Time to upgrade to LED Lights

LED lights are much more efficient than traditional incandescent bulbs, slightly more efficient than compact florescent (with much less mercury!), and have recently really dropped in price.

I just purchased a six pack of 11 watt LED bulbs designed to replace 65 watt flood lights for $27. When lit, together they use 66 watts. (And light up the kitchen better!)

Compared to the 390 watts used by the incandescent bulbs they replaced, this is a savings of 324 watts.     If they are lit for three hours a day the savings is substantial; 972 watt hours, or almost one kWh!   If we pretend the average cost of electricity is 0.10 a kWh (it’s actually closer to 0.117 for me) this works out to paying for the light bulbs in energy savings in less than a year (270 days!).     As long as the bulbs last for at least 810 hours, they have paid for themselves. (The rated life on the package claims 25,000 hours)

If we conservatively pretend the bulbs will only last 10,000 hours (9 years at 3 hours a day), they will continue to save   324 watts x 9190 hours after they have paid for themselves. This works out to 2,977 kWh, or $297 worth of electricity.   Not a bad return on investment for $27 of sunk costs.

In summary, it’s time to replace any incandescent bulbs you have with LED’s. (You may as well wait for the CF bulbs to burn out before you replace them.)

96 °F charging

I tested my 80% profile (limited to 25A) on a very hot 92 °F day. The chargers reached 65 °C but did not shut down due to overheating. (My 80% profile ends charging when the draw reaches 10A.)

A charger temp of 65 °C is still much hotter than I would like, so I don’t plan on charging in 90 °F heat very frequently.


How to convert a Denford / ScanTek 2000 Micromill to LinuxCNC / Mach3 control: Part 1 – 3 Axis control

How to convert a Denford   / ScanTek 2000 Micromill to LinuxCNC / Mach3 control

Part 1 – 3 Axis control
Video here, details below


My ScanTek 2000 ScanMill (A re-branded Denford Micromill 2000) has a dispatch date of 2005, which means that it’s main controller is a Baldor NextMoveST card. This card supports USB as well as RS232 control signals, and runs a custom (MINT) programming language that can offload machine control from the host computer.

Baldor NextMoveST

You can actually download the MintNC development tool from the Baldor website that allows you to upload custom Mint programs to the card, and could make it (for example) do some simple operations offline with no driving computer. However, I am not interested in writing Mint code to interpret g-code, so I’m going to set it up so that LinuxCNC (or Mach3) can control it via a parallel printer port.
Continue reading

Battery Bay Plan: Nissan Leaf Cells six group


This is my current CAD mockup of how I will package a group of six Nissan Leaf modules. I have 48 total modules from a Leaf battery, so I will have eight of them to distribute throughout my truck’s battery bays. I also have a CAD model of that:

Unfortunately, my existing Lead Acid (golf cart battery) bays are not tall enough to mount them vertically, so I am having to lay them sideways. This results in terminals being relatively close together in the two side bays. (I’ll probably slide a sheet of plastic between the two batteries after I get them installed to make it more difficult to drop a wrench down there and short things out. It’s only 32 volts total potential, but a lot of Amps!)

If you would like to know the specifications of a single module (and some people have asked) here is what I know:

From: http://www.electricvehiclewiki.com/?title=Battery_specs
Confirmed by Jay with a 2013 Nissan Leaf module.

lm_width=223; // mm – 8.7795 inches (measured 222 – 8.75 inches)
lm_length = 303; // mm – 11.9291 inches (measured 298.45 or 11.75 inches)
lm_thickness = 34; // mm – 1.3779 inches (measured 34- 1.3333 when compressed)
lm_weight = 3.8; // kGrams, or 8.3775 lbs

The power “bolt blocks” are 20×20 mm in size, and I modeled them around 1 inch in height. But, this height includes the space for a bolt head and a busbar on the top. In the CAD model I made them one inch tall, but in real life they are slightly shorter than that. The “bolt blocks” for the sense terminals in the middle are slightly smaller (18×18 mm) but the same height. So if you download my CAD model (link below) keep in mind that the top of the bolt blocks on the module includes a little extra wiggle room. The CAD software I use is OpenSCAD, which you can download for free. The file itself is human readable text.

BatteryPacks.scad (You may have to rename it from BatteryPacks.scad.txt to BatteryPacks.scad )

UPC Barcelona Sodexo Unity cafeteria menus

The Sodexo “Unity” cafeteria at UPC in Barcelona offers a “complete menu” for 6.60 Euros that includes a starter, main, drink, bread and desert. You could typically choose between three different starters and main courses, mixing and matching.

The common vegetables were potatoes and artichokes. Ham, bacon and fish were the relatively common meats.


Below are photos of the ten meals I ate.

Grilled fish, french fries, and kale.
Continue reading

Hexagonal wooden mirror frame

This is my finished hex frame mirror, which is the last piece of the downstairs bathroom we renovated.


I built a hexagonal wood frame out of ceder planks for the new bathroom mirror. I had originally wanted to build an irregular three sided “triangular” mirror, but once I figured out that my compound miter saw wouldn’t make cuts sharper than 50 degrees I decided I needed more than 3 sides…and 8 pieces would have been a nightmare to assemble.


I used a table saw to rip a groove in each piece to hold the glass, and then I used the flattest surface I had available (the mirror glass) to assemble and glue the pieces together. The mirror glass was original to the house, and has a 1961 date printed on the back.


Once I had the frame built, I traced out a template so that I could mark the mirror glass exactly where it needed to be cut. (Due to a few holes I was avoiding in my scrap wood, the frame is not perfectly symmetrical….)


If I hadn’t already built the frame, I would have strongly considered making a “Superman” mirror at this point in the glass cutting phase.


Playing a USB midi keyboard on Ubuntu Linux

I purchased a USB MIDI keyboard (the Alesis Q25) for use with my musical Tesla coil, but also wanted to use it to learn to play the keyboard/piano when traveling with my laptop.

The MIDI keyboard has no synthesizer of it’s own, so you need a computer that will take the MIDI data stream and play (synthesize) the actual sounds of a piano or other instrument so that you can “hear” what you are playing.

There are many pieces of complicated software on Linux that will allow you to compose music with a MIDI keyboard, but all I wanted was the ability to hit a key on my keyboard and have it make music.

The easiest set of software I found to do this was vmpk (virtual midi piano keyboard) combined with Qsynth (a GUI interface to fluidsynth). I started up qsynth and vmpk at the same time. Then I used the “edit->connections” menu on vmpk to set my MIDI keyboard as the incomming connection, and the “FLUID Synth (Qsynth1)” as the output connection.

To set up qsynth I hit the “setup” button and then in the MIDI tab selected the “alsa_seq” as the MIDI Driver and “qsynth” as the MIDI Client Name ID (ALSA/CoreMidi).
I also enabled the soundfront from the /user/share/sounds/sf2 folder.
I don’t know why I had to use vmpk to make this linkup between the keyboard and Qsynth, but so far I haven’t found an easy way to tell Qsynth to just listen to midi events from the keyboard directly. On the plus side, you can use the virtual keyboard that vmpk displays on the screen to play notes if you don’t actually have a physical USB MIDI keyboard.

Fixing the problem where gparted (parted) won’t see a partition due to a mac partition table

I purchased an external HD that was “mac compatable” but I used it with a linux system and used fdisk to put two partitions on it.

Later on, I wanted to use gparted to easily resize one of the partitions, but it refused to see any partitions at all on the disk.

fdisk could still see them just fine, but reported “Partition type: mac”

It turns out that the problem was that the disk originally came with a mac partition table in addition to (right after) the regular MBR Master Boot Record.

I noticed that the first partition didn’t actually start until 63 sectors into the disk (at the beginning of the 2nd cylinder).

Device Boot Start End Blocks Id System
/dev/sdb1 63 1171893554 585946746 83 Linux

So I used DD to copy the first cylinder to a file:

sudo dd bs=512 count=62 if=/dev/sdb of=firstCyl.bin
62+0 records in
62+0 records out
31744 bytes (32 kB) copied, 0.000715733 s, 44.4 MB/s

Looking at that bin file in an editor, I saw the string “Apple_partition_map” which is a dead givaway of what the problem was.

So, I wrote out all zeros to the first cylinder:

sudo dd bs=512 count=62 if=/dev/zero of=/dev/sdb
62+0 records in
62+0 records out
31744 bytes (32 kB) copied, 0.00165608 s, 19.2 MB/s

And then I copied the first sector (512 bytes) back from the firstCyl.bin file I had made:

summetj@constantine:~$ sudo dd bs=512 count=1 if=firstCyl.bin of=/dev/sdb
1+0 records in
1+0 records out
512 bytes (512 B) copied, 0.00183878 s, 278 kB/s

And it worked! Now gparted is no longer confused by the apple (mac) partition table that I zeroed out, and sees my partition.