A summary

Summary for July 25-31

This week not too much happened. First off, I felt not like working on the ship much, then some new toys arrived for Annie and me, and they needed some attention, too. Tried to calculate the ships weight so I can decide on the measurements of the body.

Loihi rewired Improved computer wiring: shortened IDE cable (used a vise to crimp on the plug into a new position), shortened power connection to DC/DC converter, replaced cable-salad to LEDs and power switch with custom flat cable (that took some time). Mused about placement of ventilator (into top of box?) since it interferes with the HD, which still is not fixed into place.

Played with the white LEDs, created a 3x3 array of 3 rows of 3 LEDs in series with a 100Ohm resistor on 12V. Found that the current is almost 30mA per row instead of the planned 20mA - will replace the resitors with 150Ohm ones.

Went to Merten, bought Faston connectors, skrinking tube, 150Ohm resistors (checked that out with the white LEDs: current now is 20mA - thumbs up -, so the forward drop voltage is rather 3.0V than 3.4V), new DIP sockets, solder, desoldering wick, 2 blue LEDs (for effectful lighting).

Pretty much finished all electrical connections (the main engine still needs wiring), and made a temporary connection for the 12V battery. Booted up the computer on battery power; works! Seeing it connected made me think about voltage monitoring; this is something I might want the computer to measure and react to.

Started redoing the Basic Stamp board (SVN version 2, so to say - hey, applying versions and tags to hardware might be useful!). It currently looks messy and the way I positioned the components, too much wiring is required. The new design will use less cables and replace soldered-on wires with pluggable connections. Made a new COM cable which connects between on board headers and the stamp board. This time, I used Anita's suggestion to saw out the holes, not drill them (that ripped apart the whole board last time).

Contact lense cases Looked at the one day contact lense packages I use and got the idea "they look neat, and must be good for something". So I cut out three "lenses", and, yes, they might be used for LED covers. Soldered three blue LEDs onto my 3x3 white LED array: nice and bright. So this could light the ship at night from the compass mast (on the other hand, this might screw up the compass reading - will have to see).

AC.motion fact sheet

Computer "Loihi" (IBM-compatible very low power PC)

• VIA EPIA PD 600 motherboard (Eden 600MHz CPU)
  
• 256MB DDR RAM
  
• Samsung MP0402H 40GM hard disk
  
• Windows 2000 operating system
  
• Java SE 5 runtime
  
• Netgear WG111 USB 2.0 WLAN adapter
• Parallax USB servo controller
• Custom software for remote control and autonomous navigation

plus

• Parallax Basic Stamp 2E (20 MHz)
  
• ULN 2803A power driver
  
• Parallax LCD display
  

Propulsion/steering

• Robbe 755/40 12V motor with 5mm shaft and Graupner 70mm four blade propeller
  
• Two Robbe bow thrusters, mounted bow and aft
  
• 70x35mm rudder powered by Robbe FS 251 T(orque) digital servo
  

Power

• Panasonic LC-RA1212PG1 12V/12Ah lead acid battery (about 4kg) for the computer
  
• Two Panasonic RC-R0612P 6V/12Ah lead acid batteries (about 2kg each) for the main motor (12V, in series) and the thrusters (6V tap). Servo power is provided by the main controller (BEC).
  
• PW-60 60W DC/DC converter (12-24V input, 12V/5A output)
• Two Robbe Rokraft 50uP and one Rokraft 100uP power controllers for thrusters and main motor
  

Navigation

• Garmin eTrex 12 channel GPS receiver
  
• Parallax Hitachi HM55B electronic compass module
  
• Parallax Ping))) ultrasound distance sensor
  

Connectivity

• 54MBit/sec WLAN
  
• 4 + 2 USB ports (WLAN, servo controller, nav cam, underwater cam)
  
• 1 + 3 serial ports (GPS, Basic Stamp)
  
• 2 10/100MBit/sec LAN (used in case of WLAN start-up problems)
  
• 1 parallel port (unused)
  

Hull

• 4mm birch plywood with two layers of 160g/m² glas fiber/epoxy resin shell plus white gel coat protective finish
  

Extras

• Cam on computer-controlled mounting (self-made with two Robbe S3001 servos)
  
• Underwater-cam, fixed mounting pointing vertically down
  
• Government-required lighting for motorized vessels :-) (while top ligh, white aft light, green and red lights starbord and port)
  
• Mast-mounted search lights (three 13,000mcd white LEDs front, and one each left, right, and aft)
  
• Mast-mounted effect lights (three blue 5,000mcd LEDs)
  

Thinking about the cutting scheme

Scaled some photos. Completely wrecked the Basic Stamp board in attempt to drill 3.5mm hole into the corners. 1.5mmm worked, but the larger bit cought the surrounding holes and tore the stuff apart. I'm thinking of redoing the board with an improved layout...

Realized that today I have "material-destructive hands", and decided to not cut out anything. Did calculations for the aft part of the underwater body, and got ready for cutting it out of the plywood.

Testing the main power controller, ship-Origami

Took some photos, and lost them all courtesy of the digital camera, which really needs to be replaced ASAP. Just out of pure boredom added a fancy distance bar (made with custom chars) to the distance measuring program program. Created hoods for the compass module and WLAN from Actimel bottles (it's plastic, actually).

Main engine Soldered temporary connections for the 12V battery on the big engine controller and connected it to the servo controller. It powers the servos nicely, and running the motor works as designed, too. We're getting closer...

Actually, a couple of potential problems turned out to be no problems at all. The servo controller, the servos, and the motor controllers work with each well. The 600MHz of the embedded CPU seems to be enough for W2K and the purpose.

Component placement Cut the lower deck and middle deck sections out of paper to get an impression if "it all fits" - it does. It looks like I can reduce the underwater window from 100x100mm to 20x20mm or so, having taken apart the web cam. Seeing the batteries standing so high makes we wonder if I should replace the 12V battery with a lower capacity version. Then I could lay all three batteries under the computer box on their side, lowering the center of mass quite a bit.

Idle

Nothing. Thought about using Actimel bottles as covers for the compass sensor and WLAN USB stick.

For the record: Baatz is open Mo-Fr 9-12 + 15-18, Sat 9-13; Merten is open Mo-Fr 9:30-13 + 14:30-18, not Sat; Sivas is open Mo-Fr 10:30-18:30, Sat 10-14:30.

Ideas, and creating a "CAD" tool

During a meeting at work I had a couple of ideas: underwater helicopter, jet engine driven boat, remotely controlled snow blower on tracks. Thought about lighting ("required by law": white top, white aft, red and green for port and starboard; search lights front and aft; "effect lighting" like 3 blue LEDs or so), air supply (but how to prevent water from getting in too easily), mast for compass sensor and lighting.

Changed the plug on the 12V mains adapter cable, and the plug in the computer box. Wrote BowCalc to calculate the bow measurements bases on four slider values (alpha, beta, width, and height).

Designing the bow

Bow studies in model clay (the one on the left is the winner) Developed formulas to calculate the bow measurements when width, height, horizontal and vertical angles are known. The basic shape will be that of a brick. The bow is created by making a slanted cut each side along a single plane, so that the cuts intersect to form the bow (the image of the paper model explains it best).

Wiring, continued

Created a 3 wire cable with plugs from a new plug with wires and a leftover plug (used hot glue to make cable/plug interface more stable). Recycled the plug and socket conglomerate I fabricated yesterday to connect the 2nd thruster controller. Soldered two 3 pin and one 5 pin sockets to the Basic Stamp board. Added a buzzer from an old computer, and connected it to a Darlington driver. Did some major fiddling with the cables (too thick, too short, not enough space).

Reconnected the stamp to the computer, and produced a nice short when I tried to measure output voltage on the new three sockets. After pulling the 12V supply the system rebooted properly. Tried out the FREQOUT command on the buzzer - works (to the point of getting Annie annoyed).

Tried to connect the Ping module to the 2nd socket - no reaction. Found that a fat solder spot had created a short between the signal and the +5V pin - luckily without lasting effects.

TV playback Watched some Futurama on Loihi while doing all that - PowerDVD makes good use of the hardware MPEG decoder. CPU load is at 80%, but full screen playback works without problems. Playing over the network is not possible, though. VLC might buffer reception problems, but since it does not use the EPIA hardware, playback, well, sucks.

Wiring the power supply, stripping the underwater cam

6V battery with connectors Bought cables, plugs, and sockets for connecting the compass and LCD modules to the Basic Stamp, and engine controllers and got knows what to the lead acid batteries. Bought some more plugs for charging the batteries with my ancient Robbe charger, which with 500mA is "slightly" underpowered for the job.

Tried to solder together five plugs with a 75W iron to make a multi terminal connection, and ended up with a smoking lump of solder and metal that went into the trash. Tried again with some wire, and got two six-enders for connecting to the 6V battery. Took me two hours, for chrissake!

Engine power controllers Put some plugs on a bow thruster engine and one of the small controllers (Robbe 50uP), rigged up the whole thing with battery and the servo controller, and - voila! - it works. The controller does a really good job with low power setting; the propeller starts turning very gently. I feel relieved, too, that the controller works like a charm with the servo controller.

At the model shop (Baatz at Münster) I bought two servo cable extensions for the cam turret, two brass sprockets for constructing a 1:4 servo gear (this will be the answer to "how do you get 360° ultrasound sensor rotation from a 90° rotation servo"), a 5mm shaft, and a engine/shaft coupling at Baatz.

Disassembled the Logitech web cam, which consists of a base (empty, apart from a brass weight!) and a swiveling head. The board is bigger than that of the Creative cam, but mounting is easier because it's not being held from the side but it has two decent mounting holes through the board. After removing the casing the cam is way smaller than before, and can be mounted directly above the plexiglas window.

Assembled the main propulsion system and mused about how long the shaft has to be to prevent water from leaking in. Background: grease in the shaft prevents water from pushing in. What concerns me is the length the shaft musthave: if I cut it too short the resistance from the grease might not be high enough.

Since I want the aft thruster far back and the main engine needs to be the last item on board, the shaft has to be very short. But, hey, I've got another idea: I mount the thruster engine off center (should not make a big difference), mount the thruster as the last thing, and use a medium length shaft to cross over the thruster tube. Sounds like a plan!

Fiddled a bit with the cam turret (shortened shafts). Drilled holes for cables, and put some grease on the main hub. Since I had no wires for connecting servo power I could not try out the result.

Testing the compass module and software

Electronic compass sensor Took apart the LED assembly and made yesterday and connected the compass module to the Basic Stamp. Was impressed by the sensitivity of the sensor, but it needs some serious calibration. This is were the software comes in. It took me the better part of the afternoon to get the compass software going. At some point I realized I needed to obtain a newer version of the stamp editor to get PBasic 2.5 benefits. After that the software worked as advertised. Nice: used the LCD display to display the angle.

Testing the LCD and Ping modules

Measuring distance with ultrasound
Spent the afternoon trying out the LCD and ping modules, and playing with the darlington array and red, green, blue LEDs. The LCD worked right away, and so did defining custom characters. The interfacing is ridiculously easy! I noted there is a Basic Stamp chip on the LCD and on the Ping module, which explains a lot.

The ping (ultrasound ranging) module worked right away as well. The accuracy is down to the centimeter, but it does not "see" round things such as the lamp.

I connected the red, green, and blue LEDs to the Darlington array, and connected the array to the Stamp. Focused the LEDs on one spot. The idea: mix the three components to create any hue I want. I tried to dim the LEDs though pulse width modulation, but, alas, the Basic Stamp is too slow for that. As soon as I try to get more that 6 different levels of brightness or so, the pulses become visible.

Apart from that, I possibly fried the red LED a wee bit by soldering it onto the board while the juice was on (it got grounded without a resistor - nice and bright, but too much oomph). The light cone is wider, too. The net result: the colors don't mix well.

The Parallax hardware arrives

Basic Stamp (red) board
The Parallax shipment (via Visual Data) arrived two days ago and was picked up - thanks - by my landlady. Since I wasn't home yesterday, I picked up the parcel today. Aahh, Xmas! The Basic Stamp is really tiny - though it consists of four chips on an epoxy board it fits into a 25 pin DIP socket. I fired up Loihi, started the stamp editor software, and the stamp was identified right away. Tried out a small program - works. It's not always like that...

Unwrapped the LCD display (contains a Basic Stamp chip), the Ping sensor (ultrasound distance measuring; contains yet another Basic Stamp chip), the compass module (tiny 6 pin DIP epoxy board), and - tadaa! - the USB servo controller. Wired the controller to a 6V battery, attached the servos, installed some USB COM drivers and the controller software, and - it works!

Used 1mm steel wire to build the connection between servos and devices. Was very pleased to see that turning the cam turret works as designed - I wasn't sure until now that the mechanism I came up with would actually work. The next thing would be finding a place for the wires (USB, 2 servo cables, cables for LED power later).

Picking up batteries and rudder

Shaft and propeller A colleague at work - Christoph - invited me to his BBQ party tonight, which is nice, and what is even nicer is that he lives at Emsdetten, gave me a lift to his place, and stopped over at the model shop for me. There I got the lead acid batteries (1 x 12V/12Ah, 2 x 6V/12Ah) and the rudder, setting me back by EUR80,-.

To be fair, that still was a deal, but I inhaled sharply. I'm not so sure any more if I need the capacity, and whether the computer and propulsion couldn't share power. On the other hand: the engines probably introduce a lot of erratic signals on the power supply, and I might want to execute longer autonomous cruises. So let's see how it works with weight and volume. 8kg of batteries will make the ship pretty heavy....

The shipment from the model shop arrives

Power controllers Monday a guy from the LMZ model shop at (meanwhile) Emsdetten called me; the shop will ship my ship parts shortly :-). Tuesday the stuff arrived, but since I had forgotten to leave Annie some money she could not accept the parcel. At the post office I got told I can't have it before Wednesday :-(

Main engine
So I picked up the stuff today, and from the weight of the parcel I immediately knew at least the lead acid batteries were missing (and the rudder, as it turned out). What I got: the Robbe 755/40 main engine (nice format), the Robbe 100uP power controller, two Robbe 50uP power controllers, a Robbe digital servo for the rudder, and two bow thrusters (one will actually work as a stern thruster). All in all EUR284,-, which is over EUR40,- cheaper then what the Robbe caralogue says. Nice discount! "Come again"...

Started assembling the thrusters, but found I need oil for the bearings and grease for the rubber fittings. Installed some software on Lohi for this and that - AC.motion probably will be the only model ship with OpenOffice on board :-)

Insights into IDE and LED

Loihi in the box Did some research today.

Can a 80 wire IDE cable be shortened (as in: can I cut off the second device connector)? According to the ATA spec, a cable has two device connectors (but cables with only one are sold). The minimum length for one segment is 127mm, and the first segment must be equal or longer than the second. So: I'll backup the data, cut off the second segment (the first is than longer for sure), and see what happens (it works).

Do I need to limit the current for white LEDs? Merten told me that up to 4.5V I won't need a resistor. Bzzzt - wrong! There is no such thing as "internal resistance" since a LED is a semiconductor, and once the "forward drop voltage" (the voltage the LED needs - and eats! - at mininum) is exceeded the current might rise exponentionally with voltage. White LEDs have a 3.4V dropoff voltage. With 6V and a rating of 20mA, the following resistor is needed: (6V - 3.4V) / 0.020A = 130 -> 150Ohm. For 12V that would be 430 -> 470Ohms, or 260 -> 270Ohms with 2 LEDs in series, or 90 -> 100Ohms with 3 LEDs.

How can I drive LEDs with the Basic Stamp? Parallax offers a Darlington Array at $1.50 which directly connects to the Basic Stamp and is able to drive up to 500mA on each of its 7 outputs. Neat! Maybe Merten has it. Phoned. He has! EUR0.95... So, now I can switch my LED lights on and off computer-controlled.

Cut off IDE cable after 2nd connector and folded it as neatly as possible. System still boots, have to check if BIOS still believes it is an 80 wire cable. Tried out the computer box closed. The fan is pretty noisy, even more so with the lid on. I need to verify this with a thermometer, but I get the impression cooling might be quite good because passages between heat sinks and box are pretty narrow (funnel effect).

Connected the (screaming bright red) sleep LED. Was not really needed; I saw the power LED flashes when the board goes into suspend mode. Finished one more flat serial cable for connection with the on board headers. Read up on header pin out, and soldered a flat serial cable to the Basic Stamp board. Will need two capacitors for finishing that part, though.

Wiring the computer

Wiring Loihi (onboard computer) I went to the electronics shop to get a small fan, some LEDs (white ones, 13.000mcd, now sell for EUR1,-), struts, and some plugs. Epoxied the struts into the base board, attached main board and DC/DC converter (there is a bit of tension I'd like to do without). For the hard disk I still have no solution other than attaching it with screws from the bottom.

Installed the mini fan only to realize it's pretty deep and interferes with the hard disk (it only fits at an angle). I don't feel like stacking components since I'm concerned about thermal issues (the CPU, bridges, DC/DC-converter all radiate off a fair amount of heat). The whole box is getting a bit crowded with all the cables, which wouldn't matter if they wouldn't block air flow. Though I don't really feel like redoing the box, but I might want/have to. Darn, it just needs to be 1cm higher and 5cm longer.

Started putting together the board for the basis stamp. Attached a red LED as a "sleep mode" indicator. Prepared another flat cable for the on board COM connectors (now I've got three).

Component	Connected to	Via
WLAN	USB 1	USB cable
Servo controller	USB 2	USB cable
Turret cam	USB 3	USB cable
Underwater cam	USB 4	USB cable
Compass module	Basic Stamp	Five wires
Ultrasound sensor	Basic Stamp	Three wires
LCD Display	Basic Stamp	Three wires
GPS	COM 1	Serial cable
Basic Stamp	COM 2	Flat cable
[Open]	COM 4	Flat cable

JMF cam image capture woes (mostly solved)

With the W2K "Imaging" application I can choose between the two cams, "Creative WebCam 3 or WebCam Plus" and "Logitech Camera" (which I installed first). When trying to capture an image, a preview application (specific to the camera) shows up in both cases, but only the Logitech works. Since I only had installed the drivers for the Creative cam I installed the Web Cam Control app as well. After I rebootet I installed the update, too. Capturing images with the app that comes with the cam now works, but not with the "Imaging" app.

Results of my research. VfW (Video for Windows) is old. WDM (Windows Drivers Model) is new. There seems to be a WDM wrapper for VfW, which makes WDM drivers appear as VfW drivers (in that order). The cams are not shown as an option by the USB device manager in the system tray. The Creative image capture software works with the Logitech cam, too (!). Works the other way round as well (!!). Both cams show up under "Bildbearbeitungsgeräte" in the device manager. Only the Creative cam shows up as a scanner device (WDM). The button on the Creative cam works, the one on the Logitech doesn't. The JMF registry detects the Logitech as "vfw:Microsoft WDM Image Capture (Win32):0" and the Creative as "", ahm: not at all! After plugging them in both, the Creative is shown as "vfw:Video Blaster WebCam 3/WebCam Plus (VFW):1". Does that make me happy? Not yet...

Tried out JMF Studio (comes with JMF) for the heck of it. To my surprise both cameras are now options for video capture, and it works, and (at 320 x 240) with only 20% CPU load. What the?! So: why does the stuff work with JMF Studio and not in my program!! Holy S.: after copying the properties file created by the JMF registry into the workspace project folder things suddently work... Well, thanks for sending me circling for 4h!

Weird: After starting Eclipse 3.1 suddently it didn't find any imports from the RXTX lib any more; after shutdown/restart it did. Still, I get an error now that the device object can be created. Some "jmvfw" lib is missing, and Google does not offer much on the subject. Resorted to defining an Eclipse user lib containing the JMF JARs and setting the "native library path" of that lib to the JMF dir. Works!

Now the well known JMF problems start: both VfW devices seem to refer to the same camera. More on that some other day...

Checking out the WLAN

Experiemented with the WLAN connection. Found that W2K (a way bigger troublemaker than Y2K ;-)) sometimes entirely looses the WLAN connection when I'm using TKIP and an access point; maybe the regular key changes are somehow indigestible for the Netgear USB stick/software. So I set up an ad hoc network between Maunakea (the laptop) and Loihi (the ship). Had the usual problem with pass sentences not working; switched to specifying a hex key on both sides which did the trick.

Now I should verify what I should have checked from the very beginning: whether the WLAN stays connected reliably and whether the ping time does not exceed 100ms. A bit funny to test that now, since the whole project depends on it. But I'm quite confident; if all goes wrong I'll just use an unencrypted network, and a wire LAN connection for "cold starting" the connection before I launch the boat (I have the impression sometimes the systems don't connect easily, requiring starting and stopping the wireless connections on both computers). Maybe I should add a fallback routine to the onboard software that when no heartbeat message has been received in the last 5min, the ship moves for 5 seconds and then waits for 30secs. Thus it might slowly move to a shore where it can be picked up. Fancier even: it uses the GPS to go back to the starting point.

Experiments and improving GPSLib4J

Garmin Foretrex 101 Started to work with GPSLib4J. First got some NPEs when I added a general packet listener, but no time or position listener. Found the lib makes too many assumptions... Had to modify GPSLib4J in minor ways: the Garmin packet listener now sends specific packets instead of the raw version; improved text output of PVT packets (the Garmin "all info in one" packet); improved text output of position; improved documentation (which, frankly, is a bit sparse to say the least. Added info from Garmin protocol spec to the comments).

Found that position, altitude, speed etc. varies all the time, which is good. Why? It allows me to set up my own "differential GPS". How that? Knowing my laptop does not move I know that all skipping around is due to artificially introduced and other errors. I know precisely where I am, too, if I look up my position on a map. Then I select one data set as a reference on the laptop and submit the delta to to ship every second. The ship then corrects its own data. Hm, this only works when a) both GPS receivers get cheated the same way (for a point in time t the errors on both devices are the same) and b) receive the data in sync (or is the time stamp exact enough for matching packets?). Anyways, it's an idea, which I have to try out.

Creating the software, I

Played with GPSLib4J until the GPS ran out of power, which was pretty soon. At least I know Loihi is able to connect and receive information. Ran into some NPEs depending what kind of listener I was adding. This is a bug in the lib; luckily it comes with the source so it can be fixed (the blessings of open source...).

What I could not get to work even with looking at the web quite a bit was capturing images from the web cams. I found that the Java Media Framework (JMF) is able to do the trick in theory. The JMF Registry indeed shows two devices for image capturing, but when I use the names detected by the registry to obtain the device objects, I'm unable to get them (just get a null). Some more experiments needed...

Interfacing GPS, cam, and COM ports

COM/Foretrex adapter I built myself an adapter to connect my Garmin Foretrex 101 GPS receiver to the serial port (much cheaper than buying the cable). Thanks to GPSLib4J and the RXTX serial port API I now have real easy access to GPS information!

It took some figuring, though: it took me some time to realize that

• the GPSLib4J sample code expects the Java Comm API
• the RXTX library (at least the version I tried) comes with the Java Comm API included but
  
• the Java Comm package name has been changed to gnu.io, therefore
• the sample code has to be modified and
• in Eclipse, the GPSLib4J user library I created needs the "native library location" path set to the directory the RXTX DLLs reside in, otherwise the runtime will complain later.
  

But then the self-made adapter and the sample code worked like a charm. I was pleased to find out the Java Media Framework (JMF) API allows for capturing web cam images, though I didn't try that out yet.

Since the RXTX API allows for accessing COM ports, interfacing the Basic Stamp shouldn't be too hard either (touch wood).

Connecting power, computer box II

This didn't work I started to cut out plywood pieces for the computer box. When I attempted to combine the wood pieces with each other I realized I would have a metal base plate, and somehow attached to that, a wooden frame - and a loose lid. This does not look good or stable... I looked at the metal base plate and, after some thought, threw it out. I started feeling better after I completed a wooden box, 300 x 180 x 50mm, that a) is stable b) looks decent and c) is usable.

I spent two hours on soldering out the connector on the DC converter, cutting off one plug of the ATX adapter cable, and soldering the cable directly on the DC converter, eliminating the 80mm "tower of power". The soldering holes are just a wee bit too tiny for squeezing through the soft wires. After putting some solder on the wires they did not fit "cold" :-( Then it prooved very trying to heat up holes and wires at the same time. I ended up with connections I am not proud off, and, as I said, it took way more time then it should have. That was it for the day!

Building the computer box

I went to Merten at Münster (good old-fashioned electronics shop) and bought a power meter (at EUR32,- pretty expensive, but with a couple of neat features), an experiment board for the basic stamp, and some struts for mounting the mainboard. The "pin-plugs" I bought are way too big for the USB/COM headers on the board.

Found that power consumption is 20W with WLAN. Not bad! Under full CPU load this goes up to 27W. The TV card uses 7W on its own. The laptop, BTW, uses 60W when playing a movie. Whopper!

Took me a long time to think of just how to build the box, and how to fix the struts to the box. Furthermore I don't yet how to mount the harddisk. I'm not quite happy with the design; bending rather long sheets of metal with no knowledge or special tools just results in ugly wavy and round "corners". Lots of sawing with the manual jig saw, too; at the end used the power saw, and: the cuts look quite nice with the new metal blade.

Fiddled around with struts that I cut and shortened and modified to attach the main board and the DC-converter to the aluminum "box" (base plate with sides bent up). I should have stuck with the original idea (just a base plate, and wood for the rest), and might just go back to that when I build the wooden enclosure. The general problems are stability, cooling, water proofing, and wiring.

Replaced the 40 wire IDE cable with a 80 wire cable - seems to work. Found the ATX connector cable stands 8cm high (!) when plugged into the DC-converter, and the cable is very stiff :-( Will probably solder out the connector, cut off the plug from the cable, and solder the connector cable directly onto the DC-converter. The converter, BTW, does directly plug into the main board, but it interferes with the PCI slot then. Clever design, not.

Installed more software, and studied the R&G epoxy catalog that arrived today (resins, glas fiber fabric, dyes).

Setting up the computer

Loihi rigged for setup Today the AC-adapter (220v to 12V, 5A), DC-converter (12V to ATX), IDE-adapter (3.5" to 2.5") and ATX extender arrived, which enabled me to set up the computer ("Loihi", to keep in line with my Hawai'ian volcanoes naming scheme). After verifying the HD is connected correctly (with the cheap supplied 40 wire cable :-() I installed W2K. The whole system is so quiet - no moving parts except the HD. The CPU gets pretty warm, so some additional fan might be needed.

I installed the J2SE SDK 5.0, the web cam driver, the wireless driver, and some more. I got some resistance since I didn't have any SP ready, and I know some software will not complain but simply not work if SP 4 is not installed - a to do for tomorrow.

The wireless network (USB) worked without a hitch. The web cam worked, but very jerkily and with horizontal stripes in the recorded movie (stills are OK). Hm - I hope I can find some alternative software to get better results than this! Playing back recorded TV shows (this obviously is not a requirement for the ship, but for use at home) with VLC is not very smooth either, but with Power DVD it is - maybe the latter can make use of the hardware MPG decoder. I could not get the Hauppauge PVR350 driver to install, but if I remember correctly this will be solved by installing SP4.

As it is with setting up a box, this took me over 3h... Good news, too: the model ship finally got back to me, not to deliver items but to make sure they order what I want. Arghh... Well, this time I hope it works, and I might get the stuff by the end of the week.

Finishing the cam turret, Baumarkt III

Anita and I rode to the Baumarkt and this time I bought some 4mm plexiglas for the glass bottom and metal saw blades for the jigsaw (hopefully creating less splintered wood). With a few screws and wing nuts I hope to convert my jigsaw into a small table saw. How that? Stand by...

Cam turret I pretty much finished the cam turret by drilling holes for the cam cradle, creating spacers from aluminum tubing to keep the cam centered, and custom screws made from threaded round iron to hold the whole thing.

Cam chip mounting I found out, too, that the sunshade can be removed from the cam, making it quite a bit smaller. Actually, at that size I could have made the turret smaller as well.