Saturday, October 10, 2009

Making a new rudder, part 3

Designing a rudder, part 1
Designing a rudder, part 2
Making a rudder, part 1

After lots of careful shaping of details (mostly with an orbital sander) and filling in the odd accidental gouge (using a mixture of epoxy and colloidal silica - which sets into a very tough and water-impervious substance), I was nearly ready to laminate.

From Laminating new rudder
Hardwood inserts epoxied in place, ready to take bolts

The final pre-lamination step was to drill big holes where the bolts connecting rudder to yacht will later sit, out to a radius of 5 times the bolt size: since Corecell is too soft to hold stainless steel bolts, I fixed hardwood inserts (chiseled from the remains of my old rudder) in place with lots of epoxy/colloidal silica mix to fill the gaps (actually, the wood was not technically necessary - but cheaper than epoxy, and a nice link with the old rudder, so in it went). Now, my rudder is a wood/corecell/epoxy/colloidal silica composite - and we've not even started with the glass fibre yet.

From Laminating new rudder
Rudder core, fully shaped, smoothed, faired, and ready for lay-up

Next step was to cut glass cloth to size; this is beautiful stuff - layers of four-harness satin alternating with heavy unidirectional fabric. It isn't that easy to cut, especially the 500g/metre stuff - glass is really hard. Also, the material looks almost too beautiful to use.

From Laminating new rudder
Unidirectional glass fabric, 500g/square metre

Before laminating, I need to don my protective gear - epoxy is great stuff, but not for the human body. I always wear eye protection in case of silly accidents, nitrile gloves (I'm now using cheap disposables instead of the heavy duty items pictured - latex won't do, by the way), rubber boots (because there are always drips and spatters), overalls (well, duh!) and an apron (because I tend to lean against the workbench, and my overalls nearly soaked through once).

From Laminating new rudder
Ready for Fun with Chemicals

From Laminating new rudder
Wet fabric (transparent) draped over the rudder core

My first couple of layups went pretty well - except I laid the epoxy on way too thick; extra does not add extra strength, only weight - and it is pricey stuff to be wasting. In the picture above, note the puddles of excess epoxy sitting on top of the cloth, and the messy ends at the rudder head (beside the squeegee) that I had to grind off after the resin had set. Hard to see, but important: the plywood workbench has been covered in white-faces hardboard (cheap enough to bin later) which in turn has been covered with a transparent polyethelene sheet (does not stick to epoxy).

Miscellaneous tips:
  • Clean your squeegee straight after use, while the resin is still soft
  • Buy a set of cheap brushes from Tesco for dabbing on resin wherever dry spots show up (€1.25 gets you 3 brushes in our local). Ditch after use (too hard to clean).
  • Disposable nitrile gloves give great tactile feedback - actually better than the heavy kind (plus, no clean-up). Very, very cheap at B&Q - and insanely cheap on E-bay.
  • It is much easier to use too much resin than too little
  • ...but you still need to keep an eagle eye for dry spots.
  • take your time and do careful work - resin sets fast, but not crazy fast
  • use peel ply - leaves a lovely smooth easy-to-work with surface, helps remove excess resin, well worth the money
  • an ounce of preparation (masking tape, polyethylene sheets, etc.) saves a ton of fixing up later

Thursday, October 8, 2009


Our son has reached The Age of Exploration. At last, he can pull himself upright and toddle along on his hind paws - so long as a support of some kind remains within reach. To give him some walking practice and a little independence, I sketched out a pushing/walking toy, then started digging through our wood-pile. Soon, odds and ends had been sawed and screwed into something closely resembling the sketch below (the height for the bars chosen to be just around our son's shoulders, low enough to be pushable, too high to fall over). One point to note: sanding off all edges and any rough surfaces will save grief later. With even a very cheap random orbital sander, this does not take long.

Version 1: Sled with upright handle

Full of anticipating, I presented the result of my labours to my son, who got the idea of the toy straight away, and set off at high speed, cackling with glee... and seconds later, rammed it a cupboard dead-on. No damage done, but also, no possibility of turning. The next problem was that he stood inside the sled, and tried to pull it over his own toes.

Version 2: platform to stand on, double-ended

I decided to solve the turning/reversing problem by making the pusher double-ended, duplicating the existing upright handle, and solved the standing-in-the-middle problem with a plywood platform. As I finished screwing this on, it occurred to me that I should have used bolts to secure the uprights to the base - then it would be very easy to take the whole thing apart into flat pieces for storage or transport.

The modifications were well received by my tiny test pilot: he used an upright to pull himself to his feet, then clambered grinning onto the platform, and tugged mightily at a handle, doing his best to rock the whole contraption. Which gave Dad an idea.

Version 3: build-in rocking function!

The final (so far) modification required some fairy delicate wood shaping: I made a long shallow curve from a single piece of wood, sawing, chiseling, sanding, then split it lengthwise into two identical pieces, which I glued with ordinary wood glue (do not risk snagging a screw head on your partner's tiles/carpet!) carefully to the existing runners, then clamped and allowed to set overnight.

Next morning, a very proud Dad presented his soon with the push-me-pull-you version. Son discovered that the new version made a very pleasing racket as it rocked on the kitchen tiles, and Dad was relieved to find that he hadn't overdone the rocking motion (no danger of head-over-heels).

This has been my first real foray into toy-making, at least as an adult; my son has got a lot of fun and exercise from a very simple toy, and I just as much, from the pleasure of making, of giving, and of seeing his fun.

Dads, Mums, Aunties and Uncles: to your sheds! Go build!

Sunday, September 20, 2009

Making a new rudder, part 2

This post is part of a series on making a fibreglass rudder with a foam core:
Designing a rudder, part 1
Designing a rudder, part 2
Making a rudder, part 1

Calculations done, foam cut into neat rudder-sized rectangles, I had no alternative but to start shaping the foam. I did it like this...

Step 1: mark the depth of foam to be removed. I did this by setting a circular saw to the calculated depth for a given point in the profile, then running it the length of the rudder. I kept the trenches very close together at the leading edge of the foil - the first few tracks were only 2mm apart (the saw blade width), then, as the slope of the foil changed more slowly 4mm, 5mm, 10mm. On the rear half of the foil, whose slope is almost straight, the gaps increased to 20mm, then 40mm. With all the tracks cut, I sprayed blue paint into the cuts, making sure to get good coverage on the bottoms of the tracks.

Step 2: remove the bulk of the foam. I tried chisels: bad idea. What worked really well was to take a wood saw, hold it sideways and cut along the foil at a shallow angle, keeping a millimetre or two above the bottom of the trenches. In about 40 minutes I was able to remove most of the excess foam, leaving behind a nice flat surface.

Saws make short work of Corecell

Step 3: the saw was followed by a wood plane, which cut to within 1-0.5mm of the trench bottoms - but not lower, because the plane did not cut the foam as smoothly as wood, tending to leave the surface a little rough.

Step 4: I next used a surform to remove the last of the excess foam, leaving the surface of the foil flush with the trench bottoms.

Step 5: To get a really smooth finish, I followed up the surform with a random orbital sander.

With the two halves shaped and (I hoped) pretty symmetrical, I mixed up 250mm of epoxy, then stirred in enough colloidal silica (amazingly fine white powder) to make a paste with a honey-like consistency. A squeegee was perfect for getting an even spread on the flat side of the starboard foil half. Interesting to note: although the foam looks like a sponge, it certainly doesn't act like it - it doesn't the epoxy in, but leaves it on the surface - where I need it .

Rudder core halves clamped together; polythene drapes keep epoxy off the worktop

With an even spread of epoxy covering the starboard half, I dropped the port half into position, and applied my entire collection of clamps. You never have enough clamps. And that was it for the day: all going well, by tomorrow morning, the two halves will be one - permanently.

Friday, September 18, 2009

Long Ears

In the morning, there was a small brown scrap on our lawn where none had been before. Turning over, as I thought, the corpse of a headless bat, I was startled by squeaks of fear or rage. The mouse-sized animal had curled its wings in tight around its body, and tucked its head into its chest; even its ears had curled up, looking something like a ram's horns. The face was relatively pretty, as bats go, without the weird pressed-up nose that some species have; the massive ears are the feature you'll remember, though.

Brown long-eared bat

According to the excellent "Exploring Irish Mammals", Plecotus auritus, described by Linaeus in 1758, is found across most of Europe north of the Pyreness, and also in the far east, in regions within Russia, China, and Japan. This bat likes to "foliage glean", plucking its prey from leaves or even the ground as it flys.


My tiny specimen was soaked in dew, and motionless - this last must be what saved it from our cat, who loves to chase, but tires of prey which can no longer flee. I tucked my patient away for the day in an empty toolbox, with a few crumbs of cat food for sustenance.


After sunset, I took my patient for a walk, to a wooded and cat-free place. Out of the toolbox, the ears inflated and the head looked up; then, the tiny creatures crawled from my gloved hand onto the rough bark of a large tree, hooking on tight with claws on the leading edge of the wing, while black and hand-like hind paws sought purchase.

Tuesday, August 25, 2009

Making a new rudder, part 1

Following on from the previous posts on the design of rudder foils, our hero now begins the tricky fabrication stage.

As discussed previously, my new rudder will have a fibreglass skin (6 layers) wrapped around a foam core. Not just any old foam - Corecell, a light, stiff material that is absorbs little resin (or water) and will contain any damage to a very localised area. It won't rot, delaminate or crack, and the sheet I bought from the helpful folk at MID has a density of just 60kg / cubic metre.

I've never used this material before, so wasn't sure how easy it would be to work with. First, I trimmed the lengths I needed for the rudder itself from the original 4ft by 8ft x 25mm sheet using a jig saw. No problems, and surprisingly little dust - most of the material from the kerf seemed to stick together, hinting at one potential difficultly: since this foam is a very good insulator, heat from cutting tools disspates slowly - so cutting at higher speeds can cause melting.

Next, I used an off-cut to practice my foam-shaping skills (never shaped foam before, didn't want to start by destroying my proto-rudder). First, I clamped a straight-edged piece of wood atop the foam as a cutting guide (straightness verified against a handy glass window - float glass is very flat). Then, I set the cutting depth on my hand-held circular saw and cut a series of trenches in the foam, the depths calculated from my spreadsheet describing the NACA 0012 foil I want to make.

Next, I sprayed the yellow foam with blue paint, making certain it penetrated to the bottom of the trenches.
Once the blue paint was on, it was now "safe" to start cutting away excess foam; I tried a random orbital sander, but, especially at high revs, discs clogged too quickly. Tried a sharp penknife: not too useful. A block plane worked reasonably well, and a rasp/file came in very handy for tidying up small irregularities. The paint worked a treat - you can see below how the blue lines allow me to be sure I've not removed too much material.

Shaping the leading edge of the foil was quite easy; shaping the trailing edge - much more material to remove, but accuracy is less crucial - turned out to be more difficult. I actually tried chiseling, which worked up to a point, but was slow and difficult. Not sure about the best way to do this yet - I hope to try an angle grinder soon to slice the bulk off, then follow that with plane, disc sander and rasp for the final finish.

Conclusions so far: Corecell is very easy to cut. Shaping it is much easier than shaping wood, but the same techniques won't work.

This post is part of a series on making a fibreglass rudder with a foam core:
Designing a rudder, part 1
Designing a rudder, part 2
Making a rudder, part 1

Monday, July 6, 2009

Astronavigation / Celestial Navigation, Part 2

Last night, a (briefly) clear sky gave me a chance to try a key element of practical celestial navigation. Following the example of Marvin Creamer and countless forgotten sailors of the pre-sextant era, I attempted to measure the altitude of some celestial bodies using no instruments whatsoever (partly because, aside from my telescope, I haven't made or bought any yet).

Why no instruments? Obviously, this decreases the accuracy of the observations that you can make; not so obviously, it is still possible to be accurate enough to make measurements that are accurate enough to be useful. Marvin Creamer made fairly accurate landfalls all around the world using this method, generally maintaining his latitude to within half a degree of the intended value , and any experienced navigation venturing deep into desert or out of sight of land before the coming of "modern" instruments like the sextant almost certainly relied on such techniques. So, it is possible, but is it useful?

Well, sometimes; today, people like myself who like to use the "road" less travelled have the benefit of tools like GPS (I recommend the excellent, cheap, tough Garmin Etrex); however, the very nature of independent travel in remote places means that equipment may well be lost, broken, or simply out-of-charge. By learning how to navigate without any instruments which aren't permanent (I hope!) parts of my body, I'll be able to travel with a permanent built-in backup for the GPS and the compass (you'ld be surprised how many sailors run into GPS trouble). Apart from hypothetical equipment failures, navigation is a subject that I enjoy for its own sake; there is something deeply magical about looking up into the night sky, knowing the stars by ancient names from foreign tongues, and, simply by looking, to find yourself.

Back to practicalities. At this latitude, the July sky isn't really dark at 23:00, so there weren't very many bodies to measure - simple identification was tricky, because only the very brightest bodies were visible. Even worse, not being at sea, I didn't have a clear horizon either. Nevertheless, I extended an arm and spread thumb and forefinger as far apart as they would go; for the average person, the span between thumb and forefinger will cover about 15 degrees of sky. I levelled my arm at the invisible horizon, and began to measure.

Waving my arm around the summer sky, I measured the altitude of a rising moon (itself covering only 0.5 degrees of sky, a useful checking-fact) at about 10 degrees and the bright orange of Arcturus (Alpha Bootes) at about 38 degrees. I sighted Dubhe (part of the Plough Constellation) too, although I forget the number; then, checked the observed altitudes against what Mobile StarChart app on my phone said they should be. I got the moon nearly dead-on - its real altitude being about 10.5 degrees - and Arcturus turned out to be just over 40 degrees above the horizon. The error for Dubhe was higher, about 3 degrees.

For a first attempt, with no clear horizon, these measurements strike me as acceptable; if I had been using them to find my latitude, I would've had a pretty fair chance of finding my home country. Marvin Creamer did a lot better, though, so the next time I've got a clear horizon after dark, I'll give the procedure another try. In the meantime, I'm now wondering what simple instrument I might be able to construct to get those errors a little smaller. Ideally, these will be simple instruments, such as a lost sailor on a small boat might plausibly fabricate. Also, wouldn't it be nice to measure latitude and longitude?

Coming soon...

Tuesday, June 23, 2009

Designing a new rudder, part 2

By measuring photographs and doing a little maths, courtesy of the NACA 0012 formula, I was able to discover what size and shape my new rudder (transom-hung, constant foil cross section) should be. Next problem: what materials?

My old rudder was a single piece, almost certainly a tropical hardwood, dense and strong. Marine plywood was an obvious option, except that no local suppliers had real marine-grade plywood (non-marine grade, which I have used for cabin furniture, can have voids and gaps within interior laminates, hidden weaknesses which could be fatal in a load-bearing structure). My old rudder was a single piece of tropical hardwood, which I also tried to source: a local supplier had lots of beautiful teak and iroko, but no planks wide enough to make a rudder in a single piece - they simply aren't there to be had (they've all been made into rudders already?). It might be just as well: a solid piece of timber can be sundered by a single stress-grown crack, not such a problem with laminates.

Abandoning nature, I talked to the very helpful Liam Phelan of, and began to investigate the possibility of a synthetic foam core (Corecell) wrapped in fibreglass. To get an accurate spec on which foam and how much glass, Liam suggested I talk to Martin Armstrong, chief technologist at Gurit, a firm which supplies composite materials to pretty much everybody who builds composite structures - submarines, wind farms, huge racing yachts, aircraft, etc. Martin is a busy guy, but he spent half an hour talking an amateur sailor and novice builder through the materials and techniques necessary to fabricate a composite rudder.

First, the core: A550 foam (Corecell) for the rudder core; a single 8ft x 4ft x 25mm sheet would suffice. I wasn't sure how easy this would be to shape, but Martin reassured me that it is far less dense than wood, while also having no grain; normal wood working tools would suffice, it could even be sanded into shape; a surf form might be handy. Only one problem to watch: being an excellent insulator, it is really bad at dissipating heat, so power tools should have fresh, sharp blades to minimize friction.

Layers making up my composite rudder

Then, the exterior, from which will come much of the strength; Martin specified six layers of glass cloth:

Layer 1: 290g 4-harness satin, laid at a 45 degree angle, and with a 100mm overlap both sides at the leading edge, and a similar tail at the trailing edge
Layer 2,3,4: uni-directional 500g fabric running top to bottom (no overlap)
Layer 5: 290g 4-harness satin, 45 degrees again
Layer 6: 290g 4-harness satin, 0 degrees

This post is part of a series on making a fibreglass rudder with a foam core:
Designing a rudder, part 1
Designing a rudder, part 2
Making a rudder, part 1

Tuesday, June 16, 2009

Car cleaning tip

One of our cars has a light-grey plastic trim in the interior all around the ceiling. Looks nice and bright, but not so easy to keep clean. Today, we found the solution: all the black grubbiness around areas that hands touch a lot (e.g. the sunroof controls) was removed instantly by the simple application of a popular brand of baby wipe. Looks good as new now. Excellent stuff.

Monday, June 8, 2009

Astronavigation (Celestial Navigation) for Beginners

Knowing one's way around the night sky is a useful thing, if, like me, you have a telescope and want to know where to point it, or if, like me, you have ambitions to learn astronavigation. Until last week, sunset came early enough that I could get a few minutes of practice on every clear evening, standing in my garden and counting out the stars. This time of year, the orangey-red light Arcturus is usually the first that I see; the distinctive blue blaze of Vega is to the east, and when the sun's glow has faded a little more, Pollux, Castor, and Capella (actually 4 stars, an exotic double-binary) show up nicely.

The stars that I am really watching for, though, are Polaris (the north star) and Etamin; obviously, Polaris is very useful, in that it gives a navigator a course to steer anywhere in the northern hemisphere above maybe 10 degrees of latitude (ish) - but why my interest in Etamin (gamma Draconis)? Well, it so happens that my home port on the eastern seaboard of the north Atlantic is just a smidgen north of Etamin's declination (celestial latitude), which is 51 degrees, 29 minutes, 20 seconds. Now, Polaris has the useful feature of always (where always = "several hundred lifetimes") being 51 degrees and X minutes above my local horizon; Etamin, by contrast, whirls around the sky, never dipping below the horizon, but once per day passing through the zenith - what you might call "Etamin-noon".

In practical terms, this means that were I some day to be lost in the blue vastness of the North Atlantic, no GPS, compass, sextant or chrometer to guide me home, I could use Etamin to find the latitude of home, sailing north if Etamin passed north-of-zenith, and sailing south if it passed south-of-zenith. Once at the right latitude, I would need only to keep an easterly course, and a sharp look-out for pointy rocks. Of course, measuring the fixed, non-whirling altitude of Polaris is more convenient - it can be done whenever Polaris is visible - but that would require an instrument, ideally a sextant. Marvin Creamer, an American amateur sailor and retired professor of Geography, once sailed around the world on Globestar using techniques like this and no instruments whatsoever, making surprisingly accurate landfalls.

Unfortunately, during part of the year, Etamin-noon would fall during daylight hours - but even then, other bright stars at similar latitudes could give useful hints. Which bright stars pass directly over your home port / next port? Just follow the linked query at Wolfram Alpha to see a table listing the hundred brightest stars by declination, and you'll soon be on your way. A useful tool to help you practice is the (totally free) Mobile StarChart, a java applet you can install on your mobile phone - it only has about thirty star names, but is open source, so you could add more.

Living a long way from the sea? Astronavigation can also be pretty useful in the desert, and was much practiced by people like Popski. Must learn how to use a sun-compass one of these days.

Saturday, June 6, 2009

Designing a new rudder

So, our beloved Briongloid, a 6.6M fin-keeled sailing yacht went adrift from her mooring, and spent an uncomfortable day bouncing on pointy rocks. The pounding reduced her wooden rudder to matchwood - so it's time to make a new one.

How big, and what shape? From a profile scale illustration of a Pandora International (our boat's model) I figured out the height and width - about 1.65 metres * 0.37 metres. Now, I just needed the cross-section's shape.

It turns out that the best shape for a rudder is a foil - like the shape of a bird or aircraft wing, the magic of the foil shape is that it generates lift (unlike, say, a flat surface, which only creates drag). Back in the 1930's, the boffins at NACA, the forerunner of NASA, investigated different foil types to find the best shapes for different aeronautical (and incidentally marine) applications.

For relatively slow-moving displacement craft like our yacht, their "NACA 0012" foil is the best fit; by creating a Google Calc document based on the NACA 0012 formula, I generated the cross-section above (y and x axes are not in proportion). Note the very round leading edge and thin trailing end.

Many fins and rudders taper from one end to the other, and give the leading edge a crescent profile; this tapering reduces drag by about 4% - for me, not worth the much-increased difficulty of shaping the foil.

This post is part of a series on making a fibreglass rudder with a foam core:
Designing a rudder, part 1
Designing a rudder, part 2
Making a rudder, part 1

Tuesday, April 7, 2009

Review: Imray Chart Plotter (ID30)

In the current economic climate, the demand for my services has decreased; so, less cash. Luckily, this also means more time for sailing. I'm planning a little cruise westward, which (winds permitting - hah!) will take me off the edge of my chart folio. To help my planning, I bought my first digital charts and charting software, Imray ID30, covering the west coast of England, Scotland and Wales, and the whole of Ireland. Usefully cheaper than their Admiralty equivalent (and discounted further for this Euro purchaser by the weakness of Sterling), the CD arrived promptly from those nice people at Marine Chart Services.

On first running the software, users need to register; the first step involves giving an address that must include a postcode. The tiny country I live in doesn't have those anywhere outside the capital, and even then, they aren't in the UK format which this officious little dialog demanded. After some trial and error, I discovered that "--- --" was an acceptable location(!).

Lovely coordinates, but what's the distance?

My first impressions of using the actual chart plotting software itself are that opportunities have been missed - for example, a simple tool like dividers doesn't work quite as well as it could. In the above example, I'm trying to measure the width of the anchorage at Port Magee (on Ireland's lovely south west coast). Ideally, the measurement text box would come out from its hiding place - an easy fix could be "borrowed" from CAD tools and their dimensioning widgets.

A bigger problem is (how ironic!) navigation - moving across a chart takes ages, not because the software is slow (it isn't), but because charts are often many screens wide. A zoomed-out overview in a small overlay window to show the wider context of the currently visible portion of chart would be a nice addition.

My biggest beef, however, is with chart selection. Although the "default" chart covers the full range of the folio, selecting detailed charts for harbours and so on requires browsing a list which identifies them by codes and place names. I understand the reason for this; it is the easiest way to move paper charts to a digital platform. However, it does little for usability, and isn't likely to find favour with users who are familiar with modern digital mapping, as implemented by the likes of Google Maps, Map 24, etc. Why can't I go from a whole-Ireland view to a close-up of the Cork Harbour chart in one click?

Enough with the negatives; the charts themselves look good, and I got a lot of mileage for my money - the sheer amount of data that must be collected to make these charts is astonishing. I'm already looking forward to plotting my next cruise (and with departure mere weeks away, it's not before time!).

Favourite feature so far? Definitely the print option - beautiful, instant reproductions of charts or sections of charts are only a click away (after you accept the "not for navigation" license agreement message).

The verdict so far: excellent value for money. The software is spare, but very functional, and the coverage is great, and I'm very glad I bought it.

Thursday, February 26, 2009

Cleaning a microwave oven

The levels of dried-out burnt-on food on the roof of our microwave oven were well below bachelor-tolerance levels, and, considering the dose of radiation they had received, either totally sterile or already in possession of super powers. However, with my wife due back the next day, a fab-like level of cleanliness was highly desirable...

The first attack (dish-scrubber) was easily repulsed; the fossilized remnants of dinners past appeared to have attained an inter-molecular level of integration with the substrate. Doubled the elbow-grease factor, tried again, same result. Next, considered the array of chisels, lump-hammers, power saws, angle-grinders available to me, but rejected them on grounds of safety (mine, on the return of She Who Must be Obeyed). What to do?

Then, the light bulb moment. I got a bowl full of water (just water), deployed it to the centre of the oven, and gave it 8 minutes at 800 watts. Opened the door again to a steam-blasted oven that came clean on the first wipe. Result!.

Tuesday, January 6, 2009


So there I was, tapping away at quotidian tasks, when I noticed my PC was responding with all the dash and verve of a fossilized member of Testudinidae. A quick glance at Task Manager showed that GoogleDesktop.exe was the CPU hog - apparently, it required 99% of CPU to index the work I was doing with the other 1%. Surely that couldn't be right?

Well, I love the speed and power of Google's Desktop search, so I didn't like to just kill the offending process. To get me a little extra responsiveness from my PC while I searched for a better fix, I using Task Manager to give the GoogleDesktop.exe process a lower priority. To find the real problem, I would need more data on what GoogleDestktop.exe was trying to do - so I installed the impressively capable and friendly Process Monitor (a free trouble-shooting tool from Microsoft).

This tool gives details on the interactions of running processes with the operating system, and updates in realtime. For example, I could see GoogleDesktop opening new files and folders to index them even as I created them. What was interesting was that GoogleDesktop.exe was also repeatedly accessing a file called hes.evt, even when there was nothing new to index. I deleted this file, and an instate later, a new hes.evt appeared (at first I thought it hadn't been deleted, but the new one was tiny and had an up-to-the-second creation date).

And now... CPU usage fell away to "idle" levels, and Google Desktop Search still works. Great result, but what was the underlying problem? No idea whatsoever.