After a weekend with a bunch of MIT friends in Las Vegas, and then a Sunday morning trip to see the Hoover Dam, I found myself back in Vegas filling up for the ~300-mile trip home.  However, the car did not want to cooperate, and the gas pump would shut off within 1-2 seconds of starting, sometimes leaving a small puddle above the flap in the filler tube.  I moved to a second pump at the same gas station and had the same issue.  No matter how slowly I tried to add gasoline, it would stop within a few seconds.  After about 15 minutes, I had only added about 2 gallons of gas -- that's almost as bad as charging an electric car!

What's the big idea here?  The gas tank is a fixed volume and is not under much pressure, so when I push a gallon of gasoline into it (which takes < 10 seconds at a typical gas pump), a gallon of air and gasoline vapors has to be pushed out to make space.  If the vapors don't move, then the pressure inside the tank builds, which forces the one-way fuel fill valve between the tank and filler pipe to close.  The gasoline then builds up in the filler pipe and touches the nozzle, which forces it to shut off.  (Click here to see more details and illustration of the automatic shutoff gas nozzle.)  Similarly, if no fresh air can enter the fuel tank as the car is driven, then the pressure would drop and would create a vaccum relative to atmospheric pressure.  So to summarize, air needs to be able to enter the tank slowly while driving (at a rate of at most a few gallons per hour), and air needs to be able to exit the tank rapidly while refilling (at a rate of a many gallons per minute).

Nowadays, the air/vapors exiting the tank are cleaned in order to prevent hydrocarbon vapors from simply being emitted to the atmosphere -- "Onboard vapor recovery" has been mandated by the EPA since 2000.  Now, the path looks something like this (from the Hyundai service manuals website, linked from this helpful forum post):

In fact, this shows less than half of the "plumbing" involved around the fuel tank, so this diagram was a bit confusing as we tried to attack the problem.

On this car, the gas tank filler hose is on the driver's side.  Below is the view from under the left rear tire, which shows three connections.  The left (small) tube goes to the small ports around the main filler pipe.  The middle is the main filler pipe which carries gas to the tank.  On the right is a odd-looking unit which happens to be the "liquid-vapor separator".  I can't be 100% sure, but the hypothesis is that it's just a physically-higher unit which ensures that if one were to really over-fill the tank, liquid fuel would still have a hard time crossing to the canister.

And below is the view of the same thing from underneath the car.  The large pipe on the left is the main filler pipe, where gasoline goes when refueling.  The liquid-vapor separator is the long rounded object parallel to that pipe, and in this photo, the bottom connection to it is coming from the fuel tank's vent valve (at a point high up on the fuel tank), and the top connection is going into the box on the right of the photo.  The box on the right of the photo is the charcoal canister which collects fuel vapors.  This serves two purposes: one is that during vehicle operation, those vapors can be purged into the engine intake and combusted, which prevents the release of hydrocarbons.  Second is that this is the path for air intake/exhaust for the fuel tank.  The upper (thicker) hose in this photo going to the charcoal canister from the liquid-vapor separator.  The lower (skinnier) hose in this photo coming out of the canister is for the purge system, which sends those vapors back into the engine.

After a little bit of work, we removed the canister assembly, as shown below.  This involved removing two bolts for the canister holder, disconnecting the two vapor hoses, removing an electrical connector, and loosening the zip-ties.

The canister assembly is composed of three primary elements: the canister itself, an air filter which is open to the atmosphere, and the canister close valve, which connects or disconnects the air filter from the canister.  The last two photos in this gallery show the canister separated from the air filter+CCV.

Well, after all of that, we didn't find anything wrong.  No kinks or obstructions in the hoses, no significant back-pressure or other issues.  Since most people on the forums described finding hoses filled with little carbon pellets, we knew that that sort of thing should be obvious enough, and it wasn't.  So we put everything back together, drove to the gas station, and guess what -- it filled up just fine!

An unsatisfying conclusion!  Still, that fill-up in Vegas was the hottest the car had probably seen in its lifetime, which makes me wonder whether it may have been related.  Or it could have been the two pumps I tried, or any number of other things.  A learning opportunity, and an opportunity to get more comfortable working on cars, and a working vehicle at the end of the day, so I think I'll call it a victory.

I recently made this bottle opener on the milling machine from some aluminum bar stock. As you can see from the photos, it has changed a bit since my original CAD sketch, but I've decided to include the drawing anyway.

It's very strong and light. If I were to do it again, I'd probably make it thinner than the current 0.200" thickness, and would reduce the length of the thin extended arm and increase the length of the part before the "hook" to give the user a slightly longer lever arm.

Humberto had the idea of having the step in the thin extended arm to allow it to better deform the metal bottle cap (contacting roughly at its center) instead of simply prying it off (contacting at the far edge of the bottle cap as per the CAD sketch). Any opinions on which is better?

Two photos and a sketch:

Lunch at "Bar-B-Que House" Ocean Beach (yelp link).

View from the pier at Ocean Beach, San Diego:

Tried the water, but it was a bit too cold for swimming.  Still a few surfers in wetsuits and little kids who don't care about the cold.

24-hour water temperatures:

1-year water temperatures:
This data is much noisier than I expected, but it appears that only July through September have >=70 degree ocean water temperatures.

Just released my Python command line utility for Notifo, which is an API and iPhone app that allows for easy push notifications to the iPhone. For the source code and more details, see:

http://www.nerdkits.com/files/notifo/

As a follow-up to my earlier post: http://blog.compumike.com/new-tires-now-including-a-free-broken-lug-stu

Thursday 3/25: I got new tires. They broke a lug stud and the broken part is stuck inside the wheel nut. The tire place sent me to a nearby mechanic and said they'd take care of it (paid for by the tire place). I'm informed that they'll get the parts and will call me either Friday or Saturday to do the repair.

Friday 3/26: Hearing nothing, I call the mechanic in the afternoon. They have the lug stud, but haven't been able to get the wheel nut yet. I make sure they have my number, and they assure me that they'll call as soon as they get it.

Saturday 3/27: Heard nothing -- and they're open on Saturdays.

Monday 3/29: Heard nothing. Should have called them, but got busy with work and didn't get to it in time.

Tuesday 3/30: Still hearing nothing, I called the mechanic in the afternoon, and was informed that they got the wheel nut from the dealer, but it was the wrong one, so they're sending it back and trying to get the right one. I'm assured that they'll have it on Wednesday, and they even ask me whether the morning or afternoon is better for me. I say morning, and am told that they'll likely call me around 10am, and confirm that I could bring my car in around that time.

Wednesday 3/31: It's now 2pm. So much for that 10am call. Haven't heard anything, and am about to call them. Any bets?

All for < $10 in parts, and a job that I could probably do myself in about an hour -- but really shouldn't have to.

View of Coronado and San Diego from Cabrillo National Monument:

Tidepools along the Pacific at Point Loma:

My car had Goodyear Eagle RS-A tires that were down to a sad 2 to 3/32" of tread after roughly 36K miles. I've only been the vehicle owner for less than a tenth of those miles, so I can't say much about how well they hold up, but in any case it was time for new tires.

Just picked up four new tires: "Bridgestone Potenza RE960 AS Pole Position", which were moderately priced and had excellent customer reviews on tirerack.com and discounttire.com.

Unfortunately, while they were doing the work, the shop broke one of the lug studs on the front right side. They only do tires and wheels, so they referred me to a nearby mechanic and said they'd cover it, but it was late in the afternoon and they didn't have the part on hand, so I've got to head back tomorrow. (The mechanic tosses in a "we'll also give you 10% off on an alignment" up-sell.)

Ultimately, we're talking about a $5 lug stud and nut, plus perhaps 20-30 minutes to actually get to it and do the work. Still, it's a reminder of how simple things like a 30-minute new tire installation can turn into a complicated multi-hour / multi-day affair.

Hopefully tomorrow I'll be able to report that the situation was resolved swiftly and painlessly, as an example of humble and competent customer service responding to an unfortunate incident. Of course, until then, I'm torturing myself by dreaming of all the ways that this can go wrong from here...

If you haven't heard about "Buckits" yet, watch this quick 90-second
video which explains how we're combining online payments and social
networks to create self-serve group discounts:

I need your help right now to advance this idea to the next round of
PayPal's X Developer Challenge. There are about 50 entries in the
current round, and the top 10 by popular vote advance to the final
round. Can you please take a minute and vote for the idea? Go to this
link to vote:

http://www.nerdkits.com/buckits/voting/

The Buckits system makes for a win-win situation: purchasers win because
they save money by leveraging their social network, and vendors win
because they are able to reach more purchasers.

If you're curious to see a live Buckit in action, go to:

http://www.nerdkits.com/buckits/view/VBMMRMXXG4/

If you'd like to see more systems like this that leverage the power of
your social network to help you save money and enjoy more group
experiences, then please give us your vote and tell your friends to vote
too!

Voting ends tomorrow, Friday 3/05, 11:59PM PST -- so please vote now!

We took a piece of aluminum bar and made an arm that comes out radially
from the quill collar.

It's inset so the bottom of it is flush with the bottom of the collar.
All that protrudes is the two screws. This was important to make sure
the arm wouldn't interfere with our work when milling.

It's distance is adjustable radially via the slotted path for two hex
cap screws.

We used a dial indicator and measured a deflection of about 0.002" at
the end of the arm under a heavy hand-applied force (maybe 50 pounds or
so). Hopefully this means the deflection under the drag of the slide
will be negligible!

The end is threaded radially, which is where the digital caliper will
attach.

We've been working on making a new collar to fit around the mill quill on our Smithy CB1220XL. We started with a 4" diameter by 3/4" long piece of 6061 aluminum. After a bunch of facing / turning / boring / milling / drilling / tapping / sawing operations, we've successfully made the part. We had great results with lathe accuracy, able to get diameters finished to within roughly +/- 0.002" of target.

The outer diameter is arbitrary. The larger bore diameter must fit snugly around the quill collar for a press-fit -- it is then tightened with the two hex-cap screws as shown. The smaller bore through the piece must clear the rotating spindle, but must remain close so as to provide a chip shield for the thrust bearings just within the collar.

We ran into lots of issues with work planning, some of which were due to the reduced work volume of the machine, but most of which were simply due to our original paper design taking insufficient consideration of things like obstruction by the drill chuck, tool cutting lengths, etc.

But it works! It goes on snug and is quickly tightened via the two screws. This replaces a plastic cap included with the machine. However, this version gives us the opportunity to add a Z-axis digital read out (DRO)... stay tuned!