A home-made disc recorder

Ever thought of making your own records? This is the saga of how I did it, with very little in the way of skills, tools or money

A very 21st century problem

The Mark 2 cutting head in action. A large 78rpm disc is used as a platen.

It all started with a pile of old, redundant CD-R backups. Like most computer users, I've learned the habit of regularly archiving my work on to recordable CDs, and over the years these have filled several CD storage folders. Last winter I set myself the task of going through the lot, directory by directory, file by file, reminding myself of what I had and working out what could be safely got rid of. It's surprising the rubbish you hang on to for no real reason: I was pleased to find that most of it could be binned. All the files I still wanted to keep would fit on to just two DVD-Rs.

This left the problem of how to dispose securely of over 100 CDs and DVDs, some of which contained personal information. There do exist office shredders that cut up CDs, but they are expensive; and short of taking the discs into the yard with a hammer and smashing them to pieces - which would be both messy and dangerous - I was stuck for what to do.

In early April 2011 I was clearing out some stuff including an old broken gramophone that I had rescued from the local tip, and once more I came across that pile of old CDs. Out of curiosity I ran a little experiment. I centred one of the CDs upside down on my hi-fi turntable, took the soundbox off the gramophone arm, clamped its diaphragm against one of a pair of headphones leading from my cassette player, made sure the volume was well up, and held the needle to the rotating CD, tracing out a rough spiral. When I played the spiral using the turntable's own arm (at great risk to stylus) I was delighted to hear, amidst a wild sea of appalling noises, a few bars of the original tune warbling tinnily back at me.

The success of this little trial set me thinking. With a bit of effort, and much refinement, would it be possible to make something that could do this better? What I wanted to know was: would it be possible to build a machine to make my own records? Professional disc cutters run to the tens or hundreds of thousands of pounds; there's a machine for home/semi-pro use, made by Vestax, that does one-off records, but it costs about eight thousand. Even a refurbished 78-era cutter, assuming I could even find one, would be well into the late hundreds. Making my own device, out of the materials I had in my garden shed and whatever else I could scrounge, would seem a somewhat ambitious project, no doubt!

Doing the homework

My first task: a bit of learning. Find out all you can about instantaneous disc cutting; use Google, Wikipedia, the local library, online bulletin boards. See what others have done. To paraphrase Tom Lehrer: plagiarise, but be sure to call it "research". I found a first-rate bulletin board dedicated to disc cutting, the Secret Society of Lathe Trolls, whose members' expertise ranges from decades-served professional to absolute novice. The first thing I learnt was that making a good record is hideously complicated and requires a firm grounding in electronics, mechanics and machining, which I do not have. The second thing I realised was that making a crude record isn't too hard, given the number of YouTube videos of people doing basically the same thing as I had done with my headphones and gramophone soundbox.

To sum up the early part of my research: a record-cutting lathe (such is what they are called) comprises essentially three parts:

• A turntable, on which a blank disc is placed;
• A cutting head, which converts electrical signals (the music) into vibrations of a stylus;
• A feedscrew, which moves the cutting head very slowly across the turntable as it spins, enabling a tight regular spiral to be cut.

The easiest of the three: the turntable. I play my records on a Leak Delta which cost me thirty pounds second-hand a couple of years ago. It's really a Goldring Lenco 75 with a different sticker on the top. It is one of the best (and best-selling) turntables of the 1970s, with a good heavy platter and variable speed from 15 - 85rpm.

Capturing the sound: Mark 1 cutting head

Now for the difficult bit. How do you go about making a cutting head from scratch? Well, think of what happens when you play a record. Put about as simply as possible, undulations in the groove cause the stylus to vibrate, which creates a tiny electrical current in the pick-up. This current is fed to the amplifier, which boosts the signal and feeds its output to the loudspeaker.

The cutting head has to do pretty much the same thing in reverse. If you touch a speaker's cone while it is playing, you'll be able to feel vibrations, which are essentially the same as those that the playback stylus encounters in a record, although greatly amplified. You can use a small speaker, plus some other bits and bobs, to make a primitive cutting head that will capture the speaker's vibrations and transmit them to the cutting stylus. My first try at this was based on one developed by the contributor "Seringson" (YouTube video link) and consisted of the following items:

• 5-cm tweeter, from old Sony speaker salvaged from tip;
• The pull-cord handle off an old roller blind;
• Two darning needles;
• A cork;
• A steel gramophone needle;
• Part of a light fitting: a small, cylindrical cable connector with three screw connections;
• Cotton thread;
• Small pieces of plywood, wood blocks and screws.

Details of Mark 1 cutting head.

In order to cut a mono signal on to a record, the stylus must vibrate side-to-side, not up and down. This cutting head has the speaker mounted within an inverted L shape made of plywood, at a shallow angle to the record - allowing just enough space for the turntable to pass by underneath. The angle must be shallow so that as much as possible of the speaker's effort goes into moving the stylus side-to-side. Notice that I cut some of the speaker housing off to get the speaker as close to horizontal as possible.

Mark 1 head on gantry.

Now we need some means of transmitting the speaker's vibrations to the stylus. The pull-cord fitment was hollow, and the open end I found to be exactly the same size as the tiny dust cap on the centre of the speaker, so it would fit over the top and could be glued in position. Process: drill tiny hole in other (closed) end of pull-cord fitment. Insert one darning needle (cut the point off first, for safety), use Araldite to stick needle in place and hold it absolutely steady for 15 minutes until the glue sets. The other end of this needle will be fixed into the first of the three cable connections of the light fitting. Now, fix wooden blocks to the underside of the plywood L, so the cork can be wedged firmly between them. Insert the other darning needle into the cork, and fit the eye end into the middle cable connection. Both needles will need to be bent* using pliers so that the gramophone needle - which goes into the third of the cable connections - makes an angle of about 30 degrees from vertical. This is quite a fiddle to get right, but once it is, and not until, glue the pull-cord fitment - or whatever object will fulfil its purpose - to the speaker.

(* If you do decide to have a crack at making this thing, be very careful when cutting or bending darning needles as they can suddenly snap and send bits flying off. Wear eye protectors! )

Finally, I put some screws part-way into the wooden blocks, and wound cotton thread round them and the stylus assembly. Although I used four screws, I later found that I could secure the stylus assembly quite adequately by winding thread round it and just one of the screws. The purpose of this is to prevent the drag generated by the cutting process from tearing the stylus assembly off the speaker!

Music on the move: the feedscrew

The next bit: constructing the feedscrew. This enables the cutting head to pass very slowly over the revolving turntable, in a straight line from edge to centre. I figured that, given my limited carpentry skills, the whole thing would be more likely to work if I turned the concept on its head, and attached the feedscrew to the turntable, and moved it slowly under a stationary cutting head fixed to a gantry over the turntable. The turntable thus sits in what is really a "drawer" without sides, and the "drawer" is gradually "opened" by the feedscrew.

I used a pair of 500mm ball-bearing drawer runners, and a length of M8 threaded rod, one end of which passes through a fixed hexagonal nut attached to the underside of the "drawer", and the other end of which passes through a couple of L-shaped brackets. Between the brackets is a piece of plywood with a hole drilled, and in the hole sits a skateboard bearing. These bearings conveniently will just accommodate an 8mm axle.

The next problem was how to power the feedscrew so that it turns slowly and evenly. Remnants of gramophone to the rescue: the motor still worked, so it was drafted into service for this purpose. As the slowest it will go is about 60rpm, the rotational speed of the feedscrew itself is reduced by using a large Meccano pulley, and a Meccano belt, which just happens to fit into the circlip recess on the gramophone motor's spindle. Attaching the pulley to the M8 rod was a headache, as I needed to go from the Meccano standard shaft width of 4mm up to 8mm. I don't know whether any suitable kind of shaft adaptor exists; certainly I couldn't find anything on the Web, and I have neither the skills nor the equipment to make one for myself. The problem was eventually resolved, in a roundabout sort of way, by the use of several diverse pieces of ironmongery and a lot of Araldite.

The cutter gantry works in a similar way to the feedscrew, in that another length of M8 rod is turned (by hand) to move the gantry up and down. The cutting head has to be bolted to the gantry in such a way as to meet two conditions. The cutting stylus has to be the lowest part of the head/gantry assembly (it's surprisingly easy to make a mistake here!); and the stylus point must contact the disc so that the body of the stylus, seen from above, is at right angles to the radius. If this is done correctly, the stylus point should be in line with the spindle.

At this stage I had all the parts assembled and ready to test. The next stage of the adventure would be a matter of getting to know the machine I had built, and establishing how to get the best out of it.

Preparing the sound source

In order to make a record, one rather obviously needs a sound source, be it a live microphone, a master tape, or most commonly nowadays, a digital computer file. This source also has to have certain adjustments made to it in order to produce a signal that is capable of being cut into a record. In particular, bass frequencies, if left unadjusted, would generate a groove that would oscillate so wildly as to make the record unplayable. The signal has to be adjusted, or equalised, so that the volume of these bass frequencies is reduced. Conversely, high frequencies would vanish into inaudibility unequalised, so the volume of these has to be increased while cutting. An amplifier with a phono input contains a circuit that performs an equal but opposite adjustment, so that when you play a record, you hear the original version of the signal. Since the mid-1950s, an internationally-agreed standard for this playback equalisation has prevailed, called the RIAA curve, but before then many record companies had their own equalisation standards. Some sophisticated and pricey amplifiers can be switched to these older settings for playing certain 78rpm records.

An excellent and totally free alternative to a budget-busting amplifier is the sound editing package Audacity, version 1.13.12 and later of which is capable of carrying out the "inverse RIAA" transformation prior to making a record. If you play the transformed file back on your computer it will sound shrill and tinny. Cut into a record, however, and played back through an amplifier's phono input, it should, with luck, sound like the original, untransformed file.

The music machine turns on - briefly

Armed with this dangerously little knowledge, then, I set about making my first record, on the back of an old CD. Having connected my PC and the cutting head's little speaker to my amplifier, I set the turntable running, wound up the gramophone motor to start the feedscrew, gently lowered the cutter gantry until the gramophone needle was firmly against the spinning CD and I could see a trace being made, and then played the sound file on the PC. Three minutes later I had my record.

For a first shot, it wasn't bad. It sounded more or less like the original, albeit played through a transistor radio sitting under a waterfall. A lot of high frequencies were lost, and the recorded sound level was very low, which meant that the playback volume had to be set higher than normal, with a corresponding increase in surface sound. At this point I made two novice mistakes one after the other. The first was to tweak the "inverse RIAA" setting in Audacity to try to coax more high frequency signal out; and the second was to turn the amplifier's volume up too high. Halfway through the third CD, the little speaker yielded up the ghost. End of Mark 1 cutting head.

Capturing more sound: Mark 2 cutting head

In a textbook example of trying to prove I could run before I could walk, I decided that the Mark 2 cutter would be an attempt at stereo. Back to Lathe Trolls and Wikipedia for a few gems of knowledge. A stereo cutting head has to cut two-channel sound in the one groove, and so, in professional cutters, two coils - in effect small speakers without the paper cone bit - are used. These are mounted at 90° to each other, with metal rods connecting them to a central bar on which the stylus is fixed. This assembly is itself mounted at 45° to the disc.

The Mark 2 cutting head in position on the gantry.

Now comes the clever bit. When you listen to a mono recording through a pair of loudspeakers, the sound should appear to be coming from a point half-way between the two. This will be the case if the two loudspeakers are in phase - that is, if their two cones move in or out together. If the speakers are out of phase, one cone will be moving in while the other moves out, and the sound will seem diffuse.

Remember that, to get a mono signal on to a record, the cutting stylus has to move side-to-side, not vertically. This is accomplished in a stereo cutting head by connecting the two coils (speakers) to the amplifier so that they are indeed out of phase. When the same signal - that is, mono - is fed to the coils, one will push the stylus down and out at the same moment the other coil is pulling up and in. The net result is that the stylus is moved laterally, which is what we want. (If the coils were in phase, the stylus would only end up being moved up and down, and the result would be a "vertically cut" or "hill and dale" record. Such records are playable, but only using a suitable pick-up cartridge that has been rewired for vertically cut discs. Some early 20th century records were made this way.) If the two signals start to differ - that is, if a stereo element is introduced - the coils will start to do different things, and a vertical component will thus be introduced to the movement of the stylus.

I also decided that the new cutter would be made of aluminium, rather than plywood: another learning experience for me, as I had never worked in any metals before. I ordered a small sheet of 3mm aluminium and a couple of very cheap 5cm tweeters off eBay, paid a visit to B&Q for a few other items of hardware, and set about knocking stuff together.

List of parts for Mark 2 cutting head:

• Two small aluminium oblongs (90 by 85mm);
• One small aluminium triangle;
• Length of 1mm aluminium with 90° angle, for joining oblongs and triangle;
• Some M4 nuts, bolts and washers, ditto;
• Small Meccano L-shaped bracket;
• Large heavy-duty L-shaped bracket, for fixing cutter to gantry;
• The two tweeters;
• Two small cones, made from tough but pliable plastic sheet;
• Two plastic sticks from cotton buds;
• Part of aluminium drink can;
• A gardener's plant tag;
• Cylindrical cable connector, salvaged from Mark 1 cutter;
• A gramophone needle.

The cable connector and the aluminium strip from the drink can, with holes marked for drilling.

The aluminium strip from the drink can is wrapped around the cable connector, and has holes drilled into it at strategic points to allow the needle, cotton bud sticks and fixing screws to go in. Out of sight in the picture is part of a plastic plant tag, the kind that gardeners stick in the ground to remind themselves where they've put things, and this is sandwiched between the wrap-around layers of aluminium strip, and bolted at the far end to the Meccano bracket.

The cotton bud sticks were glued on to the plastic cones using my indispensable tube of Araldite, and then bent and cut to fit into the cable connector so that the cones covered the speakers' dust caps. As before, the final step, after a great many trials and dry runs, was to glue the base edges of the cones to the speakers.

(There was one further refinement. I fitted a fuse assembly into the cables leading to each speaker, in the hope that this might prevent any more overloading disasters. Somewhere I found that a "quick blow" fuse of 0.5 amp would be about right for these speakers, which are rated at 8 watts. This may not, of course, make a ha'p'orth of difference, but I feel as though I've discharged some sort of duty by making the effort. Update, December 2011: Yes, the fuses do work. After moving my equipment I reconnected something not quite right, and a horrifying wail from the cutter head was followed suddenly by an equally horrifying silence. The fuses in both channels had blown; the cutter coils were fine.)

Better results are obtained if heat is applied to the blank whilst cutting. Professional set-ups use a coil of wire wound round the sapphire stylus, but I settled for a simple 100w incandescent bulb - one of the most difficult items to find in the UK now that Nanny has banished them from the nursery - and this is switched on over the rotating blank a few minutes before starting the cut. Since taking these pictures, I've moved it slightly so that it is closer to the cutting head.

The completed lathe with Mark 2 cutting head.

Trouble with torque

But will it fly? I stuck to mono for these early tests, and as soon as I started to play the first CD/record made by the new head, it became obvious that there was a huge improvement in the sound, but still some loss of top-end frequencies.

One other problem that became apparent around this time was that the playback was too fast. With the cutter disengaged, I had set the turntable to 45rpm for recording, but had to play the resulting disc back at about 42rpm. The Leak/Lenco turntable, fine beast though it is for playing records, was never designed to cope with the extra drag on the turntable caused by cutting. I found that with the cutting head up, a speed of about 48rpm gave more or less the right results, as the cutter, when engaged, would slow it down to 45. Professional systems use a turntable with vastly more torque, or rotational oomph, than a domestic record player, and this problem, in general, does not arise.

I then had a go at half-speed mastering: slowing down the Audacity file and the recording turntable to half speed while making the cut. The idea here is that the high frequencies, so difficult to cut into a record, drop an octave and in so doing become easier to cut. As far as the high frequencies were concerned, I got some frankly stunning results out of doing this; but at slow speed, the performance of the Lenco, burdened with the cutting head, became unacceptably wobbly. Instead of 22.5rpm, I had to start the cut at 26. The resulting disc still ran too fast at the start and dropped like a stone in pitch during play, so at the end it was flat by about a semitone.

In the end I settled for a compromise: having slowed down the file appropriately using Audacity, record at 33 1/3 (measured at 35 with cutter disengaged) and play back at 45. This means that the turntable has enough momentum to withstand the cutting head pressure, reducing the wobbliness, and there is some improvement in high frequency recording.

The dilemma of the horns: cutting vs. indenting

Throughout this article, I've talked freely about cutting records, which is not, strictly speaking, what my home-built lathe does. A cut record is made by removing material from the surface of the blank disc, which is what high-end disc recorders do, using a chisel-shaped stylus. Cutting a record produces a trail of swarf, or chip, which has to be continually removed from the disc surface before it gets tangled up with the stylus. A tiny vacuum cleaner, mounted adjacent to the stylus, removes the chip as soon as it is produced. What my lathe does - and the cheaper disc recorders do - is merely indent: no material is removed from the surface; it is simply pushed aside.

Perhaps a somewhat cumbersome analogy will help. Imagine you go out to clear your path after a moderate fall of snow. You take a shovel and a large wheelbarrow. As you proceed down the path, you shovel up the snow into the wheelbarrow. When the barrow is full, you wheel it somewhere convenient and dump the snow out. This is akin to cutting, since the snow is removed.

Now imagine that instead of shovelling, you use a simple snowplough, a wedge of wood attached to a broomstick, and walk down the path pushing it ahead of you. The snow is not removed, but is just piled up at either side of the path. This is similar to what happens when indenting a record. The material piled up on either side of the groove is known as the horns, because if you imagine a cross-section of the groove, this displaced material sticks up like a pair of horns. The uncut surface between grooves is known as the land. When you play a cut record, there are no horns, and the playback stylus drops immediately into the groove; playing an indented record, more care must be taken to get the stylus into the groove, as there is a high likelihood that it will skitter about on the land, trapped between adjacent sets of horns.

Media study: what to record on

An 8cm record. It plays at 45rpm, and this side runs for 1 minute 25 seconds.

With my lathe back in action, the hunt was also on for different materials to try out as recording media. Redundant CDs/DVDs/CD-Rs seem to be made of a relatively soft material, because they are the easiest to indent, and the sound level is reasonably high (although still quite a way below that of a commercial record). Sound quality drops away markedly if the groove is continued more than half-way in from edge to hub, and many record players will not allow the playing arm to move closer to the centre than this.

Clear polycarbonate, the stuff used for greenhouse and conservatory windows, as well as model-making, is widely recommended on the Web for disc cutting and indenting. I have, to date, used the Makrolon and Lexan brands, ordering it in A4 sheets 1mm thick and making discs of 8, 15 and 21cm diameter. The 1mm sheets can be easily cut to size with scissors. Records made in Makrolon or Lexan end up a fraction quieter than those made on old CDs. I haven't been able to spot any difference in the quality between recordings on the two brands.

A record made from a plastic box. It sounds dreadful.

The 8cm discs are really a bit of a joke, as they will only allow about one minute per side, and the sound is fairly lo-fi. Plus, some record players can't play them. 15cm records allow two to three minutes at reasonable quality; I chose this size because it's perfect for pop music and two such discs can be cut out from one A4-size sheet, which works out at about 60 pence per record. 21cm records, the largest that can be cut out from A4, are therefore something of an extravagance; I can get about 5 to 6 minutes on a side, but having the source and turntable slowed down to 33rpm means the cut takes about 8 minutes, which is sailing dangerously close to the limits of the gramophone motor that powers the feedscrew.

Getting into truly experimental mode, I tried a PVC record sleeve (didn't work at all), and part of a stout but flexible plastic box that my mum's Royal Wedding plate came in (playable, but terrible quality). British LP sleeves made in the 1960s are coated on the front with cellophane, and I had a go at making an 'instant picture disc' by recording on a piece of sleeve. The results were playable, but almost unlistenable.

Some Lathe Trolls have successfully made recordings on a certain brand of plastic picnic plate sold by Safeway in the USA. Ever the optimist, I bought a set of four brightly-coloured plastic plates from Tesco and attempted a recording, having cut the raised edges off, on the one whose colour I would least like to have in my kitchen: a truly emetic shade of purplish pink. I did get a recording, but it was accompanied from start to end by a horrendous squealing sound. These plates were made of polypropylene (recycling code 5).

Drums of CD and DVD blanks usually come with one or two dummy CDs with no coating, and these can be used double-sided. Usually they are made of the same material as the genuine CDs, in which case no problem, but some are made of polystyrene (recycling code 6). These latter will record, but will play back with a lot of hiss.

Mid-May 2011: some thoughts so far

As I come to the end of this phase of my project, I'm more than satisfied with progress so far. To be frank, it never ceases to amaze me that the thing works at all. That said, making a record successfully is still very much a hit-and-miss affair, and there is still a shopping list of problems to be solved as and when finances will allow:

• Speed of turntable not rock-steady. Possible solution: find a turntable with more torque.
• Because the records are cut with a steel gramophone needle, they have to be played with a (light-weight) 78rpm stylus. Possible solution: find a way of producing a finer cutting stylus, so LP stylus can be used for playback.
• Sound is still very quiet compared to commercial pressings. Possible solution: use speakers with higher power rating.
• Difficult to gauge cutter pressure using the vertical feedscrew, leading to miscuts. Possible solution: allow cutting head some vertical movement, and use a spring to counterbalance it.
• Feedscrew motor only runs for a few minutes. Possible solution: replace gramophone motor with an electric one.
• Stereo doesn't seem to work very well, if at all. Possible solution: forget about stereo, as cutter works best when fed mono signal.

Finally, here's an MP3 file of extracts taken from two Lexan records I made today (15 May 2011), of out-of-copyright recordings: Elephant Tango by Cyril Stapleton, and Route 90 by Clarence Garlow. I've also knocked up a brief video which shows the lathe in action, and playback of another out-of-copyright pop hit: Hit & Miss, by the John Barry Seven plus Four. For the sake of sensitive ears, the first part of the video, showing the cutting process, is silent.

---

June 2011: Some real science!

By the last week of May I had developed what I could call a working record lathe. The machine produced playable records, but there were obvious areas for improvement, as noted above. With some apprehension, and feeling something of a poor relation because of my lack of any kind of practical engineering or electronics skills, I subscribed to the Lathe Trolls bulletin board, introduced myself, said what I was doing and put out a general SOS for advice. Within a day or two my appeal had elicited several helpful responses and suggestions; one contributor even e-mailed me a technical article on stylus geometry for indented records (or, to use what appears to be the term in common use, "embossed records"). I'll admit to not fully following every last word and equation, but the upshot seemed to be that a finer embossing stylus would improve frequency response. So I set to, with my basic kit of a handyman's drill, a jeweller's glass, a whet-stone, a keen eye and a steady hand, and made several attempts at sharpening some steel gramophone needles. The idea is to create a "truncated cone" on the stylus tip. I found that an angle of 90° at the cone's apex yielded a groove that was difficult to track with the playback stylus; it skipped easily. A slightly acuter angle eliminated this, and there was certainly some improvement in sound, but the most obvious improvement was a considerable reduction in surface noise.

My next task was to do something about the feedscrew mechanism, which had been codged up from old bits of Meccano and a series of incompatible parts held together with Araldite glue. This meant that the groove spiral was not regular and even occasionally bumped into itself on successive revolutions of the blank. Having paid a visit to eBay and sold a few more vintage records out of my treasured collection (a long story...), I had enough to invest in some properly made parts: a pulley wheel, a belt the right length, and some bearings and collars. This upgrade also entailed the replacement of the M8 threaded rod with an M6 one, as I could not find, anywhere, a pulley with set-screw the right size for an 8mm axle.

The next tip from Lathe Trolls was to make some tests on what my home-built cutting head was actually capable of recording. So far I'd just been having fun, making copies of a few of those well-loved vintage records - but now it was time to do some serious testing. I put together an Audacity sound file consisting of tones at various frequencies, starting at 125 cycles per second and doubling up to 8000. The same sequence was repeated several times, before and after applying the "inverse RIAA" transformation. I made a record of the sounds, and then replayed the record into Audacity for analysis. The cutting head performed surprisingly well on low frequencies, and did so up to 2000 c/s. The 4000 c/s tone was much quieter, and the 8000 just about audible.

High speed spin cycle

At this point I decided to go down a path that I had considered earlier, but not yet followed: 78rpm. My earlier attempts at half-speed recording had been thwarted by the unsteady performance of my record player's turntable at slow speed: it wasn't capable of maintaining 22½ rpm and overcoming the drag caused by the cutter. At 45rpm, though, the additional momentum largely eliminates this problem; and Audacity is capable of speeding or slowing the source to any desired speed. I made some test records cutting at 45, with source suitably slowed to give correct playback at 78, and from these made some adjustments to Audacity's "inverse RIAA" curve.

By recording at four-sevenths playback speed, the "window of recordability" for frequencies is effectively shifted up almost an octave. This means that some lower frequencies will inevitably be lost, and to minimise this I made the bass reduction part of the curve much shallower.

The upswing in sound quality and volume was a very pleasant surprise, and these 78rpm records will play with either an LP stylus or a 78 one. The sizes I'm using now are 122mm (a gnat's whisker smaller than a CD) which allows just over 2 minutes per side; 180mm (the same gnat's whisker larger than a 7 inch record), on which about 3'30" can be accommodated; and 210mm, which gives about 5 minutes.

I'm still looking for a black medium to record on, in order to make the finished article look a bit more like a "proper" record. I tried a sheet of black acrylic, but the thinnest I could find was 3mm, which is monstrously thick for a record, and the recording quality was much poorer than with clear polycarbonate. Black polycarbonate does exist; I have not yet seen it for sale to the UK, though, and its price is several times that of the clear material. A possible fudge might be to use two sheets of clear 0.75mm Lexan, one-sided, and glue them back to back with a sheet of black paper in the middle!

More soon...

---

Mid-July 2011: update

I haven't made any changes to the lathe itself since June, but this update is just to give a brief round-up of some tweaks and tests I've carried out.

• Stylus sharpening. Like almost every other aspect of this project, my technique is primitive: fix raw gramophone needle in chuck of DIY drill-driver, and hold at an angle to a whet-stone to sharpen the point to a neat cone. Best results seem to be gained from a stylus with a cone with apex of about 50° to 60°; angles outside this range tend to cause tearing of the disc.
• Media. Black polycarbonate remains elusive, but I did manage to get hold of some white (the colour is called 'opal' by the manufacturer, but it's white) of thickness 2mm. The recording turned out very noisy, like a shellac 78; my guess is that this is caused by the pigment, and would be exactly the same if the material were black. Hunt for black polycarbonate therefore called off!
• Circular records. Up till recently I cut out my records with scissors, after recording. This gave a passable but not very professional appearance. The Olfa CMP-2 circle cutter will cut circles in 1mm polycarbonate, up to 30cm diameter. (It can't quite manage to cut 2mm poly.) My first efforts with this tool resulted in decidedly oval discs, but I think I'm getting the hang of using it. The centre spindle of the CMP-2 is a sharp conical point at the base of a rod about 4mm in diameter, and to get a circular circle, you need to drill a 4mm hole and push the spindle of the CMP-2 in far enough so that the cylindrical part of the spindle fits snugly in the hole. You may need to put a bit of cardboard under the thing you're cutting, and you'll definitely need to anchor it with clamps. In the case of polycarbonate discs for records, the 4mm hole can be drilled out to 7.5mm afterwards.
• Picture discs. I have made a couple of these, by using two sheets of thin (0.75mm) Lexan and sandwiching the picture between them. I still have to find an adhesive that spreads evenly and dries transparent. Vinyl floor tile adhesive leaves a very visible splodgy mess. The thin Lexan at 0.75mm can of course be used as normal, recording on both sides, but it is a bit too floppy to be used for records greater than 18cm diameter.
• Labelling. A good record label will be informative and attractive - it can be treated as a miniature artwork in its own right. Some record players with an automatic return mechanism may reject prematurely if the groove continues to within a diameter of 80mm, and thus the groove of any records intended for distribution (of which I have not made any yet!) should keep out of this inner zone. Self-adhesive labels can be bought in A4-size sheets, and labels designed with a good graphics package such as Inkscape. An inexpensive light-duty circle cutter will give much nicer results than trying to use scissors - although I still have to cut out the centre hole with a pair of nail scissors, not having found a 7.5mm hole punch! You can also print on plain paper, and glue the labels by hand. Copydex is fine for this, but it stinks to high heaven. Be warned that Pritt Stick appears to work, until you flex the records slightly... and then the labels fall off.
• Higher-fi. The ever-dependable Audacity sound editor can reduce speed and pitch of a recording by any desired percentage, and I have carried out some experiments with cutting at 33rpm for 78rpm playback, with more tweaks to the 'inverse RIAA' curve and a further improvement in high frequency reproduction. The drawback is that the feedscrew of the lathe, powered by the clockwork gramophone motor, runs independently of the turntable speed, and is already at its slowest setting; thus 33rpm recording uses more 'real estate' per minute of music than faster speeds: about three minutes can be accommodated on a 21cm record.

I'm also considering making a 'demonstration disc' available - this will consist of various excerpts of royalty-free music sourced from various corners of the Web. Coming soon, I hope...

Some home-made records:

Upright: two picture discs and a normal one. 23cm diameter.

Angled: two standard-sized 'singles'. 18cm.

Flat: three 21cm discs and one each 12, 8 and 6cm diameter.

All these records are 78rpm except the 6cm one, which is 45rpm.

---

August to November 2011: lathe Mk.2 takes shape

Towards the end of the summer I was finally convinced that I had done all that I could to the Mark 1 version of the lathe. One problem that had been nagging me particularly was that the gramophone motor, which powers the lathe's feedscrew, was really running too fast even although it was wound down to its slowest setting: I could just about get three minutes' worth of music on to a seven-inch record, and then only by taking the groove right up to the label. In order to take the project to its next level, I would have to see about replacing the clockwork motor with an electric one. This, of course, would mean spending some money™, not just on a motor, but also on a battery or D.C. power supply to run it. I wanted to use new, properly machined parts for the driving mechanism, rather than any more battered Meccano; gears, bearings and shafts are readily available on Internet, from a number of online stores, but these parts don't come cheaply.

My first plan was just to replace the clockwork motor with an electric one, and use some kind of rheostat to vary the speed. This approach would have worked reasonably well if I'd been happy to settle for a record that just consisted of the sound track. However, "real" records have other features, such as run-in and run-out grooves, and a locked, circular groove at the innermost point. If I wanted to attempt something like this, then I would need some kind of shifting gearbox, as it wouldn't be possible to alter the motor's speed enough by electrical means alone.

It was, then, time to find out exactly what a "proper" record looks like. My research led me to the following list of standards for a so-called "seven-inch" record:

• Diameter: 175mm (actually six and eight-ninths inches!);
• Run-in: 4mm of radius;
• Start of music track: 167mm diameter;
• End of music track: 107mm diameter, minimum;
• Locked groove: 96mm;
• Label: 85mm;
• Centre hole: 7.4mm

I had never realised, in nearly thirty-five years of collecting records, that there were so many standards.

Having made a strong cup of tea, and with these requirements in front of me, I began work on designing my motor/gear system so that the finished records would adhere - with luck - to at least most of these standards. The most important point to emerge from this was that a seven-inch single has only 30mm of radius in which to say all it wants to. Put another way, a three-minute pop song will require 10mm of record per minute.

Things were starting to get complicated, and around this time I came to two decisions. The first was to try to design a better cutter head, and aim to make 45rpm records instead of 78s. I had cut a number of 78s on the old lathe, using 45rpm to record, reducing the source speed appropriately, and this had yielded good results. What if I wanted to play back at 45? The answer, derived by simple arithmetic, is 26rpm for recording. Some tests on the old lathe established that this should work satisfactorily if I could improve the cutter head somewhat. With this information, I flung together an Excel spreadsheet listing recording and playback times, lines per millimetre, the times to advance the feedscrew by one millimetre. Eventually I established that I would need the motor to turn between 50 and 100 times per minute, and that, in order to do both sound track and run-out spiral I would need 25:1 and 1:1 gear ratios.

A finished 6 8/9-inch (175mm) record, created using the Mark 2 lathe. This particular one is featured in my YouTube demo

The design for the simple, two-speed shifting gearbox was roughed out in half an hour. It consists of three shafts, with the shifting one in the middle. The input/shift gearing drops the speed by 5:1, and, depending on the position of the shift, the shift/output ratio is either another 5:1, giving a net reduction of 25:1; or 1:5, yielding no effective change between input and output speeds. This was the easy bit. Much trickier was locating a suitable set of gears in order to put this together. I had to plan meticulously what to buy, and research all the parts thoroughly before placing my order - I'd get to the end of my shopping list and find that one of the gears was specified with a wider bore than the others, which would mean using some kind of shaft adapter or trying to do my own precision machining - for which I have neither the skills nor the specialised equipment. Eventually I located a set of '1.0 module' gears with a consistent 4mm bore across the whole range of sizes.

The second decision was to build a completely new machine, instead of trying to graft the new parts on to the old one. After several days of banging about in my woodshed I had the basic frame together. The main difference was that the motor and gears would be situated under the moving platform, rather than extending in front of it, and this made the whole thing that much smaller: the Mark 2 machine is exactly half a metre square.

I have put together two YouTube videos that explain the construction of the new lathe. The first gives a brief overview and a demonstration of a CD being turned into a record. The second is a more detailed description, with commentary, and demonstrates a 7-inch record of four samples of royalty-free music sourced from various sites on Internet.

The Mark 2 lathe embossing a record.	Close-up of record grooves being embossed.

---