For those of you that don't know, Bob Pease, author of Analog Circuits and Troubleshooting Analog Circuits , is an icon in the analog design world and his Pease Porridge section in Electronic Design Magazine has been going on for longer than I've been an engineer, and possibly longer than I've been alive. Well, this recession claimed another victim and he got laid off from National Semi. It's understandable if National were going bankrupt, but they just bought another company. They could at least have given him a nice emeritus title and let him keep his National Semi email address instead of simply handing him a pink slip. What a stupid decision by corporate management to lay off a world renowned, highly-respected, analog guru at an analog/mixed-signal company.

I just felt I needed to rant about that. It really bugs me...

One of the things that was taking up my time last week was working on an article for Product Design Magazine. It's kind of fun to actually write for a magazine, especially one that I used to read when I was in the states. This one was completely off-topic from wireless sensor networks and I wrote it in a fit of rage after hearing about all the engineering jobs that were disappearing in the US.

I actually thought that this article wouldn't get accepted by the editor when I submitted it. But it ended up getting published and it turns out that the feedback was pretty good. I was actually touched by some of the stories I received from the readers. Seems like I'm not the only one that's unhappy with the way corporate America treats its employees.

Here's the link if you want to check it out:

Link

p.s. You might notice that I managed to squeeze in a shoutout to the Contiki project. 

...Contiki in tha house...woot...woot...

Well, it's been about a year since the site started and also since I've embarked on this open source project. Let me just say that it's been quite an experience. In this short year, I've learned many things…about Zigbee, wireless sensor networking, the WSN community, programming, business, and myself. So I thought I would put together a little review…the year in Zigbee…

One of the main things I've learned is that I was naïve to think that I'd be able to finish this project in under a year. Way back when, I was poking fun at a company that took 18 months to write a Zigbee stack and then finally gave up. I can completely understand what they went through, and the hubris that caused me to think I was better than that. Programming is just a small part of writing a Zigbee stack. The real time sink is the learning curve. I grossly underestimated how long it would take me to understand the specification from top to bottom and that, I would have to say, took the majority of my time. 

The one saving grace is that I made a good decision in choosing the Contiki OS for the framework . It probably saved me a couple of months of debugging and has all the facilities that I required in writing the stack. I still haven't achieved a mastery of it, but I do have a good grasp of the parts that I need and it keeps on coming to my rescue whenever I get into a hairy situation with the stack. It's also one of the reasons I'm able to keep the code size under control.

Also, my understanding of the Zigbee spec right now is quite advanced, if I may toot my little horn. There are a lot of subtle complexities which I understand now, and for all its bloated obesity, the Zigbee spec does a fairly decent job of covering everything needed for interoperability. Most people just think of Zigbee as a halfway decent mesh routing mechanism, but that's just a very small part of the picture. The real complexity lies in the fact that along with the routing, you need to handle device discovery, service discovery, network management, and also specify device behavior for interoperability. And as much as I bitch and moan about the spec writers being crazy, unrealistic, ivory-tower robots who don't have any understanding of the fragile emotional state of a programmer, I don't think there's currently a better solution that can handle all of the requirements for device interoperability. Actually, if Zigbee with its interoperability and 6LoWPAN with its TCP/IP infrastructure could somehow get together, then I really think that it has the potential to redefine how people think of the internet.

Some of the other things I learned by starting this project is that a website takes a lot of time to maintain! When I first started the site, I thought that I could just blog about whatever strange topics would come into my head, but I quickly found out that I just didn't have that many opinions…or thoughts. At least not enough to provide constant blog posts. I really admire people that can blog every day or at least a couple times a week. I don't seem to have the brain capacity to generate a lot of content, mostly because my life is pretty boring. I pretty much stay home almost all the time and either write code, spend time with my family, or service the side job.

Luckily, I found a way to keep the site somewhat interesting and it was kind of by accident. I was trying to stay on top of all the developments going on in the WSN industry, especially around nine months ago when it seemed like a new protocol was coming out every week. I set up a bunch of RSS newsfeeds (the list is a monstrosity) and would monitor it for any news that came up. One day, I decided that I would just start posting the news on the site as well. At first, I tried to post the news along with my opinion about it, but I quickly ran out of opinions. So I ended up just posting the news and a link to it . The news eventually overran my front page so I ended up creating a separate section for it and adding its own feed. That turned into the most popular part of my site.

The strange thing was that at the time I was doing the newsfeeds, I needed so many sources because there wasn't a whole lot of news. There was the occasional media frenzy when someone would announce a new protocol, but after that, it would go quiet again. I think I would be lucky if I could get one piece of news every two or three days. Now, there's so much news about the WSN industry that it's a struggle keeping up with it, especially with all the green talk, energy efficiency, energy harvesting, and the Obama administration.

One other thing that kind of surprised me was that living life as an open source developer doesn't mean taking a vow of poverty. I was originally preparing for the worst…I mean cutting all my expenses, loading up on instant ramen, trying to only shower when needed (that was actually by preference, but my wife objected), but things didn't turn out too bad. After I had originally quit the company I was working for to do the project full time, I ended up working out a deal with them to stay on for two days a week at 40% of the salary. That was enough to cover the rent and bills and give a little something to the wife. As the project got more known, I started getting consulting and job requests…actually, quite a few of them. I ended up having to turn them down because I needed all my time just to work on the stack. I finally ended up taking an offer from a local disti to help them out a couple hours a month. This was when I had the grand plan for the disti makeover . That turned out to be a disaster. The problem was that the huge recession actually hit the Japanese consumer electronics industry really hard, and the company went into crisis mode. There just wasn't much time to implement the things that I had suggested since they were fighting for survival. I ended up leaving the company, mostly to save them some money and I help them out for free from time to time.

Ahhh, but back to the poverty myth. So the interesting thing was that I had set up a corporation last July , on the recommendation of the Software Freedom Law Center . Apparently, even open source authors have liability risks from free software, although I don't know who would want to sue an open source developer. They'd be last on my list of juicy targets to squeeze a couple million out of. After I had set up the corporation, I did most of my spending through it, ie: the corp leases office space (my room) from me at standard Tokyo office rates (which happens to be the cost of my rent), dinners are paid with the corporate credit card (as long as you keep a record and the dinner discusses something related to business ie: I brag to my wife that I got three new visitors that day, then it's legit), and equipment purchases are also on the card. Also, my income from the side work goes into the company as revenue. It turns out that since corporations are taxed after expenses, as opposed to employees who are taxed before expenses, even though I'm made roughly 50% of my original salary, my spending power was almost exactly the same. I actually eked out a small profit last year. It was something like $700 and I actually need to pay $75 tax on it. Don't laugh, that puts me ahead of 95% of the tech industry. Strange…the inner workings of finance…

As I mentioned, even just one year doing the project has taught me a lot about the industry, the spec, programming, and plus I got to meet a lot of really nice people. I think this past year has been my preparation period to get everything set up. This next year, I'm really hoping to turn the software into real, working code. I'm really anxious to see the finished product and I think that an open-source Zigbee stack can lead to a lot of innovative designs that have a positive impact. So this year, I'm focusing on making the transition from vaporware into reality…which leads me to the next post…

In a morbid sort of way, the global recession and the US financial meltdown is probably the best thing that ever happened to the wireless sensor network community. The dire financial straits that the US got into forced the government to implement an economic stimulus package which is targeting key instrastructure industries, one of which is energy. One of the areas in the energy industry that is going to be improved is the US electrical grid which will be upgraded to a "smart grid" that can handle renewable energy, thus lessening the dependence on foreign energy sources and carbon-producing coal generators. Since renewable energy is notoriously spiky, the smart grid will need to implement demand response to help smooth out supply and demand issues on the grid. Implementing demand response requires that electrical meters be upgraded to "smart meters" that can communicate bi-directionally with demand response devices. And of course, one of the communication links to the demand response devices will be wireless, and most likely based on 802.15.4.

So in a roundabout, chain-of-events, six-degrees-of-separation, butterfly-flapping-its-wing sort of way, the global financial crisis is probably the best thing that could have happened to 802.15.4 because now, there is a real, solid, high-volume application for it. Regardless of which protocol wins the meter sockets, all 802.15.4-based protocols stand to benefit from this turn of events. Before the financial crisis and all the hype about smart meters, 802.15.4 was pretty much just in the realm of academia and model rocketry circles

The fact that there is a solid application for 802.15.4 can't be emphasized enough. Bluetooth was a dying protocol until some cell phone company decided that it would be kind of cool to try and make a headset connect to their phone wirelessly. Getting rid of that little black wire that hung out of your ear launched Bluetooth firmly on to the wireless map, and validated the business models of all the struggling Bluetooth companies (at least the ones that survived). It also ushered in a new era of people that look like they're talking to themselves.

However the more important thing is that having a solid, high-volume application causes the price of chips to drop like a rock. As volume ramps, the incremental cost of a chip decreases, and for smart meter applications (assuming limited government bribery), the cost savings will be used to drive the chip prices lower to compete for the design wins. Although it's bad for the vendors, it's great for us!

To continue the Bluetooth example…in 2000, Cambridge Silicon Radio's Bluecore 1 chipset cost about $8.50 in quantities of 1 million . In 2002, CSR's Bluecore 2 chipset cost about $6. And the price pretty much hovered around $5 until around 2004 when Bluetooth enabled cell phones started getting popular. Now, you can buy a Bluetooth chip in 1000-piece quantities from DigiKey for about $2. That means that the real price companies are paying for Bluetooth radios is probably close to $1.

Now that 802.15.4 seems to have found it's one solid, volume application, it looks like the same pattern might happen. Currently, 802.15.4 radios can be found on DigiKey for about $2.50 in 100 piece quantities. If the smart grid upgrade and the smart metering programs work out as they're hyped to be, then it can easily drive the cost of 802.15.4 radios well under $1. In fact, if volumes are high enough and sustained, I think we can see $0.50 radios in a few years. At least with the help of some industrious, fabless semiconductor houses in Taiwan.

So in a strange and twisty way, the wireless sensor network community should thank the global financial crisis because finally, we can have a low cost radio that will make wireless sensor networks very cheap to implement and the whole world can benefit from their efficiency improvements...and I can finally make a cheap, wireless dog feeder and become a millionaire...

I got an email a couple months ago from the editor at Product Design and Development Magazine to discuss 802.15.4 and its applications a bit. Basically, he wanted to know what impact 802.15.4 would have in 300 words or less. It was kind of tough because there's so much to talk about with protocols like Zigbee, 6LoWPAN, ISA100/WHART, and RF4CE. So my first draft was about three pages written in a serious, engineer voice. I scrapped that and left out all the protocols since I figured that people didn't want to hear a lot of technical jargon specific to the industry.

My second draft was shorter, but still had a serious, professor tone and felt like I was trying to teach people about 802.15.4. I thought that it was kind of arrogant and stupid since I'm not really an authority on 802.15.4 (I just know the Zigbee part) so I scrapped that draft too. Finally, I said fuck it and wrote it like a normal post on this blog. I also threw up some links to projects that I find interesting and kind of weird, since I figured that the usual applications (smart energy, remote controls...) would be pretty well represented by the others. If the topic was about 802.15.4 and the impact it would have on product features, what better way to demonstrate this than by linking to some wireless drink coasters? That's a new product feature right there...

Anyways, so you can check out my little blurb at the magazine . I had initially wanted to send a picture of me and my dog dressed up for Halloween, but my wife told me not to be stupid. So I threw on a coat and scarf and took a picture of myself in front of some of my lab equipment. Here's a secret...I was only wearing boxers from the waist down and my hair was kind of dirty and weird-looking so I ended up wearing a cap. Hmmm...didn't turn out too bad...

FreakLabs

Link

p.s. The 'do on the guy from Libelium is pretty cool...

Man, I always knew the US was pretty rough at the hiring/firing dance, and the Japanese companies seem like noobs in comparison. If you look at CNET's Tech Layoff scorecard, you can see the US tech companies shedding people here and there, but the Japanese companies are shedding people in droves...massive layoffs that are going to seriously hurt the domestic industry and stop Japanese kids from even contemplating any type of career in engineering. It's going to take a long time for the engineering industry to recover from such a bad image, if it ever does. Sad... :(

Sony
12/09/2008
16,000
NEC01/30/200920,000
Panasonic02/04/200915,000
Pioneer02/12/200910,000

 

This is my second week in California and it has been much more hectic than I expected. The first couple of days was pretty relaxing. Hung out with my sister and her husband, met up with some companies in the Bay area, and just kind of kicked back.

But I got a horrible phone call late in the week and found out that my dad went to the hospital with heart problems. I immediately went to Southern California to get a handle on the situation. Since then, I pretty much alternated between the hospital lobby and going to my dad's house to take care of his two pit bulls. Finally, my dad had the heart surgery and everything went well. He just got out today and is happily back at his house playing with his pit bulls.

This experience taught me a lot of things. The first one is that I have a lot of respect for doctors and nurses, because they have to go through that kind of thing everyday and spend their lives surrounded by extreme emotions, sadness, and death. Also, I want to use the work I've been doing with the wireless software to help other people. I can't save people with my hands like the doctors can, but I think wireless sensor networks have a lot of potential to help people and improve their living conditions. I'm also looking forward to the upcoming personal home and healthcare (PHHC) profile coming from the Zigbee Alliance.

Wow...what a trip to the US!

Getting ready to board a plane to California today. It's going to be a nice break from the cold winter weather in Tokyo. It's been freezing lately and it's getting tough to even walk the dog, but I just checked the weather in Cali and it's like 27 deg C in the daytime. That's like mid-80's in Fahrenheit! So I'm packing my sunglasses, sandals, and JTAG emulator and getting ready to work on my tan. I'm going to be up in Berkeley visiting my sis and some friends until Friday (1/23) and then heading down to LA to see my parents. Wow...gonna be interesting conversing in English again. Talking to my dog doesn't count...

'Til then, California here I come...right back where I started from...doo doo doo...

One of the things that really irritates me is the difficulty I have obtaining 802.15.4 transceivers. The problem is that since I'm in Japan, the only access I have to prototype quantities of any chips that aren't sold in Akihabara is through DigiKey. There are no distis that cater to individuals here since they will only service companies that with a minimum order quantity of something like 1000 pieces.

The problem occurs because DigiKey is a US-based company and all inventory ships to Japan out of the US. That means that they can't ship any chips with export restrictions on it, ie: 802.15.4 chips. This also applies to Bluetooth and any other wireless radios that include encryption on it. The problem is so bad that one of the reasons I'm making a trip to the US is so that I can stock up on radios. 

Well, I'm hoping that those days are behind me. Recently, meaning as of October 3, 2008, the US Bureau of Industry and Security simplified and loosened the items that are bound by export restrictions . Here are the highlights as it applies to 802.15.4 chips:

Encryption Simplification Rule of October 3, 2008 (73 FR 57495)

Summary of Amendments to the Export Administration Regulations

...

Defines wireless “personal area networks” and “ancillary cryptography” and excludes these items from review and reporting requirements.

...

License Exception ENC (Section 740.17)

New paragraph (a) - No Notification or Review Required

  • Short-range wireless items not controlled under Cat. 5 (§§ 740.17(b)(4)(i) and 742.15(b)(3)(ii))
    • =100 meter range
    • Examples: IEEE 802.11 and 802.15.1
    • May self classify under 5x002 or 5x992 as appropriate

 ...

§ 740.17 (b)(4): Items excluded from review requirements

  • Short-range wireless encryption functions
  • Foreign products developed with US-origin encryption source code, components or toolkits
  • Wireless “personal area network” items
  • “Ancillary cryptography”

I tried to buy some radios around October/November but DigiKey still wouldn't sell them so I'm not sure if the info has filtered down to them yet. I'm already going to be in the US next week so I won't have a chance to test this out until after I get back to Japan and order some parts. 

The reason I'm posting this is because people that live outside of Europe (Farnell) and the US (DigiKey, Mouser) have the same problems as I do of trying to get radios. I get a lot of emails from guys in India and Korea who want to buy radios off of me in quantities of 10 or so since the distis there also have minimum order quantities. I'm pretty sure its the same for any other country that doesn't have a big audience of home electronics enthusiasts.

Anyways, I thought it would be interesting to share this bit of news and let people know that it may soon be possible to order the chips from companies like DigiKey. 

Happy New Year to everyone!

Wow, it was a complete news drought for the past week and a half, since pretty much the whole tech industry shut down for the end of the year. I didn't realize how addicted to WSN news I've become. The news feed I put up looks like it mostly benefits me since it keeps me on top of things in the industry. People I meet are usually surprised that I know what's going on, especially because I look like a bratty geek and I'm mostly isolated in my room in Japan with no friends to speak of out here. Thank God for the Japanese government's initiatives to bring high speed internet to the home. My life would be a lot more boring without a 100 Mbps connection to Youtube. 

Since all the companies were shut down during Christmas/New Year's, I decided that I would take some time off too. I figured it wouldn't be very fair to myself if I was the only one working so I took a break and studied up on my RF hardware design. RF design is one of those things that always vexed me. Its filled with all these crazy math formulas and bizarre HAM radio terminology like "coplanar waveguide" and "LC balun" and the only people that really seem to know about it are the the guys with long, white beards and trucker caps that operate their own shortwave radio stations and go by the name W3ZX.

So I decided to "slay the dragon" or "swallow the frog" or whatever  the metaphor is for addressing your fears and spent the extra time reading RF texts, appnotes from every major 802.15.4 vendor, and researching online. Now I can say...hey, it's not that bad. Those of us in the 802.15.4 community that don't really need to design our own RF chip have it pretty easy. If you just want to integrate a chip on to a board, then you pretty much just need to know about transmission lines (there's a ton of online calculators), impedance matching, and possibly some basic balun and filter design. Of course that's the distilled info from about 10 days of researching and pulling my hair on the subject. 

Anyways, the main goal is to free myself from making exact copies of the vendor reference boards so that I can develop custom hardware that fit the apps that I want to make. Hopefully, I can also sell some dev boards or modules in the future to make some side cash so I can stop eating the weekly stew. The weekly stew is the big-ass pot of stew I make once a week and eat for about five days straight. It's great for saving money, but not recommended if you're a foodie or gourmand.

Since I'm going on about the future, here's my main goals for this year:

1) Get the stack certified - Instead of saying I'm going to finish the stack, I pretty much believe that there's no such thing. There's always profiles to add or features to implement. Hence, I'll say that I want to get it certified. It's much more clear that way.

2) Make development boards - This is mostly for myself, but it will probably make the stack easier to evaluate. I'm currently using the Atmel Raven boards which are feature packed and low cost, but it's a pain in the butt to customize. I also want to be able to span different combinations of MCUs and radios since that's the main benefit of having an open source stack.

3) A bunch of instructional posts - Okay, "a bunch" is not a very clear goal but I do want to finish my detailed Zigbee tutorial series, some instructionals on RF hardware design (based on the mistakes I'm planning to make doing my own boards), and finally start up the rapid prototyping section that I've been wanting to implement for a while.

Hmmm...I guess there's a bunch of other things I want to do this year as well (ie: survive), but  they don't really come to mind at the moment. Hell, those three will keep me occupied for at least a couple of months anyways.

Nice to be back with a clean slate for 2009. Hope the world doesn't implode...

Ciao...

By the way, I'll be in Berkeley from 1/19 to 1/23 and Orange County from 1/23 to 2/1. Throw me an email if you're interested to meet up.

Merry Christmas and Happy New Year!!!

Hope everybody has a good rest this holiday season. I'm sure everyone needs it...In the meantime, I'm going to try and sculpt a Christmas turkey out of rice and seaweed...