do and don't work if we think we have to work harder my booklet#7, chapter 7 (www.malotaux.nl/booklets)."
More on the Demise of Print
I'm still getting responses from readers about how on-line resources are replacing printed material. Print is under assault from another direction: this week a couple of vendors told me that they are getting rid of their written manuals and replacing them with videos. Sometimes there's nothing better than a decent video that demonstrates some technique, but, in general, I feel they are overused. One can flip through a printed manual. It's far faster to read something than to watch a YouTube stream. And how in the world does one search a video?
James Irwin responded to a comment that praised libraries as repositories: "Part of what is missing in the comment is that "free" at a library is based upon previously paid taxes. (Not that I am against it, but lets compare apples to apples.) The broader money base for a library generally means lower costs to individual users.
"I am not sure what it costs to maintain a digital copy of a magazine for a year, or "forever", with enough access bandwidth to make it useful. Even though I grouse at the single copy price I understand it. (I just like easy payment options.)
"In addition, electronic copies have the same problem as music. Photocopying a magazine reduces the quality and copies of copies are even worse. Thus a company can be "assured" that copies do not proliferate. With electronic copies the publisher loses much control and future income. (Not sure how to handle this, but having a link in an electronic copy that I can send to others interested seems to work.)
"Paper lasts for 50 years? True, provided one has humidity controlled storage -- costly, just ask a library.
"I like the electronic magazines. The easier storage and ability to search makes them so much more useful that the hundreds of articles in my file cabinets that can take me hours to go through.
"Anyway, I believe the paper vs electronic is like "Cards" vs "Cubs" -- who knows who is better? Not I."
Tools and Tips
The tools page I maintain (http://www.ganssle.com/tools.htm ) is composed of submissions from Muse readers. Please keep sending your love-it/hate-it recommendations.
I came across a free handbook from TI entitled "Handbook of Operational Amplifier Applications." It's a freebie 94 page .PDF which starts with the basics of op amps and takes the reader through all of the commonly-used applications. Highly recommended, and it's available here: www.ti.com/lit/an/sboa092a/sboa092a.pdf .
The handbook can't cover every possible application, of course. Back in the early 70s, when microprocessors were still expensive, I designed an analog circuit that computed the square root of the sum of the squares of three analog inputs for a colorimeter. By placing transistors in op amp feedback loops one could exploit non-linear semiconductor behavior to compute logarithms, which greatly simplified implementing the math. The Art of Electronics (Horowitz and Hill) shows one way of doing this.
We used 40J op amps from Analog Devices for this and other applications back then. The 40J was about a cubic inch of potted electronics and was notoriously unreliable. On one trip in 1973 I must have replaced 50 of these in eight or ten countries. The customers were all angry, as was I after a month of 40J swaps. Boarding the plane home from Tokyo I wasn't happy to be assigned seat 40J on the 747.
And the seat was broken.
Let me know if you're hiring firmware or embedded designers. No recruiters please, and I reserve the right to edit ads to fit the format and intents of this newsletter. Please keep it to 100 words.
Joke for the Week
Note: These jokes are archived at www.ganssle.com/jokes.htm.
Tom Mignone sent this:
One more programming language entry for the jokes - it was largely used in embedded Air Force projects such as plane avionics and weapons systems. I programmed in this language for the MX 'Peacekeeper' missile system project in the mid 80s.
Jovial: As hard as you try you cannot write a program that makes shooting yourself in the foot with a nuclear device funny.
About The Embedded Muse
The Embedded Muse is a newsletter sent via email by Jack Ganssle. Send complaints, comments, and contributions to me at firstname.lastname@example.org.
The Embedded Muse is supported by The Ganssle Group, whose mission is to help embedded folks get better products to market faster. We offer seminars at your site offering hard-hitting ideas - and action - you can take now to improve firmware quality and decrease development time. Contact us at email@example.com for more information.