Those of you who have seen the excellent time traveling anime Steins;Gate, will be familiar with the Divergence Meter. The Divergence Meter is a device brought back from 2036, invented by the future version of the series' main protagonsist. It is an array of eight nixie tubes that uses an unknown method to determine the "Divergence Value" of the current "world line" (time line) in relation to the world line of departure, which has a default value of 0.0000000 The Divergence Meter is used to detect what time line the universe is currently in. Throughout much of the storyline, the protagonist tries to break the 1.0000000 barrier (create a world timeline change so significant as to raise the first, leftmost Nixie tube value from a 0 to a 1) to escape from the "Alpha range" of timelines (as indicated by values below 1.0000000), thus avoiding a series of events from ever taking place (how such a device would work has been the subject of much speculation amongst geeks, such as this blog post here). Needless to say, this device inspired geeks everywhere to want one, and indeed some (very expensive) "functional" props were produced for sale. Functional as in: showing the divergence values that feature in the series, generating random values and, inevitably, simply telling the time and date, so you can use it as a desktop nixie clock. But geeks being geeks, several people also created their own circuits, so people can build their own, mainly around the affordable IN-14 nixie tubes: Divergence Meter Project – Waicool20's Site Divergence Meter Project (brotoro.com) The latter is a popular project, with Nixie Keith selling some PCBs for it on Tindie. And it is this project that I decided to build myself. The physical appearance of the original Divergence Meter has a purely functional ghetto-mod aesthetic, but I decided that I wanted to make a Steampunked version: something that would look like a Victorian/Edwardian era scientific instrument that would accompany the lady or gentleman chrono-adventurer on their time travels. I pictured a device in brass and black walnut, looking a bit like a carriage clock, made with the traditional Victorian materials and methods of those days. The result is below. It looks to fit in quite well with my (real) Victorian microscopes. Here some more detail shots: A build log will follow. Meanwhile you can see the whole story on Flickr: Divergence Meter | Flickr
Sooo... As mentioned above, I ordered some PCBs from Tindie. They came in different colours as the batches rotated; that month's flavour was red (would have preferred black, but hey). They come either as is, or with SMD components pre-soldered for those who struggle with that. The SMD components are Supertex HV5622 32-Channel Serial to Parallel Converter chips; little miracles of 10mm square that can handle nixie voltages of 180V and about three tubes each, and much preferable to the larger (and harder to obtain, these days) Soviet K155ID1 nixie drivers which can only drive one tube each. However it is not difficult to solder them on yourself! You just use soldering flux paste (flux comes in anything from a paste to a very low viscosity fluid) to hold the components in the right position, load up the head of the iron with solder, and then you swipe in one, smooth motion across the contacts. As you can see through the magnifier, that makes for a clean solder. The kit recommends a specific pre-manufactured HVPS, the Taylor Electronics 1364 (http://www.tayloredge.com/). Tom Titor used this approach because these things are a challenge to design yourself, with various component placements having unpredictable effects, Luckily they are fairly cheap (about $15,--). The PIC running the show is a PIC16F628A. The prototype board used a Dallas DS1307 real-time clock chip to keep the time and date, with a CR2032 battery backup to maintain time even if the board is not powered up. However this was found to be wildly inaccurate, gaining one second every eleven hours (even though the code allows you to adjust how fast the clock runs), so the main board layout now also accommodates a Dallas DS3232 high-accuracy real time clock chip which should stay accurate to within two minutes per year. This is another SMD component and finicky to solder; even more so the teeny-tiny SMD capacitor that came with it. Below the completed boards: The original circuit runs on a 9V power adapter with a 9V battery as backup. The idea is that when unplugging the clock, you can throw a switch to go to battery, so you can use the clock as a cosplay prop if you want. The circuit itself runs on 5V however, with a LM7805 1A 5V regulator (TO-220 package) regulating the voltage down. This seemed to be an energy inefficient way of doing things... I preferred to run the circuit from a 5V USB adapter, with a rechargeable Li-Ion battery as backup. Adafruit sells a nice little PSU/charger circuit that connects to a USB power adapter and a Li-Ion battery. While the adapter is connected it puts out a steady 5.5V and keeps the Li-Ion battery charged; when the adapter is unplugged it seamlessly switches over to the battery (until it runs out, of course). When the adapter is plugged back in, it switches back to mains feed and recharges the battery. So I left out the unnecessary LM7805 and bridges the contacts instead. The switch connects directly to the Adafruit circuit which will handle the mains/battery switchover, and therefore act as an overall on/off switch. This allows me to turn the tube display on and off at will (without affecting timekeeping, as this is backed up with the CR2032 button cell). Incidentally, the clock allows you to set what hours the display should come on and switch off again, so you can turn off the tubes overnight when nobody is likely to be watching them (in my case, they come on at 8:00am and turn off at 11:00pm at night). This also extends their lifespan (which in any case is rated at 12,000 to 200,000 hours easily, and there are reports of nixie clock owners who have ran such tubes for over 5 years now, without problems). Power consumption is not a problem; nixies are cold cathode technology, basically, and calculations show that running a nixie clock for 30 years will cost you about £150,--* Tom Titor sells pre-programmed PICs, but of course you can program one yourself with the assembly code downloaded from his website, using a PICkit 2 or 3. Conveniently you can do this with the PIC already seated on the board as Tom Titor incorporated a ICSP (in-circuit serial programming) connector on the main board. PICkits are cheap (not to mention, lots of unlicenced knock-offs exist) and always a handy tool to add to your geek arsenal. Anyway, flashing it was easy. Next we move on to the nixies. The circuit was designed around IN-14's as these are ubiquitous and relatively cheap. Tom Titor also has a PCB schematic for IN-18's (if you feel rich...) but you'd have to make your own; Tindie does not sell these. * future crazy-ass energy price rises excepted...
On to the nixies. IN-14s are some of the most commonly available (and cheapest) cylinder shaped nixies (others being the lozenge shaped IN-12s), and since this project needs eight of them, this was Tom Titor's preferred choice. They are ideal: not only do they display the full range of digits; they also have two dots in each bottom corner (most nixies have just one, if any at all). Tube size is 19mm diameter x 55mm and digit size is 18x12mm. The tube has long directly solderable wires, a firing voltage of 170V and keep-up voltage of 150V, at a current of 2.5mA. It has three brothers: the IN-19a, IN-19b and IN-19v, which are symbol nixies. This makes for a very flexible set of display options in multimeters, meteorological, acoustic or lab displays. The main source for procuring NOS (New Old Stock --i.e. unused and straight from the original box) nixies is surplus stock languishing in former Soviet warehouses in Ukraine, where they were once manufactured. There is a teeming trade on eBay and Etsy of Ukrainian dealers who roam the warehouses for surplus stock of nixies, oscilloscope CRT tubes etc. Although this summons images of dodgy chain-smoking guys in faded leather jackets and Ray-Ban knock-offs called "Yuryi" or "Andrei" stalking the Zone for abandoned military tech to sell on the black market... ...I have found them remarkably professional, prompt and reliable in dealing with them, and have bought IN-14s, IN-19s, IN-18s, IN-13s and even am oscilloscope CRT (for another project) from them without any issues. Prices of nixies have been steadily rising over the last decade as nixie clocks have become more popular, and although good ol' Soviet efficiency means that tubes were overproduced by the thousands and became obsolete before they were ever used, stock is gradually dwindling. The larger IN-18 tubes that cost me perhaps £60,-- each in 2019 are now £80,-- each, or more. The war in Ukraine has not stemmed the flow of trade, but it has slowed it down and prices have risen even more of late (Non-Soviet tubes are even rarer, like the desirable, large Z566m tubes which currently cost £90,-- each, and the very large Japanese Rodan CD47 is pretty much unobtainable now --they used to go for £500,-- each back in 2015). Now you know why I bought so many: stock up while you can! IN-14's come with a plastic base, often of varying colours depending on manufacturing source and batch. This is the first thing I needed to address... Black vinyl die is our friend! Much better: Soldering nixies with 12 thin, bendy wires is tricky. To facilitate placement on the board, I lowered the plastic collar a bit and then cut the wires into progressively decreasing lengths, starting with the anode wire which is clearly marked on the tube, and also easy to make out on the PCB. This ensures correct orientation of the tube. It is then possible to guide each wire into its hole sequentially, using the lowered plastic collar as additional aid for alignment. Care must be taken during soldering that the tubes remain upright and perfectly parallel to each other; once they are soldered they have very little give in them. Et voila:
On to the wooden case. I opted for a nice black walnut to go with the brass, with a steampunk aesthetic and Victorian joinery details, which meant box joints for the walls and ogee bevelling on the base. Trend did a set of miniature dolls house routers for model makers, a set of which I was able to pick up cheap on eBay (they are not made anymore). I also found a 4mm box joint router bit. Metal or wood, it's all the same for the mill... Using a more standard set of routers, the base was hollowed out, including a recess for the battery (I started with a single cell one, but later decided to go for a double cell battery): The walls were cut from some 6mm thickness black walnut sheets and the box joints cut using the router (6mm deep, i.e. wall thickness). The side walls were propped up 4mm relative to the main walls (using an engineering spacer block) to ensure the joints aligned and interlocked correctly, and the whole setup backed with a piece of sacrificial wood to prevent splintering during the cut from front to back. Using a router I cut a 3mm deep, 3mm wide groove in the bottom of the walls, which thus slot in the lips of the base as shown. And here the PCBs set up on the base with some upstands I turned from 6mm brass bar and fixed with 3mm brass screws. I later decided to mount the PCB not to the base, but the brass top plate. This was so that if the clock ever received a sharp knock sideways against the edge of the brass top plate, it would not end up shearing off the tubes... In fact I changed the design several times as I went along building it, and even now would have done a few details differently in hindsight. Live and learn. And a test run of the circuit using my bench power supply. It worked! Clock mode and Divergence Meter mode: You will note that I am wearing a glove as I manipulate the buttons on the back of the circuit to summon different divergence values. This is because some of the 170V lines feeding the nixies run quite close to the edge of the board. High voltage circuitry should always be handled carefully!
Next: the brass top plate, made from 2mm brass sheet (which is rigid --and heavy-- enough). Email communication with Tom Titor informed me that the nixies are spaced exactly 21mm apart (centre to centre) on the board. Getting this precisely right would be crucial. The nixies themselves are approx. 19mm diameter (being glass blown tubes, very small variances are expectable). I decided to drill 20mm holes, partly because that's the milling bit I've got, and partly because the 1mm tolerance would allow for inevitable small variances in the upright position of the tubes, and for the tubes to move very slightly without cracking. So first the brass plate was covered in blue layout fluid and marked out with a caliper and compass: The plate was then clamped on a piece of sacrificial wood and milled with a 20mm milling bit. I needed to pause every few holes to allow the brass to cool back down, as it expanded and warped as it got hot. This would have thrown off the measurements... Having drilled the main holes, I also drilled and countersank the holes for the PCB mounting. Getting the alignment right was easier now I could test fit the PCB. The brass was polished and straight-grained with a soft Scotchbrite pad and grade 0000 steel wool. The countersunk screws were brushed radially using the lathe (details matter!). And the result: Tested on the wooden case. The glow of the nixies works well reflected in the brass:
Because the Divergence Meter is essentially a travelling companion for the lady/gentleman chrono-adventurer, I wanted to evoke the feel of a carriage clock. As such the protective bracket surrounding the nixie tubes would double as a fold-up carriage handle. Here is where some design (re)considerations were made. Originally the idea was to mount the PCBs to the base with brass upstands, and glue the walls together using small oak staves in the corners (oak because it is sturdy, and because that's what I happened to have to hand).The wooden base and brass top plate would then be screwed into the staves using brass wood screws, sandwiching the wall assembly in between. However, this would mean that the whole clock would effectively be held together by the corner staves. With those being made of oak that would have been robust enough, the weight of the clock would hang from these staves by the brass wood screws. It did not feel elegant. Moreover it would mean 8 screw holes in the base (4 for the PCB; 4 for the staves) or, if the PCB is preferably mounted to the top brass plate, 8 screw holes in the brass top plate, which again felt inelegant. So I decided that the base would be connected to the brass handle assembly, which would support the weight of the clock, with the walls sandwiched between the top brass plate and the base. This would minimise the screws in the top plate and the base to 4 each, and make for a sturdier assembly. The hinges incorporate some hinge parts that I originally made for Ada (and then remade as the lever mechanism changed slightly). The other parts were similarly made from 10mm square brass bar. And more milling (using my home-made positioning stop, to ensure that both parts are cut and milled identically)... And more milling... Until we have the finished parts, test fitted with a 4mm hinge pin: The hinge pins were cut from 6mm brass bar. I was aiming for what engineers call a friction fit: tight fitting parts that produce a joint which is held together by friction after the parts are pushed together. Trial and error established this to be 4.1mm for 4mm holes. Once gently tapped in with a hammer, they were cut to size with a piercing saw. The result is a nice, almost invisible hinge pin once sanded flush with the hinge. Next I milled the crossbar from 6mm brass bar, putting in some accents along the way. And test mounted in place to see how the assembly offers protection to the delicate tubes, and works when folded up as a carrying handle: The crossbar is wedged between the upstands, but also screwed in tight with 3mm countersunk screws. The bracket is screwed to the top plate with hidden 6mm diameter upstands, to which the base will eventually be connected via a set of horizontal feet made of 10mm square brass bar, tapped with M4 screw holes.
Thanks for the log. I like how the mat went from pristine to gross. All it needs now is some black brasso residue, half of a peanut, and a suspicious curly hair. I'd say it's done it's job.
The USB PSU/charger is position with a bit of Ash, which will also hold the battery in place. My new set of Japanese chisels comes in handy here. All this was done at this stage to determine the final positioning and cable arrangement. It turned out the USB PSU needed to be flipped over to clear the components on the PCB above. All these little things can catch you out! Here the final construction is visible: the bottom base screws to the brass brackets which screw into the carrying handle. The walls are sandwiched between the top and base plate. The whole design allows for easy disassembly and repair if necessary. The next challenge was to fix the USB PSU/charger circuit to the case in such a way that it would withstand the rigors of cables being pushed in and pulled out. After some thinking I came up with a pin construction which pins the PCB firmly to the base, while also offering M2 screw holes to fix the brass cover plate surrounding the USB port (this cannot be screwed to the back wall which needs to be able to lift clear from the base in disassembly). It involved making two very small, but very precise identical parts from 6mm brass bar. The holes were tapped with a tiny M2 tap: Next the brass cover plate made from 1mm thickness brass sheet, cut with a piercing saw (old skool) and the micro USB port opening milled and then filed with a needle file to the exact shape and dimensions required. The parts and the assembly. The two 2mm holes in the brass cover are for the battery charging/battery low LEDs. The screws are M2 countersunk screws, again radially polished on my lathe. The Japanese chisels came in handy again to make the recess for the USB assembly. Note the holes for the mounting pins. Once the PCB is screwed in place at the back and the wall lowered in position, the whole assembly is firmly fixed in place and can withstand the stresses of plugging and unplugging. The Adafruit circuit comes with four surface mounted LEDs: Power (blue LED), Battery Charging (yellow LED), Battery Full (green LED) and Battery Low (red LED). This was inconvenient, as I wanted LEDs in the brass plate, not on the PCB. I also decided that I only needed two: Battery Low (orange rather than red), and Battery Charging (yellow). The reasoning was that the nixie display would obviously indicate when power was present, and that when the battery is not low (or charging), it is obviously full. So the Power and Battery Full LEDs were redundant. All SMD LEDs were desoldered (unfortunately they did not survive this process), and small wires were carefully soldered to the pads of the Battery Low and Battery Charging LED, with 1.9mm LEDs at the other end. I also soldered a 2.0mm JST connector to the PCB, which will connect to the corresponding connector wired to the main PCB. This again makes potential disassembly for future repairs or maintenance easier (after all, that CR2032 battery will need replacing every few years). The final assembly below, with LEDs superglued in place (OK, I admit that is not a very Victorian technique, but in my defence they did use glues). You'll note that the brass plate is already lacquered on these photos; more on that later! And in place:
The clock has two push buttons for controls, and one slide switch to toggle power. Of course plastic would not do! For the push buttons I decided on brass buttons with pearl inlay, turned from 10mm brass bar. Finnicky work again.... The result. The pearl inlay discs are 9mm diameter, and cheaply available from any jewellery dealer. Test fitted on the PCB (which helped me determine where the holes in the walls needed to go. I decided to mount them recessed, which would be the case in a travelling instrument, to protect them from glancing knocks. The sliding switch got a similar treatment: a knurled knob turned from 10mm brass bar, mounted recessed on the switch: And here buttons and switch test fitted in place: Finally, the battery retention clip, cut by piercing saw from 1mm brass sheet, and drilled and countersunk for 2mm screws: Screwed in place and test fitted: On to the brass lacquering, which turned into quite an adventure...
That's good when the mess is not coming from junk piling up and turning into a storage space. Never before has a usb port looked so upset about it's existence.
Most genuine Victorian scientific instruments are easily recognisable by their unique brass lacquering. This was obviously to protect the brass from tarnishing, but also to make it look more lustrous. These lacquers were made from shellac diluted in alcohol, with dyes added, such as annatto, turmeric or dragon's blood, or a combination thereof, giving the brass a gorgeous golden hue. Occasionally roasted walnut shell was also used, and possibly other dyes; unfortunately many Victorian craftsmen and workshops did not keep records of recipes or techniques (although I am trying to get my hands on a few historical records), so we don't know for sure. What we do know is that different workshops often used lacquers with their own signature colours, and a knowledgeable antiques dealer can recognise not only who made a certain scientific instrument but also in what year, just by looking at the colour. Naturally I wanted to give my steampunked clock the same treatment. Some restorers actually mix their own lacquers, and as I have been researching this subject I found that it is probably not hard to do. However Kremer Pigmente does a shellac according to a known 1892 recipe and I happened to have a bottle of this, acquired some time ago (I have since found another company called Horolacq which does a range of different colours for professional clock/watch restoration). The first tries at lacquering brass were a disaster. It has the viscosity of water, but dries very quickly. You cannot brush it out with a brush --it will mark and streak. You cannot retouch it, nor apply it in layers, as each new application dissolves the previous one. It has to be applied in one go, and allowed to settle and even out, in a totally dust and moisture free environment. If a dust fleck or hair gets stuck on, you have to clean the whole piece with alcohol and try again. If it does not settle evenly, you have to clean the whole piece with alcohol and try again. If there is too much moisture, it turns opaque and you have to clean the whole piece with alcohol and try again. It shows up the tiniest scratch or flaw in the brass, and lacquering small pieces with corners, grooves or holes is tricky as that is where the lacquer wants to settle. It also will drip down to the lowest gravitational point. This was harder than I thought. In the end I turned to two people for advice. One is Penny Thoyts, a professional antique microscope restorer whose supreme crafts(wo)manship can be admired on her website. The other is a Dutch engineer, Tatjana van Vark who is clearly a genius, having restored her first oscilloscope at the age of... 14. My email correspondence with them was enlightening. From Penny Thoyts: Hi Robin, Your projects look wonderful, such good fun! Lacquering is hard. Warming does help but it's definitely warm not hot. I use a desoldering gun or hot air gun. Hot plates get too hot and ovens cause the lacquer to go brown. Traditional lacquer was done by warming the piece with dry heat ie they placed it in front of a hot fire, not an oven or hot plate. I warm it so you can still touch the metal. 50C ish. After I have got the lacquer on I let it dry for a few minutes then give it another blast, up to about 200C to help it set. The most important thing of all is the humidity. Any water in your lacquer and it will go cloudy. If the air is too humid it will go cloudy. It is much easier to lacquer on a bright sunny day in the UK. I have a dehumidifier in my workshop. (note: This is true: some brass musical instruments restorers say that they will only attempt lacquering jobs in Spring or Summer, never Autumn or Winter). Getting the correct amount of lacquer on the brush is the tricky bit. Dryer is better than wet when you start out. You must not go over the same area twice and you must be fast. I don't know of any books about it unfortunately. I really ought to write one. I have seen it mentioned in a couple of old books but nothing particularly helpful. It's just trial and a lot of error. Tubes are the easiest to do as you can pop them on the lathe. Large flat areas and funny shapes can be a real pig. Soft, flat ox hair brushes are essential. You'll never get it on with a paint brush. For funny shapes I often use cloths dipped in lacquer rather than brushes. I always use these: https://www.halfords.com/motoring/car-cleaning/sponges-brushes-and-buckets/halfords-soft-polishing-cloths-x-4-172395.html They are the only ones I can find that are soft enough. Correspondence with Tatjana van Vark was in Dutch, but the gist of it was: Skill in this indeed difficult art can only be acquired by continuing to experiment with endless patience. Such skills are not verbally transferable. Just start again until the result is acceptable..... Continuous cylindrical surfaces can be beautifully varnished in the lathe with a cotton ball dipped in the varnish. I wish you good luck ! So armed with the lathe and heatgun, ox hair brushes and Halfords soft cloths I tried again and learned: Soft Halfords cloths dipped in shellac work much better than the oxtail brushes (furniture restorers write that they use cotton balls wrapped in soft cloth). Large flat surfaces are best done cold, so the lacquer gets time to smooth out evenly. Do the underside before you do the top side that is most visible. Cylindrical items are best done spinning slowly on a lathe, and left spinning while they dry so the lacquer doesn’t settle at the lowest point. Then use a heat gun to warm the items up to let the lacquer harden (without causing it to bubble or discolour; smell is a good guide here...). After ten to fifteen attempts I actually got reasonably good at it. This made me wonder how Victorian craftspeople would have done it. I imagined that they might have used rags, or perhaps sheeps fleece, as they would often use what was cheaply to hand (they also used cream as a metal cutting fluid). Might the lanolin in the fleece facilitate a better transfer of lacquer onto brass? I think I will get my hands on some and try it out. Incidentally in my research I found a reference of interest, citing recipes for 19th century finishes: Hazel Newey, Aspects of the Conservation of Early Scientific Instruments and Apparatii, dissertation submitted for Diploma in the Conservation and Archaeological Materials, Institute of Archaeology, London, 1976. Correspondence with Penny Thoyts again: Hi Robin, Glad your lacquering is coming along. There are a lot of tricks to it. I find that when lacquering tubes with slots or holes cut into them, a sheet of paper inserted down the tube is just absorbent enough to soak up any runs. Where there are holes in flat surfaces, sometimes a very gentle touch with a dry cloth or tiny piece of foam on the edge of the hole can prevent the ring effect you often get. It really is trial and error. As long as you aren't planning on starting a rival business lacquering microscopes I'm happy to help where I can. I'm not convinced about lanolin, people have gone to great lengths to dewax shellac and degrease tubes before lacquering. I'm not certain that sheep grease will help, but it's certainly worth a try. It can do no harm to give it a whirl. Chamois can be quite oily, maybe that would be worth a bash too, I have certainly used real chamois with jeweller's rouge for final polishing. It works very well for that purpose. If you can't get unprocessed wool, women's cosmetics are currently having a lanolin revival, it's in everything at the moment. It was terribly out of fashion for a long time. The references you sent is quite good, the thesis by Hazel Newey is held at UCL and they don't seem too keen on making copies for people. I shall have to send the librarian a grovelling email. I can't seem to find Hazel Newey anywhere, it would appear she has retired, I may be able to track her down if I'm lucky. Attached a picture of today's lacquering. It went on first time! (It was heartening to note that even an experienced professional like her still finds it a difficult job to get right first time!) After using 2/3rd of a 250ml bottle of lacquer, and at least 15 attempts, I think I finally got the hang of it. All the photos in my log are the end result of a week of trying and cursing and trying again. Cloth is definitely the way to go (although it soaks up a lot of lacquer). I had to disassemble the hinges and then put them back together (and push in the pins) without scratching the lacquer, which was a challenge, but I found I could push the pins home with the engineering vice (which is machined with perfect 90 degree angles and smooth surfaces, and applies an even clamping pressure throughout). I used greaseproof baking paper to line the vice; this seems to protect the lacquer without sticking to it.
Out of curiosity, what will you be doing with your workshop and tools when you eventually move to the EU?
My tools are all coming with me. I am looking for a property with a big-ass barn which will give me plenty of space to amass even more tools make all sorts of stuff.
The lacquering. As described above, I used a Halfords soft polishing cloth dipped in shellac to lightly swipe the brass top plate in one stroke... This seemed to deliver a nice and even coating. Even then I had to do it three times because of a random fleck of dust of fibre. :grrr: I did the underside first, then mounted the spacers for the PCB to act as legs, and did the topside. Then I used the heat gun to gently warm it and encourage the shellac to harden. The hinges were disassembled and screwed to a temporary holder to clamp in the lathe clutch, and while slowly rotating brushed in shellac, again in one smooth swipe using a cloth. They were left to dry while rotating, to prevent the shellac from settling at its lowest gravity point. Then I gave them a blast with the heat gun again. The small bits of the hinge were held with improvised wrapped wire. They were by far the most finnicky bits. The ends of the hinge pins were done with a cotton bud while clamped in a self-closing tweezer. The crossbar was held by a screw clamped in the lathe clutch and a live centre in the tailstock. Again, one smooth coating while it was rotating (trying to avoid a spiral effect), and then blowing it with the heat gun, still slowly rotating away. And the result! You can see the contrast in colour with the mounting pins under the plate, which are untreated. Of course all the visible screws got a coating too, with the cotton bud. Next: assembly and lacquering of the case.
The case walls were glued together in stages (with wood glue, naturally) with oak staves in the corners... To fill the small gaps in the box joints, I mixed black walnut wood dust with wood glue and covered the joints: After drying the corners were sanded level. In hindsight, I suspect a general wood filler would have worked better; the wood glue mix seals the grain of the wood and affects subsequent oiling and staining. I had to do quite a bit of sanding to remove the surplus. In Victorian times, wood would be treated with linseed oil and then lacquered in shellac. In fact, the first varnishes were linseed oil boiled with a shellac alcohol solution. The linseed oil offered protection from moisture and solvents and alcohol (since shellac dissolves in alcohol); the shellac offered toughness from scratches and prevented linseed oil from oxidising and darkening. Antique furniture restorers still use this combination. Shellac has a rather glossy finish; sometimes this would be enhanced with French polishing; sometimes it was toned down with a mild abrasive and a beeswax coating. Fortunately, using shellac on wood is as easy as it is hard using it on brass. In fact, woodworkers say it is impossible to mess up. A chimp on LSD could do it. A squirrel on crack could do it (I did not have chimps, squirrels nor hard drugs available to test this). It can be brushed on with a brush, or with a cloth. Again, you cannot layer coatings; the second would simply dissolve the first. But woodworkers recommend oiling the wood, then sanding lightly, then applying a coat of shellac, then sanding the surface again and applying another coat to meld with the first. Usually two coatings suffice. They also recommend dipping the cloth in a bit of mineral oil before using it to apply the shellac to prevent it sticking to the wood. This seemed to confirm my earlier ideas about using wool fleece, with a high lanolin content. I decided on using raw linseed oil rather than boiled (which is tougher, but also darkens the wood), and a button shellac which gives black walnut a warm, almost mahogany glow. Then I would mildly sand it with 0000 steel wool and give it a buff with a furniture wax made with wax from my own beehives. The linseed coating already brought out the wood nicely: The button shellac was dissolved in denatured (i.e. undrinkable!) alcohol. Some restorers claim that 98% pure (cooking) alcohol creates a warmer colour, but that stuff is harder to get and a lot more expensive. I made what is called a #2 cut solution (i.e. 2 pounds of shellac on 1 gallon of alcohol, or in metric: 24 grams of shellac on 100 cl of alcohol), which is considered good enough for general purposes. Dissolving the shellac is a slow process and takes about 24 hours or more; some use a magnetic stir plate to speed up the process (you can pick up a basic one cheap for about £30,--, but the internet has logs by people making their own home-brew version from an old computer fan and some magnets. Needless to say I have already started building one from some redundant PC parts I have lying around). The end result really brings out the grain of the wood: Next a light buff with 0000 grade steel wool, and then a buff with my home-made furniture beeswax.
And the final assembly. Here you can see the positioning and wiring of the electronics, and how everything fits together. Once the brass brackets are screwed into the support posts (which in turn screw into the carriage handles, they push up against the oak staves and wedge the walls against the brass top plate. The base is then screwed into the brass brackets in turn, and wedges up against the bottom of the walls. the whole thing is kept together by this brass "skeleton" surrounded by the wooden walls and bottom. Everything put together, and the first test run. Nice. The buttons and switch were also lacquered with a cotton bud, and then glued in place with (a tiny dab of) Bison Kit. This is a well-known all-purpose tough glue in the Netherlands, less known over here. It dries to a tough rubbery consistency rather than creating a brittle bond like superglues do, and is great for bonds that are subject to tensile stresses. And the final result here: The carriage handle function: And its current place in the living-room: And that's the log! It's been great fun, and I've learned a lot, for Ada and future steampunk projects which I am already lining up.
