Lab Notebook: Two Digit Floating Point Static LED Display

This "Lab Book" posting is my thinking about a design problem I've been considering.

Problem:  I want a way to display a time interval that can be photographed as part of a test fixture.  I'm trying to convey a time value, which can vary from fractions of a microsecond to seconds.

Ideas:

The value that I'm trying to convey can vary over a large dynamic range, but it will always be a "round" number.  For example, it might be 10 milliseconds, or 85 microseconds, but it will never be 10.85 milliseconds.  Therefore, I really only need a couple of digits to convey the value, plus something to convey the metric time units. Using a two digit LED display, including decimal points, together with a three additional LEDs representing microseconds, milliseconds, and seconds, I can convey 99 seconds down to 10 nanoseconds.  For example, here's every decade of 85 from seconds to hundreds of nanoseconds:

85 seconds
8.5 seconds
.85 seconds (850 milliseconds)
85 milliseconds
8.5 milliseconds
.85 milliseconds (850 microseconds)
85 microseconds
8.5 microseconds
.85 microseconds (850 nanoseconds)
.08 microseconds (80 nanoseconds)

Normally, the precision of a number is expressed in the number of significant digits with which is it written.  For example, 85 microseconds +/- 10 nanoseconds would be written as 85.00 microseconds.  But here, since there are only two digits, the precision needs to be explicitly stated, rather than conveyed in significant digits. Since the timing resolution is independent of the scale, there will probably just be a note on the fixture to the effect of "All times are exactly as stated, +/- 10 nanoseconds."

Because the display is intended to be photographed at any possible exposure time, it needs to be static, with all active LEDs on at all times.  It can't be a multiplexed display, the way you'd normally drive a large number of LEDs, because at exposure times shorter than the refresh rate of the complete display, you would only see portions of the display lit in the photo.  With 2 digits of 7 segments each (14 LEDs), plus decimal points (two or three* LEDs), plus three time scale LEDs, that's a total of 19 or 20 LEDs that need to be driving simultaneously.

*Why would I want the third, rightmost decimal point?  It is not specifically needed, but since the middle and leftmost decimal point locations are so integral to conveying non-full-digit values, it might be useful to explicitly include both leftmost and rightmost so there is always a decimal place on and it is not unclear where it resides:

85 milliseconds displays as 85. milliseconds, to explicitly differentiate it from .85 and 8.5 on the display.  This also indicates another problem with this scheme, however, because virtually all 7-segment displays include the decimal point at the right side of the digit. (HEY! A survey of Digikey attempting to verify that assertion turns up this part:

It includes a "Left Decimal Point" and "Right Decimal Point" on the same digit, suitable for use as my leftmost digit.  The dual-decimal point feature is sufficiently unusual that it doesn't appear to be mentioned anywhere in the description of the part on the Digikey website...  If this or some other dual-decimal point digit is not sufficient for my needs, I'm going to end up having to improvise a left-most decimal point, or I could just add a third digit, and then only ever have right decimal points, but also 7 more LEDs, and a higher explicit precision display.  I think I'll make that decision based on other considerations, like brightness, size, color, etc.

Anyway, to make the 19 or 20, or possibly 27 LEDs statically illuminated, I'll need 27 output bits, so I'll probably just drive them directly using a 74AC164 serial in / parallel out shift register.  (If I end up wanting to drive my display electronics at 5V but still have a 3.3V MCU driving everything, I might end up instead using a 74ACT164, which requires a 5V supply, but has TTL input switching levels, meaning that my 3.3V inputs would have more margin.  Though, really, even 74AC should work fine at 5V with 3.3V inputs.

Surviving a Nuclear Attack

I am hopeful this is information nobody will ever need. But it seems like one of those things that's better to know than not. Plus, maybe it will be a good conversation starter at your next dinner party...

Obviously, if you are too close to a nuclear explosion, you're dead. But there's an intermediate range, at which there's a lot of steps you can take to improve your chance of survival, both immediately and in the longer term. Obviously, the prepper / survivalist market has written a LOT about this, but here's my short "What To Do" immediately after a nuclear detonation. Also, please note that this is not comprehensive, nor am I an expert. Much of this is common sense, but it does include some info that surprised me when I learned it. Ultimately, you are responsible for your own safety, so try to keep calm and do what you can to stay alive.

The First Instant:  Seek cover from the flash and immediate radiation exposure - If you suddenly experience a flash brighter than the sun, DO NOT LOOK AT OR TOWARD IT. In fact, turn away from it, and get as much cover between you and the flash as possible. If the detonation is a high yield explosion, the flash may last many seconds. The flash isn't just visible light. It is intense radiation across the spectrum from long-wave thermal IR to gamma. Closer in to ground zero, the flash will ignite fabric and paper. There are photos of Japanese atomic bomb survivors with the patterns from the fabric of their clothes seared into their skin by the differential absorption of thermal radiation. Flash fire is an immediate hazard as things around you, and possibly your clothes, may burst into flames. If this happens, try to put out / remove the fire as quickly as possible, even while seeking cover.

The First 10-60 Seconds:  Seek cover from blast. Time is of the essence: If you're too close to the explosion, you won't have sufficient time to seek cover from blast effects, but then if you're that close, you may well be incinerated anyway. Try to get away from and seek cover from things that will become shrapnel when the blast wave hits, particularly glass. Windows will shatter inward and the glass shards will embed in everything they hit. In the immediate aftermath, medical care may be unavailable, and you want to try to minimize the risk of injuries that can get infected. Glass shrapnel can also be difficult to remove because it is not as easy as metal to identify in x-rays. If you get to cover, and then it seems like no blast wave comes, DO NOT LOOK OUT TO CHECK. Your sense of time may be distorted by the stress of the situation, and you don't want to stick your head out just as the blast sprays you with high speed debris. If, for some reason the blast doesn't reach you, you are still better off staying under cover anyway (see "The First Three Minutes" next.)

The First Three Minutes:  Stay shielded from the fireball. The primary nuclear reaction in a detonation is over in about the first 30 nanoseconds. However, during that initial nuclear reaction, many different radioactive materials are produced, and they continue through their radioactive decay chain. Some of the materials have very short half-lives. This means that the fireball and subsequent mushroom cloud is still emitting a great deal of ionizing radiation. The activity rate drops off over time, but staying under cover for even three minutes will significantly reduce your acute radiation dosage. Gamma radiation is very penetrative, so the more material you have between you and the fireball, the better. Ideally, it would be something REALLY thick, like a mountain. Usually there's not a convenient mountain between you and the blast to duck behind,  but if you can get behind a landscape feature, or a large concrete structure (as long as it won't also fall and crush you) it will help. And anything is better than nothing!

The Next Few Hours: Threats include fires, structural damage, and the immediate fallout. About 20 minutes after the detonation of the Hiroshima atomic bomb, the burning of the large number of wooden structures combined into a firestorm. In the case of Nagasaki, however, there was insufficient fuel for the firestorm to form. The thermal energy of the blast can also vaporize a lot of water from the surface, which can condense on the fallout particles and drop back to the ground as "black rain."  It is extremely radioactive, so if possible, avoid exposure to it, and absolutely do not drink it. Unfortunately, many of the burn and radiation victims after the atomic bombing of Hiroshima were desperate for anything to drink, and so consumed the deadly black rain.

The Next Few Days to Weeks:  Now you're trying to limit your exposure to the longer-lived radioactive contaminants, like iodine, and minimize their damage to you. If you have them, take iodide tablets according to the instructions. This is especially important for children. It will saturate the thyroid gland with non-radioactive iodine and limit the absorption of any radioactive iodine from the environment. Taking vitamin C may also help your body deal with the damage of radiation exposure. Try not to inhale, eat, or drink any fallout. Food in sealed containers will remain safe even if exposed to radiation. If you have a shelter, stay in it. If you don't, you have to decide if it's better to try to flee the fallout area, or stay under whatever improvised cover you have. If you do have to go out, wear a dust mask, or make an improvised face cover. Before you go back into a shelter area, strip off contaminated outer clothes and leave them outside. Clean the fallout off your skin, but make sure not to embed it into your skin by over-scrubbing.

Finally, any event that results in you needing this information will also mean that the world has changed. At least a lot, possibly catastrophically. But try to help others if you can, and focus on staying safe, healthy, and alive.

Plastic Deformation of the Word Führer

I'm currently reading "I Will Bear Witness - A Diary of the Nazi Years, 1933 - 1941" by Victor Klemperer. Klemperer was a German Jew who lived in Dresden with his non-Jewish wife. On February 13, 1945, he was finally ordered by the Gestapo to report for deportation (and almost certain death) in three days, on February 16th. That night, however, Dresden was firebombed by the Allies, and the chaos saved Klemperer's life as he escaped the city with thousands of other refugees.

In his entry for July 14, 1934, more than 10 years before the final spasms of World War II in Europe destroyed his home city, Klemperer wrote (translated from the original German), "Observant Jews purify vessels that have become tref [ritually unclean] by burying them. In the same way the word 'leader' [Führer] will have to be buried for a long time before it is pure and serviceable again."  Now, 82 years later, and 73 years past the end of the Nazi Third Reich, the word führer still carries connotations of evil that it may never shed. The Mirriam-Webster English-language definition of führer is "Tyrant." Though it once just meant "leader" in German, under its use as Hitler's self appointed title, it changed; the definition was bent too far, and now may never spring back to what it once was. Even in other languages non-native German speakers know the word / title that evoked such passion, and for many, terror, so many years ago.

What other words will history change out from under us?  Leave a comment...

Christmas Recipes

Here's some Christmas treat recipes that I've been meaning to post for ages.  I hope to get others posted soon!

Homemade Caramels

(Special thanks to my cousin Stacy for this recipe. I'm not sure where she got it!)

1 Cup Butter (2 Sticks)
4 Cups Sugar
2 Cups Light Karo Syrup
2 (12 oz) Cans Evaporated Milk
1 teaspoon Vanilla

- Generously butter pan
- Add butter, sugar, and Karo syrup to a large saucepan over medium heat until mixture begins to boil (5-10 min)
- Gradually add evaporated milk, maintaining boil
- Stir until firm (this taks a while)
- Add vanilla
- Pour into pan and refrigerate
- Cut and wrap

Peanut Butter Blossom Cookies

Another favorite Christmas season treat, the cookies with a Hershey Chocolate Kiss candy in the middle of each one.

I got this recipe from my mom, though I think she got it from a Hershey's Kiss package sometime in the past.

48 Hershey’s Kisses (see note below)
½ cup shortening
¾ cup creamy peanut butter
1/3 cup sugar
1/3 cup packed light brown sugar
1 egg
2 tablespoons milk
1 teaspoon vanilla
1 ½ cup flour
1 teaspoon baking soda
½ teaspoon salt
Granulated sugar

- Heat oven to 375 degrees.
- Remove wrappers from chocolates.
- Beat shortening and peanut butter in a large bowl until well blended.
- Add 1/3 cup sugar and brown sugar.  Beat until fluffy.
- Add egg, milk, and vanilla; beat well.
- Stir together flour, baking soda, and salt; gradually beat into peanut butter mixture.
- Shape dough into 1-inch balls. (See note below.)
- Roll in granulated sugar; place on ungreased cookie sheet.
- Bake 8 to 10 minutes or until lightly brown.
- Upon removal from oven, immediately press a chocolate kiss into the center of each cookie.
- Remove from cookie sheet and cool on wire rack.  Cool completely.

Makes 4 dozen cookies. 

Note: I modify this slightly by making the dough into ¾ inch balls and using more Kisses as needed.  This increases the ratio of chocolate to cookie, which is always a good thing.  I also reduce the baking time to about 7 minutes (for the smaller cookie) and use baking parchment paper on the cookie sheet, instead of baking directly on the cookie sheet. 

Bad Lockbox Design

This is from my backlog of previously unfinished posts. I'm trying to go back and finish some up. This was from early 2012:

I was recently staying at the Hsinchu Sheraton hotel in Taiwan, and as with most hotels, there was a lockbox / safe in the room to store your valuables while you're out of the room. This particular lock box had an interesting security vulnerability: Every button on the keypad produced a friendly, and uniquely identifiable, press-tone.

Hotel Lockbox with Button-Identifying Tones

Someone wanting to know the combination for the lockbox would need only hide an audio recording device anywhere in earshot of the box to record the tones. From that recording, they could determine every button pressed when either setting the combination or subsequently opening the box. There's a good reason that lock keypads usually use a single tone to indicate button press on all the keys! It's helpful to have audio feedback that you have, in fact, pressed the key, and that you haven't bounced it. But having a unique tone for each button is only slightly more secure than having every key announce itself with a recorded voice enunciation as you press it: "EIGHT!" "SIX!" "SEVEN!" "FIVE!"  That, by the way, is a level of dumb I haven't yet encountered, but I'm sure it's out there somewhere.

Anyway, I wonder if the Hsinchu Sheraton has upgraded these in the last 6.5 years?  If you've stayed there recently, or you've seen some other "security" device design fails, post a comment!

Fun With Homophones

Native English speakers are generally familiar with many of the common homophones, words that sound the same but are spelled differently and often have radically different meanings:

To, Too, and Two

There, Their, and They're
Your, You're, and Yore

Raze and Raise - This pair has the distinction of being homophones AND antonyms!

When I was in elementary school, we were taught that these were homonyms, but some definitions require homonyms to have both the same pronounciation (that is, be homophones) but also have the same spelling (and thus be homographs).  An example of such is pole (a long round support), pole (a fixed point to which other points and lines are referred - the origin of a polar coordinate system) and Pole (a native of Poland.)

Here are some less common homophones:

Undo and Undue - The popularity of the word "undo" has increased markedly since the advent of CTRL-Z type undo features in many software packages. For me, the concept bleeds over into "real life." I have found myself, immediately after breaking something or otherwise making a poor choice, thinking "Undo!  Undo!"  Undue, however, remains a much less common, though occasionally very useful, word.

Away and Aweigh - Here again is a really common word and its much less common homophone.  Indeed, many people don't even realize that the naval song "Anchors Aweigh" is not, in fact, "Anchors Away."  Last year in a pop-up Halloween Store I saw a costume for "Sexy Sailor Girl" labelled "Anchors Away."  In that case, it could have actually been a somewhat clever naughty double entendre, but it was probably just a mistake.

Wet and Whet - As a verb, whet means "to sharpen," so a knife-sharpening tool is sometimes called a "whetstone." Since a sharpening stone is often used with water on it, it is both a whetstone and a wet stone.

Do you have any favorites homophones?  Do you think these should all be called homonyms?  Leave a comment!

Wrecking Ball Analogy

Imagine that right outside your house hangs an enormous wrecking ball.  And when I say enormous, I mean REALLY big.  Like far bigger than your house.  The chain from which it hangs extends out of sight into the clouds, and for the purpose of this analogy, there are always clouds, so you don't worry too much about what the other end of that chain connects to.  It has always been there, since long before your house was built.  It just hangs there, a few inches off the ground. It usually doesn't move, or at most it sways a few inches either way if the wind gets really heavy.  But your house is about a foot away from it, so no problem.  Then, one day, there's an earthquake.  The shaking stops in less than a minute, but you go outside, and notice that the wrecking ball is now slowly swinging away from your house... Five, ten, twenty feet, and it's actually speeding up as it moves away.  What would you do?

I would hope, when you read the situation as I've described it, that you instantly realize that the wrecking ball swinging away from your house is enormously dangerous.  This is the start of a catastrophe that will destroy your house, because the ball will soon come back, with its unstoppable mass, and you need to get anyone in the house out immediately. You would not, I hope, take the time to explore the spot uncovered by the wrecking ball moving back, or follow the wrecking ball away to see exactly how far it goes. You have only minutes, possibly only seconds, to take life saving action.

You may have already guessed where I'm going with this.  Replace "wrecking ball" with "ocean" and the situation I'm describing is the start of a tsunami - a "tidal wave." Interestingly, the physics of the situation are pretty comparable.  Often the first indication of an impending tsunami, other than the earthquake itself, is the ocean "rolling out," "receding" or "dropping." Often people are intrigued and follow the water out.  This is not a good idea.  If you ever see the ocean dropping quickly, you should get away from it and seek higher ground.  And take as many people as you can quickly convince to join you.  Because that ocean, like the swinging wrecking ball, is going to come back  And when it does, it is probably going to destroy everything near the coast.

If you think in terms of possible outcomes, seeking high ground on even the possibility of a tsunami is the winning strategy. If you run away, and nothing happens, you might look a little silly. But if you don't run away, and the water comes in, you may well end up dead. It seems like an easy decision to me. So if you see the ocean suddenly dropping, get away, seek high ground, and warn everyone around you to do the same immediately.

"Fuel, Air, and Spark" as a processor analogy.

One EE I worked with years ago used an analogy that has stuck with me when debugging embedded systems.  He said, "A processor is like a car engine, it only needs three things to run: fuel, air, and spark."  The fuel is power: you need the correct voltage on all the rails.  In some more complex systems, you need to carefully respect the power sequencing rules, but more generally you need the correct power.  The air in this analogy is the code and data, drawn in from memory and through the engine of the processor in the unending march of the program counter.  Sure, the analogy is imperfect because you loop back through and jump around within the code, but it needs to be there, or the processor never has anything to execute.  And finally, the spark is the clock.  In an engine, the spark triggers the combustion in the cylinders, and in a processor the clock advances the sequential logic of all the state machines that together make up the overall state machine of the processor.  Without spark, the fuel-air mixture in an engine never burns.  Without the clock, a processor never advances.

This analogy makes debugging an embedded system or any sort of processor system remarkably straightforward.  If it isn't operating, it's because something is wrong with the fuel (power), air (code / data), or the spark (clock.)  A voltmeter can quickly tell you if the power rail has the approximately correct DC voltage, and an oscilloscope will reveal transient deviations in voltage.  A 'scope will also let you check your clock signal, both for frequency and electrical quality.  And if you have an external address/data bus, you can watch the memory transaction directly.  For an embedded system with all internal memory, your examination of the code / data may need to be more indirect.  But if you get all three things, fuel (power), air (code), and spark (clock), together, and the hardware is not fundamentally broken, then the processor will run.

Ewwwww!! Sharing Germs At The Venetian Hotel in Las Vegas

Note:  I have a ton of half-written posts, so I'm going to try to finish up and post some of them.  Here's something from the start of 2017...

I stayed at The Venetian hotel in Las Vegas while I was at CES in January, and noticed something really gross...

The water glass in the bathroom had a layer of crud in the bottom of it.  Maybe it was just hard water residue, maybe it was toothpaste film from some prior guest.  Whatever it was, I wasn't going to use the glass.  But I decided to take a picture of the crud for posterity, and then I set the glass back down next to the sink:

Bathroom Drinking Glass Jan 6

That, however was not the really gross part.  This morning I noticed that there were two "new" glasses, upside down, in the corner.  I decided to check if the new ones were any cleaner.  That's when I saw this:

Bathroom Drinking Glass Jan 7

Ewwwwww!  The same stain is still partially there.  But only partially.  And that's horrifying.  Because it means that housekeeping must have "washed out" the original glass, there in the bathroom, and then put it back.  And there is no scenario in which the person cleaning the bathroom also washing the glassware is a good thing.

When I checked out shortly after this, I spoke with a manager.  She said that the policy is that the glassware is replaced in the rooms if it has been used.  Clearly this isn't what happened.   She also suggested that perhaps this was two different glasses, but if you look at the shape of the blob in the upper left corner of both photos, it is clearly the same mark:

This original blob is clearly the same stain...

... as this one that had been partially washed off the next day.

Had I known that I would be doing "before and after" comparisons, I would have tried harder to get a better image for the original picture.  Nonetheless, I'm completely confident that this is evidence of a really disgusting housekeeping practices, at least between January 6 and 7 in my room, at the Venetian.

WTH?! Kingsman - "Shoot the dog."

This weekend I watched "Kingsman: The Secret Service" on HBO.  It was a generally fun, over-the-top spy movie with tons of absurd situations about which I happily suspended disbelief and just went with the story.  One thing in it, however, really rubbed me the wrong way.  The lead character Eggsy, a candidate to join this super-elite spy organization, has been given a dog to train and raise as part of his own spy training.  The last test of his readiness to become a Kingsman is when he is handed a pistol and told, "Shoot the dog."  He is unable to, and is summarily dismissed, then upbraided by his mentor and told that he's thrown away this amazing opportunity.  The gun, it turns out, held only a blank, but he didn't know that.  Wait? What?  What, exactly, was the point of this test?  If every other member of this organization has "passed" this test, then they are all either psychopaths, or pathologically willing to follow nonsensical orders. In either case, they probably shouldn't be running around the planet with weapons and the autonomy to do horrible things at will.  I sincerely thought, as I watched the scene, that it was a test of character, and that the "correct" answer was to refuse to shoot the dog.  After all, they've spent much of the film up to this point emphasizing that these agents are "gentlemen."  But nope, apparently they want psychopathic gentlemen.  It really damaged my enjoyment of the entire film.

As an aside, at point-blank range, even a blank can cause injury or death.

So I Want to Manufacture in China... Who Should I Use?

A friend recently asked me on behalf of a client if I had any recommendations about finding a contract manufacturer (CM) in China. Here's the lightly edited version of what I wrote back to her:

This is (as you know) a massively involved question.  It depends enormously on the nature of the product...  Volumes?  BOM cost?  Level of Tech?  The CM you'd use to make 10,000,000 greeting card modules on paper one-layer boards is very different than you'd want to make 1,000 pieces of a 19" x 19" 20-layer HDI PCB with $20,000 worth of parts on it.  In between is an entire spectrum of possibilities. China is also so entirely about relationships that you REALLY need to have an introduction and preferably some prior contact with anyone you want to work with.  You should also plan to have boots on the ground there often or even continuously, especially early in the process.  Ideally that representative is someone Chinese who speaks the language and knows the culture, and has relevant relationships already in place.  Although you then also need to be careful that those pre-existing relationships don't mean that "your" guy is more loyal to your vendor than to you.  A good place to start that I just heard about last week from a co-worker of mine, and which sounds interesting is Dragon Innovation.  I should emphasize that I have not used them myself, but I've heard good things about them...

Another question I would ask is "Are they sure that they want / need to manufacture in China?"  The benefits of building there are shrinking, especially at the lower end of the volume scale.  I have found more and more CMs in the U.S. that are willing to take on medium volume manufacturing at lower cost if you can give them lead time.  They use those jobs to fill in gaps in their "primary" quick-turn prototype assembly.  One place here in the Seattle area that I use for protos but which also does smaller end production work is Out of the Box Manufacturing.  I use them to assemble boards, but they can also do full final product assembly and test.  You can also "split the difference," and do the more commodity aspects of your manufacturing (e.g. bare PCB fab, or PCB fab + assembly, or shooting plastics) overseas, and then do final assembly, calibration, and test here in the U.S. You can also look into working with U.S. firms that have subsidiaries or partners in Mexico or elsewhere in Latin America.

A local design-and-project management firm that we use for a lot of stuff, and with whom I've had good luck, is Monsoon Solutions. They have relationships with fab houses and assembly houses in the US as well as internationally, and can help you find the right balance of cost, speed, and international entanglement.  They will also maintain inventories of part kits and help actually design products as well, if you want.  Of course, you pay for their expertise, but I've found it to be valuable.  Obviously I've got a current bias for Seattle sources.  Many of my Bay Area leads for this stuff are somewhat stale, since over the last few years my employers have either had direct relations with industry-specific ODM firms in China (hence my frequent travel!) or didn't manufacture in China at all...  There's also a fairly recent profusion of manufacturing consultants on all this stuff, arguably making it much easier to get started in HW production.

Anyway, this only touches on some of the myriad issues.  It doesn't even mention issues of Chinese vs. U.S. law , intellectual property protection, supply chain integrity, international shipping, and so on. As usual, the correct answer is "get expert help with this, because it's a monumental undertaking, and it's worth getting right."  Good Luck!

The Evil (Non-)Phonetic Alphabet

I am a big fan of the NATO Phonetic Alphabet when reading character strings to people, especially over the phone or radio. It was designed for optimal intelligibility, and is a standard that most people are at least vaguely familiar with from movies and television.  Plus you get to sound like a secret agent or military scout calling for air support.  When I was talking with my dad the other day, though, he mentioned hearing someone on the radio say, jokingly, "'C' as in 'Chanukah'."  This got me thinking about what an optimally bad / confusing / ambiguous phonetic alphabet would look like.

This is a work in progress, but here's what I've go so far:

Aisle
Bdellium
Czar
Djibouti
Euphrates
Fjord
Gnome
Herb
Isle
Jalapeno
Knee
Lladro
Mnemonic
Ngabu
Ouija Board
Pneumonia
Qatar
Rwanda
Sphygmomanometer
Tsunami
Uvula
Vraisemblance
Wrinkle
Xylophone
Ypsilanti
Zlotys

Have any suggestions for even better (worse) words to convey letters?  Comment below!

Maker Faire 2014!

I'm just home from Maker Faire 2014, and it was spectacular.  This is the first year that I've exhibited there, and I am exhausted, sore, sweaty, smelly, and immensely happy.  My exhibit was an LED Shutter Timer which allowed people to see the exposure time, shutter travel speed and shutter direction of their camera, whether it was the electronic rolling shutter on a camera phone, or the travelling-slit curtains of a DSLR.  I talked with people almost non-stop both yesterday and today, and I met an amazing collection of creative, friendly, engaging and engaged people.  If you're reading this because you saw the URL for this blog on my table, feel free to look around, and soon I'll have all the technical information about my project posted.  In the next 24 hours though I need to shower, sleep, and try to recover my voice!

If you went to Maker Faire this year, please leave me a comment and tell me the coolest thing you saw there!

"Fine, continue paying your debts!"

I got a robo-call this morning from someone telemarketing their debt consolidation and reduction services. These places usually provide some combination of loan aggregation and paid negotiation with your current creditors to reduce your balance owed, at the cost of trashing your credit rating for a while.  When I have time I will sometimes play games with telemarketers, because I abhor their business model.  Sometimes I'll even try to convince the person with whom I'm speaking that they should quit and follow their higher aspirations in life, which I'm sure never included sitting in a boiler room call center selling random people things that they don't need.  Today, however, I was in a hurry, so I skipped straight to, "Do you check your calling list against the federal Do Not Call list, of which you are now in violation?"  The woman paused for a moment and said, in a snotty tone, "Fine, continue paying your debts!" And then hung up.

I was amused and somewhat disgusted that of all the things she could have said, she chose to insult me for being someone who continues to pay his debts.  I wonder if she's ever considered that the logical extension of her work and apparent world view is the fundamental breakdown of credit systems everywhere.  After all, there is an enormous amount of unsecured credit extended every day, to individuals and business, in order to keep the economy buzzing along.  Could we live without it?  Sure, but everyone would have to maintain a larger cash reserve, and all large purchases would have to wait until after the funds for the purchase were acquired. You'd also have to send payment for all purchases in advance, or at least through some sort of escrow transaction. Imagine if every time you ordered something online, you had to send payment before the seller shipped.  And it would have to be a check, or money order, or bank transfer, because in a world with no unsecured debt, there can be no credit cards or even charge cards.  But these excellent tools do exist to allow us to smooth money over time, because people, for the most part, continue paying their debts.

I understand that there are times when circumstances make paying debts impossible.  Risky investments, such as venture capital poured into start-ups, are made with the full understanding that the money very likely will never be paid back.  In that case, the investor is taking on substantial risk in exchange for substantial potential upside.  Sure, you're 90% likely to lose your investment, but it only takes one early investment in Google to pay for all your investments in pets.com.  But here I'm talking primarily about unsecured personal debt, such as credit cards.  And yes, if you owe more in interest each month than you're able to pay, the financial hole will just get deeper without bound. In that case, a negotiated reduction in interest rate is preferable to a settlement, which is preferable to bankruptcy.  And bankruptcy is far preferable to protracted destitution.  But writing down debts should be a last resort.  When possible, we should aspire to avoid unnecessary debt, and when it is a necessary and useful evil, we should then continue paying our debts.

A Bad Good Idea...

I am a fan of old 8-bit computers, so when I found an original TRS-80 in the surplus electronics store where I buy things like stepper motors, I had to take it over to the test bench and see if it actually worked.  It did power up, but while I was connecting it up, I discovered this:

There are three connectors on the back of the keyboard unit which are labelled, in raised black-on-black letters, TAPE, VIDEO, and POWER.  All three are identical 5-pin DIN connectors.  This means that any of the cables will fit into any of the connectors.  And they are on the back of the keyboard, so you're very likely to be plugging them in blind, by feel.  This means the chance of plugging the wrong cable into the wrong hole is pretty high.

As with all consumer products, the engineer that designed this computer was desperately trying to keep the cost as low as possible.  By using the same connector, there was only one unique external connector part number on the bill of materials, so their purchasing volumes for that connector were three times as high, meaning they get better volume pricing. They also needed to create only one connector footprint symbol for the printed circuit board.  But from a human factors point of view, this is a terrible decision.  It is absolutely certain that people are going to plug the wrong cable into the wrong connector sometimes, and their computer would be mysteriously not working.  Indeed, depending upon the pinout of the connectors, the computer might even be damaged.  If it is still under warranty, they'll be angry and demanding of a repair, replacement, or refund.  And if it is out of warranty, they'll be angry and they'll tell all their friends about your terrible quality.

When you design something, it's important to think carefully about all the ways it could be used incorrectly.  You can't always outwit someone who is determined to do something wrong, but you want to try to make the correct way of doing things the path of least resistance.

PCB Copper Thickness

When specifying the thickness of copper on printed circuit boards (PCBs) the usual metric is "ounces."  One ounce copper means the thickness you get if you flatten one ounce of copper out to one square foot.  The chart below shows the thicknesses of various ounce weights of copper in both mils (thousandths of an inch) and microns (millionths of a meter).

Copper Thickness (in Ounces)	Copper Thickness (in Mils)	Copper Thickness (in Microns)	Resistance per Square (in mOhms)
1/4 Ounce	0.34 mils	8.5 microns	2.0 mOhms
1/3 Ounce	0.44 mils	12 microns	1.5 mOhms
1/2 Ounce	0.67 mils	17 microns	1.0 mOhms
1 Ounce	1.34 mils	34 microns	0.5 mOhms
2 Ounces	2.68 mils	68 microns	0.25 mOhms
3 Ounces	4.02 mils	102 microns	0.167 mOhms
4 Ounces	5.36 mils	136 microns	0.125 mOhms

The table also shows the approximate "resistance per square," which is the edge to edge sheet resistance of any size square area of that weight of copper.  This is very useful for quickly calculating the DC resistance of a PCB trace by decomposing it into a series of squares.  For example, a trace 10,000 mils (10 inches) long and 10 mils wide is made up of 1000 squares, each 10mils x 10mils, in series.  If that trace is 1/2 ounce copper, the DC resistance would be 1000 squares x 1.0 mOhms / square = 1000 mOhms, or 1 Ohm.  If you are running high current through that trace, 1 Ohm can cause a substantial voltage drop.

Super Easy Giant Chocolate Fondue

This is a bit out of the normal realm of my engineering projects.  Last week at work we had a Halloween Party at lunch, with each department kicking in some activity or fun thing for everyone to enjoy.  I volunteered to do a chocolate fondue.  My goal was to make enough for everyone there (about 100 people) to try some if they wanted, and to make it relatively quick and easy to prepare.  I found a recipe here and decided to scale it up.  The original recipe calls for four cups of chocolate chips, which is 24 ounces.  Since Costco sells a 72 ounce bag of chips, I decided to do a triple batch.  This worked out great, because the recipe also calls for 1 and 1/3 cup of half and half.  Triple that and you get 4 cups, or 1 quart!  So it really did come down to:

1 giant (72 Ounce) bag of Nestle semi-sweet chocolate chips from Costco
1 Quart of Half-and-Half

Melt together in a crock pot.  That's it.

I used a 6-quart crock pot, and I might have gotten away with a smaller one, but I was glad I had some room to stir.  I also did use the Reynold's liner bag, which made clean-up very easy, but was probably otherwise not necessary.

To get it ready quickly, I deviated from the source recipe's instructions:  I poured both ingredients into the crock pot and put the stoneware pot (but not the heater base) into a microwave on high for 5 minutes to get it started melting.  Then I put it in the stoneware back into the crock pot heater and set it on high.  I stirred it almost continuously to prevent burning, and in about 15 minutes it went from the consistency of chocolate milk with large melty lumps to a smooth, very nice dipping chocolate fondue.  At that point, I set it back to "warm," which kept is just right throughout the party.  I also stirred it occasionally to make sure it stayed smooth.

This produced about 100 ounces, which is enough for 100 people to have a reasonable taste, or 20 people to go quite nuts.

For dippers, I put out mini snickers, marshmallows, Oreos, pretzels, and strawberries.  Far and away the most popular item was the strawberries, and I ended up wishing that I had bought much more than the two quarts I did.  Lots of other things also work well with chocolate fondue, such as pineapple, apple, and poundcake.  Really, what ISN'T better dipped in chocolate?  One of my clever co-workers took one of the apples from the bobbing-for-apples area and cut it up for dipping.

It's fun, it's quick, and it's an impressive amount of molten chocolate!

I crashed my flashlight...

My dad recently bought some Techlite 200 high-intensity LED flashlights at Costco, and gave me two of them.  They are amazingly bright and well constructed, so they quickly became my favorite hand-held flashlights.  I was reading this extremely detailed review of them on Amazon, and learned that I could skip the somewhat jarring strobe mode when turning them off by just holding down the power button for a couple of seconds.  I tried it, and it works, but after doing it several times, my light would no longer turn on.  I had to remove the cap to disconnect the battery, causing a hard reset of whatever control circuitry they have in there.  I don't know exactly how they implemented the electronics, and I'm not willing to sacrifice this neat flashlight to find out, but I crashed my flashlight...

Attempts to reproduce the failure on the other Techlight 200 I have were unsuccessful, so it is apparently not an inherent design flaw, just an edge marginality.  Still, I crashed my flashlight!  There is some sort of cautionary story in here somewhere...

"Revolution" TV Show.

This entry is some thoughts about the new television series "Revolution."  This latest invention from the imagination of "Lost" creator J.J. Abrams premiered Monday, September 17th, on NBC.  The pilot episode was released early on Hulu, so I took a look at it.  I'll try to write this entry without spoilers, or at least without any more spoilers than the ads and website for the show provide.  Still, if you prefer to avoid even a hint of fore-knowledge, you may want to read this after you watch the pilot episode.

"Revolution" is set "fifteen years after the blackout."  In a video on the show's website, Abrams says, “The question the show asks is, 'What would happen if everything powered by electricity suddenly turned off?'” Because I'm interested in worst-case scenarios, disaster preparedness, and knowing what to do when the zombies come, I'm intrigued by ideas like this.  But this a problematic scenario...  People are really ingenious and given the proper motivation (say, the loss of civilization as we know it), they would come up with some pretty clever stuff.  Let's consider how...

While it's true that we use electricity for an enormous number of things, we can still do a lot without it, as we did in the past.  Energy can be harvested mechanically in the form of movement, with windmills and water wheels or water turbines. In 1890, George Westinghouse suggested that the energy of Niagara Falls would best be transmitted to Buffalo not as electricity generated on site, but as compressed air. Compressed air can also be used to store mechanical energy in high-pressure gas cylinders.

Furthermore, a lot of technologies that we think of as “electrical” don't have to be. In many cases, we use electricity to provide mechanical energy that could be provided by other sources. Refrigeration, for example, uses electricity to turn a motor to run a compressor. But there's no reason the compressor can't be turned by a water wheel or a windmill. Or a steam engine...

In the event of a “Revolution” scale blackout, the world would quickly see the return of the steam era. Boiler technology and steam engines would be the next big growth industry. Coal and wood fired steam railroads would quickly provide critical transportation links, first on a small scale as museum exhibits were pressed back into service, but then growing as fast as more equipment could be built. Central steam engines in factories could provide mechanical energy for all sorts of tasks, such as machining, cooling, cutting, building, and so on.

And if the “Revolution” world really just precludes electronics, it wouldn't be limited to steam engines. Diesel engines use the compression of the piston to ignite the fuel-air mixture, without any electrical spark plug. And they can be built to be started using only compressed air. It may take a lot of tedious pumping, but you could prime a compressed air system with a hand or foot pump. Then, once you've got the engine running, you could use it to drive a compressor to charge up the compressed air cylinder for the next time you need to start.  All without electricity. Virtually all large scale modern engines are highly dependent upon electronic engine control modules  (ECMs) to manage every aspect of their operation. But that's in order to make them as efficient as possible. You can trade away some of that efficiency for a much simpler, all mechanical control system. That's how they all used to be. And there would be enormous motivation for people to return to and advance these sorts of technologies.

Refrigeration also illustrates the case where the current state of the art is only one of many ways to achieve the desired goal, in this case of making things cold. Anyone with a propane fueled refrigerator in their camper knows that you can keep a fridge full of food or drinks cold just fine using a small flame to drive an absorption refrigeration cycle.

A permanent loss of electricity would limit the production of some materials. Aluminum, for example, is refined using an electrolytic process, without which, it is becomes a very precious metal. Fortunately, there is already a significant stock of metallic aluminum in the world that could be recycled almost endlessly. But anything that depends on an electrolytic manufacturing process could no longer be made.

Otherwise, I can think of only a few key technologies that would be impossible to at least approximate in a “Revolution” world: high-speed computing and high-speed, long-distance, and wireless communication. And even some level of both computation and communication can be done mechanically. Mechanical calculators existed for years before modern electronic logic circuits, and I'm sure the world would see new heights of sophisticated mechanical computers being built. It would be a steam-punk enthusiast's dream come true. I can also imagine a telegraph system based on mechanical modulation of something like a metal bar, in order to send messages significant distances at the speed of sound in metal, or around 13,000 miles per hour. Or perhaps we'd just see the return of other pre-electronic communication technologies.  Sure, neither of these are as good as what we have now, but they are still a major step above the pre-industrial, agrarian civilization portrayed in the show...

T-Mobile Customer Service Fail

I was in Seattle recently, helping work an event booth at the Bite of Seattle, and we needed to upload a bunch of photos.  The available free WiFi options were pretty poor, so we purchased a T-Mobile wireless hot spot. They offer high performance network connectivity, and contract-free data plans that can be purchased or re-charged easily.  As in so many situations, it seemed like a good idea at the time.

Once we got the hot-spot set up, we had network connectivity but it was insanely slow.  I put my phone in airplane mode and connected only it to the hotspot, then ran the SpeedTest.net app, and got 13 kbit/sec down, and 9 kbit/sec up.  That works out to 2.96 hours to upload each photo.  Obviously, that was not suitable.  At other times, I did get somewhat better performance.  When I thought to get a screen capture (Thank you, Ice Cream Sandwich!) it was up to 31kbit/sec.  Now each upload would be slightly less than an hour!

Rather than throw more time and effort down this particular rat hole, I decided to take the easiest of consumer outs and just return the product that clearly didn't meet any reasonable user expectation.  At my local T-Mobile store here in Santa Clara they gave me a prompt and courteous refund of the purchase of the hot spot, and didn't even try to ding me for the $25 restocking fee they warned of when I purchased it.  But they wouldn't refund the $35 for the useless 3GB data plan at which I had barely sipped.  For that, they said, I would have to call T-Mobile Customer Service.

What followed was a customer service experience surpassed in badness only by the U.S. Postal Service.  The Postal Service was worse only because the folks at T-Mobile pretended to care about my annoyance.  Over half an hour on the phone I was transferred from tech support, to customer care, to prepaid refill, to payments, to refunds.  Although T-Mobile runs the network, conveys the data, tracks the usage, and can recharge the account when needed, they had no mechanism to verify (and thus refund) the 3GB data plan purchase I had made.

At the end, they said that I would have to go to the T-Mobile store where I purchased the plan to get a refund.  That, I pointed out, was in Seattle.  Well, then, I'd have to go to a local T-Mobile store.  That, I pointed out, was who sent me to call them.  Ultimately, they won, in that I gave up, and decided to instead just tell them they suck, and expense the $35 rip-off.  I will yet contest the charge on my credit card, since they didn't actually provide the service I paid for.  Even then, I may or may not get my money back.  At least the Post Office eventually gave me a refund in stamps...