A week in the shop, and a week in Nantes

Return of the weeknotes, though I fear the next couple of weeks will be sparse guitar work wise due to August being a month of travel. The week before last I had family visiting for much of the week that kept me busy, and then last week I was on actual vacation in Nantes, France, enjoying a combination of the annual arts festival and just switching off and indulging in things like cooking local produce from the market in our rented apartment. Between these events I failed to write up what I did before all this week, so here it is below.

This coming week I’ll be mostly doing paid software work whilst I'm at home, and then I’m in Helsinki for a few days (tagging along with my other half who is out there for work), before heading to EMFCamp, a weekend long hacker festival where I’ll be talking about how I got started in a new domain again. If you happen to be at EMFCamp do say hello!

Anyway, what is happened last couple of weeks, and normal service should resume in September.

I started the previous week with two guitars hung up and ready to move forward: one made from parts the CNC router had tried to destroy and I’d patched, and the actual bits from the first of the commissioned offsets (both of which need names). What unites them both at this stage is that they need fretboards next.

In the past I’ve radiused fretboards by hand using a radiused sanding block, but you have to be very careful not to introduce any bias as you do so: it’s far too easy to take a little off one side more than the other at the diagonally opposite corners, due to how your arm moves relative to your body even when you think you’re pushing straight. The common solution to this is to use a jig to make sure you don’t go down too much at any point, but I wondered if perhaps I could use the CNC router to radius the fretboard roughly, and then just use sanding block to remove the machine marks rather than the bulk of the wood.

Whilst it sounds simple enough, the worry I had is that, unlike most people, I use a laser cutter to etch guide lines for my saw to cut the frets. Being an optical device a laser cutter has an ideal distance above the cutting surface to ensure it is properly focussed: if I radius first then my cutting surface will no longer be flat, and thus I can't guarantee it'll be in focus at the edges. This might mean the grooves I etch aren't deep enough to sit the saw in to get it started.

To test this I got a bit of scrap wood and used my radius block on it shape it like a fretboard, and then I used the laser cutter at various powers to etch the 0.5mm slot I’d normally use to create my saw blade guide. You can see the results here:


First thing to note: because I used ply wood and just did a quick job of the radiusing you can see the contour lines show the biasing effect I’m talking about. As I’m right handed you can see despite thinking I was sanding straight I’m actually going sort of from bottom right to top left slightly. This is what I’m trying to avoid with all this talk of jigs or CNCing.

Next thing to note: the radiusing had no notable effect on the slot I etch, proving my fears were unfounded. The fall off at either side is only around 1.5 mm, and that isn’t enough to cause the laser to lose focus. I did however find that I had to be much more careful when cutting the slots with the saw. Compared to cutting the slots before radiusing the board it was much easier to slip out of the grove, as you have a smaller contact point with the wood when sawing. 

Overall it worked a charm though, albeit with the caveat you need to be extra slow when sawing the slots. The frets you see in the above picture are held firm without glue, so that’s a green light on this approach in general.


Next up then is to get the CNC router to make a fretboard with a radius, or more specifically to get Fusion 360 to generate the toolpaths to do so. This turned out to be much harder than I had anticipated, and I lost easily a day of work just trying to tweak toolpaths to ensure Fusion didn’t do something I didn’t want.

Screen Shot 2018-08-20 at 20.30.50.png

Fusion 360 in general is more aggressive with its tool moves than I like for guitar work: it’ll happily assume it can cut through 10mm of wood in a single pass, whereas typically I want it to move more slowly to ensure a nice finish. It took me quite a while to get Fusion not to do occasional plunges through the material, which whilst not incorrect would have not left me with the finish I wanted. It was a nice reminder that whilst I’ve used Fusion quite heavily for the last year it’s mostly within a set of regular operations, and outside that I’m still a bit of a newbie.

Once I had a toolpath, I then set out to make a test fretboard. I found a old maple fretboard blank I had in stock, and set about cutting it out, trying to fit this in at the last moment before I went on vacation. At this point I also had family visiting, and my brother and brother-in-law asked to watch, so I set them up with PPE and got started.

Initially things went well, but in the end due to a combination of being tired and feeling watched I made a mistake, resulting in my fretboard blank being ruined. As is always the way with a new set of toolpaths, I end up tweaking them on the fly as I learn more about what each pass is doing. On this occasion I realised that the bulk clearing paths for the fretboard didn’t account for the initial levelling paths I’d created to get the blank down from 9mm to the 6mm required. So, rather than watching the CNC router cut air for a good 10 minutes, I aborted that part of the run and went back to Fusion 360, tweaked the tool paths to start further down and set things going again. Unfortunately, I failed to double check what the updated toolpaths would do, and failed to spot that I’d ended up generating toolpaths that cut behind the headstock transition and through the fretboard itself. Doh.


In the above image you can see on the left the headstock transition is a weird island on its own rather than a ramp down at the end of the fretboard. At least this was my mistake rather than a failure of the CNC router: me being wrong is generally easier to fix than the CNC router being wrong. But all the same, as this was the last free evening before heading to France, I was quite frustrated not to have made at least one fretboard before I went away. As ever, this just reinforces the measure, measure, cut mantra: due to being tired and the pressures of being watched I’d not double checked my toolpaths, so I only have myself to blame.

I quickly ordered a couple of more blanks before going away, so I’ll see if I can sneak in time this week to do another test cut of my fretboard design.

However, in general we had a fun week with my family around. I introduced my nieces, Leah and Sarah, to 3D printing, letting them pick an item from thingiverse (they went for a Pokemon) and printing it out there and then. I think it's one thing to hear of 3D printers, but it's another thing to actually see a thing you picked appear out of nowhere.


With luck that'll inspire them to make things themselves one day.

I also took an afternoon out to take my brother Tristan and Sarah to a rehearsal studio for an afternoon where we had quite a fun jam session. Tristan is a very talented guitarist, playing with IKARI, and ordered and plays the second guitar I ever made, and Sarah has been taking guitar lessons for quite a while now, so it seemed a reasonable idea to collect up the guitars I had and my amp and find somewhere we could make some noise.


Tris hadn't yet had a go on my amplifier, so he gave that a going over. and seemed to like it for the cleans on his IKARI parts (I have to confess, I wasn't aware he did cleans, but you learn something every day ;).

It was a bit last minute, so I failed to secure a left handed guitar for Sarah in time, but she did a valiant attempt with the right handed guitars we took along. In addition to playing some of the stuff she already knew, I managed to teach her the blues scales enough that we could have a simple jam along together.


Tris also taught me the rhythm part to Ocean Waves and we jammed that together. A great afternoon of fun just playing guitars with family, and a reminder of why the guitars are worth all the effort!

Much to my surprise, although I took the vacation time to totally switch off, there was a surprising amount of cool guitar stuff at the art festival. First was Fluides, by Céleste Boursier-Mougenot, where in a netted space filled with dozens of small birds that would come to rest upon a number of face up mounted guitars, which were wired through delays and reverbs to amplifiers. Each time a bird would land on a guitar or shuffle along the neck you’d get a lovely ambience sound piece.

In Daniel Firman’s exhibition, Outside, one of the pieces was called Drone, which featured three telecasters being spun at speed from their headstocks, with weighted spindles that would occasionally strum them. The guitars were each tuned to a different range, so the effect was somewhere between a didgeridoo and Mongolian throat singing. 

Finally, I was surprised to find out in the rural areas south east of Nantes a giant Les Paul emerging from a round about:


It turns out that Hellfest, a massive annual heavy metal festival, takes please each year in the middle of the Loire vineyards. All the villages in this area compete on decorated roundabouts - I assume this one must be one of the more unusual ones!


Wilderness highlights

Every year for the last few years my other half and me join a few friends for Wilderness Festival, an event spanning four days and a wide array of cultural activities: music, spoken word, theatre, dance, swimming in the lake, a very silly cricket match, and lots of nice food and drink. Also, being in a field outside somewhere in Oxfordshire, the cell network can’t cope with the sudden influx of 30000 mobile phones and you have no Internet access for most the event, so you can just switch off and immerse yourself in all the fun thing on offer.


What’s on music-wise varies from year to year. In particular, there’s one stage (this year called The Hereafter Stage) which seems to get a very different set of programming each year, so hasn’t always been to my tastes; but this year it was mostly focussed around blues rock, so Laura and I spent a lot of time hanging out there. Pretty much all the bands were small up and coming bands, none of which I’d heard of before, but now I’m slowly working my way through the list I build up. Here’s a few of the 17 bands we saw over the weekend that I recommend you check out, just in the order I saw them, so you can make up your own mind about which ones are the best.

  • Elles Bailey is a great blues vocalist and made a great start to the festival with her songs
  • Cosmic Strip were more psych-rock than blues-rock, but their general spirit fit in well on the Hereafter stage. Sonically very different to most things we listened to, with more ambient sounds and delays segueing to more straight rock, but great stuff all the same. They also did a cover of a Dandy Warhols song, which is a sure fire way to seal the deal for a big Dandy’s fan like myself.
  • John Fairhurst did some amazing solo guitar work, just him and his resonator guitar, lots of blues textured fingerpicking songs, often without lyrics but still very captivating (which solo guitar can be quite hit and miss with). We unfortunately missed his electric set later on the last day, but definitely a very talented guitarist worth checking out.
  • True Strays are a crazy bunch: we saw them late afternoon, after they’d driven up from Brighton starting out at 6am having just finished a festival they were hosting, and still put on a rocking show full of energy. More at the rock end of blues rock, it was still great music with heart; their singer/guitarist rocked some great slide guitar on another resonator guitar. They’ve just put out an EP on band camp, so you can go listen to that. 
  • Whiskey Shivers are a five piece from Austin, TX that describe themselves as Trashgrass. For their last two songs, having already done a high energy set that got the crowd on their feet,  they got off stage and into the crowd and just played acoustically, finishing with a gentle rendition of Daneil Johnston’s True Love Will Find You In The End, which was such a touching way for the last act of our festival to end.

A week in the shop and in a field

Last week I cut the neck for guitar #5, the much delayed first commission custom build, and this week I cut the body. As stated last week, I still don’t trust the CNC Router at Makespace particularly, but between the desire to make some progress and my now better understanding of the machine itself, I felt time had come to re-roll the dice. I was more nervous with the body compared to the neck for two reasons: firstly, although the wood from which the neck is cut is more expensive in monetary terms, a body blank represents half a day to a day of work by time I get to this point, so is more expensive to me personally; secondly, the body requires more cuts and more complex cuts than the neck, so more opportunity for the CNC Router to get confused and do the wrong thing.

However, like the proverbial kettle that never boils, it seems that by watching the CNC router like a hawk for the entire process (my thumb hovering over the abort button at all times), I was able to stave off failure, and the body came out all okay.

There's much relief on my behalf that this is done. Now all I need to do is make the fretboard and we’re on route to getting this guitar onto the hand making part.

A couple of learnings from making the body. Firstly, I utilised the discovery I documented in last week’s notes that the machine’s ability to find the origin after it’s been powered off and on is more accurate than people suspected, to within a tenth of a millimetre. This meant I was able to make the guitar body over two days due to being short of time, rather than having to do it all in one session (CNC routing a body like this can easily take 4 or 5 hours all in). As the saying goes, the proof is in the pudding, and after I finished machining the body the seam where it was cut from the front side on day one and the rear side on day two is very close: as closed as I’d expect had I flipped it anyway, so I’m very happy there.


The other learning is in the triangular tab shape there. Most design tools insert rectangular tabs to hold pieces being cut out in place, but the down side of rectangular tabs is that the tool is briefly paused in the X/Y plane whilst it moves up and then later down, and at this point it will wobble slightly, leaving a groove that has to be sanded out. I spotted Fusion 360 has an option to make triangular tabs, which mean that the tool never has to dwell in one position, and so there’s no ghost grooves around this tab.

Next up for this body and neck now will be sanding. I finally gave in trying to sand by hand, and upon the advice of Matt I’ve got myself a proper orbital sander along with some professional quality sanding disks, so hopefully making the body nice and polished will be a lot easier.

This was about the only progress I made in the shop, as I spent the later half of the week watching live music for a change over at Wilderness Festival down in Oxfordshire. In the past there’s not been a wealth of music that has been of the style I’d normally go for (though there’s been good stuff that I’ve discovered that way), but this year they had a new stage, called the Hereafter stage, which was programmed with lots of great blues rock bands.


I’ll write up a list of my faves here shortly once I’ve caught up from being in a field without Internet for 4 days.

A week in the shop

This blog seems to be rapidly turning into a blog about CNC routers rather than guitars, for which I can but apologise, but it just reflects the reality of my situation at the moment, in that it continues to occupy my time as I try and find a way to get a reliable, repeatable, and efficient way to make guitars. Although I’ve convinced myself I can make bits by hand now, it’s not particularly efficient, and I do think quality wise certain bits are more reliably/repeatably produced on the CNC router, even if I’m using it for mostly roughing things out.

This week I decided that the misbehaving CNC router at Makespace is what it is, and given I’ve failed to find an ready alternative I’m going to try push through at Makespace with what I have. Before doing any guitar building on it, I had two tasks I wanted to get done on the CNC router care/maintenance front. Firstly, I spent an afternoon creating a new sacrificial bed for the CNC router. What should have been a simple job turned out to be more complicated due to the way most CAM tools work. I’ve documented all of this in another post, so if you’re interested head over there, but the end result was the CNC router had a nice new sacrificial bed that was actually level, unlike the old worn one which was all over the place.


Having levelled the bed, I then set about another experiment, which was to try measure how accurately the machine is in getting back to the same set of positions on the bed after you switch it on and off. You might assume that the CNC router knows where it is at all times, but that’s not the case: when you power it on each time the controller logic knows nothing about the location of the cutting arm - it has no way of measuring the arm's location at any point in time. Instead when you power it on the controller will move the arm into one of the corners until it can do so no more. Once it’s done this it then knows where the origin is on the bed, and from then on it keeps track of how how much it has moved the arm by, so it can then keep track of the arm from here on out. But if I switch off the machine and ask it to find the origin again, how accurate is that? My motivation is that it seems a lot of the router failures we’ve had are with more complicated multi-stage designs, and so if I can reset the machine between stages perhaps it’s less likely to ruin my pieces. 

The way the CNC router detects that the arm is in the corner is done with a set of physical switches: the CNC Router sets the motors going towards the origin and when the switches are pressed it stops, so how accurate the origin is is down to how accurate these switches are. Opinion seemed to be divided amongst people I asked, some suggesting it might be as much as 0.5 mm out after a power cycle, which would be a bit too inaccurate for my use case. But there’s nothing like testing something yourself, so I made a simple design referenced at the machine origin and just ran it on the same bit of wood mounted in the same position 5 times, between each run I’d power off the machine and get it to reset itself.


The end result, whilst not hugely rigorous, indicates that the reset is accurate down to around 0.1 mm, which actually for guitar building is acceptable. At some point I’ll repeat the test with a more accurate measuring device than the callipers I used, but the initial results are promising. Knowing I can turn the machine off and on again between phases starts to give me a route to more confidence in getting the machine to make something.

But I’ve finally had enough CNC router naval gazing. We're drifting into August and I've still not produced the pair of commissioned guitars I’d aimed to have completed by May, and I decided to take the risk with the CNC router as it is, tampered by all the knowledge I've gained about how the router works and what its limits are in the last few months, and start on guitar #5 again. This was the original request to build an offset guitar that has pretty much defined all my guitar building this year. It was this guitar that the CNC router started failing on, hence the delay and why this blog is now more about CNC routers than guitars.

Even though I don’t understand why the machine fails properly, we have some ideas, and I do know that when it fails there is usually some sort of quick indicator before it does so, so if I just watch the machine like a hawk through the cutting process, there’s a good chance I can stop it if it starts going wrong. I decided to go all in, and broke out the nice bit of birds eye maple destined for the neck of guitar #5 and cut the neck.


The birds-eye maple is really lovely up close:


All went well for the most part. I did abort one of the CNC router phases due to a suspicious noise happening at one point, and so did have to re-start that stage, but an hour or so after I started I had a neck cut out.


So, some success after all our failures up til now. Next up will be the body for guitar #5. My aim is to get the main bits for this guitar cut out this coming week before I vanish to Wilderness festival. 

I did also find some time to take the damaged body and damaged neck that I’d been repairing and sand them down to get the neck to fit the body spot on. This here is a picture of the neck held in just by friction alone, which is always a good sign.


I'll be interesting to compare this body, where I did the comfort carves by hand, with the one I hope to do on the CNC router in the coming week, and see which is more pleasing :)

I wanted to try a fuzz pedal a while back (I was playing lots of White Stripes at the time) so I bought a fuzz pedal kit from jed's pedals (obviously someone like me struggles to buy ready made things when they can have an excuse to build something), and I've been trying to 3D print a case for it. However, the epic heat wave has really been making it hard: the design is not really any different to the ones I did for the clock a few weeks back, but I’ve had about a 75% failure rate, which I put down to the high temperatures causing the printers to not want to play ball. I have finally managed to get a case printed out though for test fitting:


Seems strong enough, which I wasn't sure about. Fun thing about the lettering: that font I've used for the name "Fuzz Zero" should have straight edges, but instead it's got this cool wobble on the lines. I'd love to claim this was some original graphic design on my behalf, but it's a pleasant side effect of me setting the laser cutter to run too fast for cutting outline, and that decaying wave you see is the mirror in the laser cutter wobbling as it is forced to change speed too quickly (I'd set the machine to etching speeds, but then set it to do outlines by mistake).

Clearly plastic and acrylic don’t make for the best case in terms of shielding, but eventually I’ll probably make this out of a nice wood and copper shield the innards. Next is to find a spare half hour at some point to solder the board up!


How to level a CNC router sacrificial bed with Fusion 360

I’m now one of the maintainers of the CNC Router at the Cambridge Makespace community workshop, and one of my first tasks as maintainer has been to replace the sacrificial bed that is mounted on the CNC router.


The sacrificial bed is the large wood base that is permanently mounted on the router to which work pieces are then in turn mounted (which is nice and fresh in the above picture, taken after I'd replaced it). Although in theory you can mount things to the metal bed on the router, for the majority of jobs done in a space like ours (particularly given our machine is only rated for wood and plastic work) a sacrificial bed which things can be screwed to readily simplifies machine use somewhat.

The downside is that over time the sacrificial bed is worn away by people screwing things into it and cutting through into the bed. At which point you face an interesting challenge - how do you make a new bed that is by definition exactly the size of the working area of the machine? It turns out that doing this in Fusion 360 required some non-obvious trickery, so I thought I’d write it up. This is what a well loved sacrificial bed ends up looking like:


The first thing I did when I started this task was look for existing design files that I could re-use. Unfortunately they seemed to have been lost over the course of previous CNC router owners coming and going, so I had to make my own. Thus I broke out the callipers and fired up Fusion 360 and recreated the bits of the CNC router I cared about: The metal base onto which the sacrificial bed is mounted, plus a model of the existing sacrificial bed layout wrt screw holes for mounting etc.

Screen Shot 2018-07-27 at 09.54.57.png

The bed is made from 16 metal extrusions, which are nicely symmetrical, but with the notable feature that there is no center slot, so the two middle screw holes have to be staggered either side. But that aside it’s a fairly simple thing to model. To try and and save the next person having to recreate the design yet again, I shared this design with the Makespace community - a nice upside to Fusion 360’s cloud based nature.

Having made the model, the next step was to make the CNC router make the new sacrificial bed. This is done in two parts:

  1. Firstly, we mount the new bed on the old bed and make the counter sunk screw holes to let us mount the new sacrificial bed on the metal base
  2. Secondly, we then face off the new sacrificial bed to make sure it’s completely flat

For those not familiar with the term, facing off just means making sure the material is absolutely flat, so we mill down the surface to remove any variation due to imperfections in the material or the base its mounted on.

The first stage was easy enough. Although the sacrificial bed is quite large, the mount holes are by nature within the area of movement for the CNC router. Thus I simply screwed the new bed onto the old sacrificial bed, cut the mount holes, and then was able to quickly swap the old sacrificial bed out for the new one.

The second stage though proved quite problematic: here we need to go edge to edge on the new sacrificial bed and the material size is actually 20 mm bigger than the movement size of the CNC router. This is okay, as I have a 1 inch (25.4 mm) wide router bit for levelling jobs like this, so the movement of the machine head plus the diameter of my router bit means in theory we can face off the new sacrificial bed without problem. The problem however is trying to get the software side of things to understand those constraints. I spent a good 90 minutes with both Fusion 360 and Vectric VCarve Pro trying to find an elegant way to express what I wanted to do with no joy. The fundamental problem is that neither bit of software understands the constraints of the machine you’re using, just the constraints of the material.

Screen Shot 2018-07-27 at 09.27.40.png

Look at the above picture as an example: here I’ve told Fusion 360 that I want to face off the sacrificial bed using my 1 inch router bit. You can see that it’s made this nice pattern that relies on going off the side of the material and back in again. I can’t do that in this case as I’m at the limit of my machine’s bed here: there’s no room in which to execute that turn, and I failed to find a way to express that constraint to Fusion 360 in any of the options.

I tried some other tool path generation techniques in Fusion, for example using its adaptive clearing method to just say make the surface of the material match the model, which should be equivalent to facing off here. Here I hit another issue with Fusion 360, in that it still wants to enter the material from the side. Even if I try explicitly telling Fusion I do want’t to have a lead in, it still has some partial path coming in from the side. 

Screen Shot 2018-07-27 at 09.33.44.png

After much fiddling, I managed to stumble across a solution. I reasoned that if the sacrificial bed was treated like a pocket, then surely it’d not execute a lead in manoeuvre, as in pockets you'd normally hit a wall doing so. Thus I set my piece origin to the center of the material rather than a corner as I’d normally do, and used the 2D pocket path, and indeed there was no lead in issue, but it still insisted on going outside the material at first:

Screen Shot 2018-07-27 at 09.39.02.png

Because there are no walls on the edge of this pocket, Fusion 360 reasonably assumed it could do that. In the end I realised that I could control the step over amount for the spiral shape it is making, and by playing with that number to make it create a slightly tighter spiral I managed to get a tool path that remained within the material near perfectly.

Screen Shot 2018-07-27 at 09.42.00.png

Thus I was able to execute this as my face off pattern and I got the nice finished sacrificial bed you see in the opening picture of this post.

Ultimately the problem here was that I couldn’t find any way to teach Fusion 360 about my machine limits (and it’s not alone in this issue, it’s just that I happen to use Fusion 360, I hit similar issues in my brief play with VCarve Pro to see if it was more sensible). I was getting to the stage where the irrational part of my brain was going to write some software to generate a bed levelling pattern based on machine limits rather than material limits - and the whole point of getting into woodwork was as an escape from writing software!

The other frustration with this kind of problem is that it’s not until you try run your tool path on the CNC router that you have any sense of whether it’ll work or not: do you get the dreaded "x-axis out of range!" message or not. I did manage to work out that using a combination of ncviewer.com to visualise my tool paths and reading the low level g-code I could take a good stab at whether a design would work or not, but reading g-code is not for the fainthearted, and ultimately the whole process was a user experience nightmare for what should have been a simple enough task.

A week in the shop

This last week was not the most productive week in the shop, as I spent quite a bit of time doing software contract work, which is how I pay the bills currently. I did get into the shop for a little bit, so here's a look back at the week.

First up I tried to plug the hole in the neck that the CNC router damaged a while back. Unlike with the body the hole in the neck isn’t a regular 2D shape, it was a irregular gouge taken from the side when the router bit snapped as the CNC router erroneously drove it into the bed. After clearing it up a little with sand paper and a small file, it looked like only a small amount of the damaged area would be visible once the neck was carved. If you look at the picture below then it’s only the top 2 or 3 mm that’ll remain post neck carve.


Given the small area that needs patching I thought perhaps I’d try plug it with a mix of sawdust and wood glue. Initially this looked promising:


But once it dried it looked a lot less good:


I suspect I’ll live with this for this neck, given there’s other areas of pain I need to deal with and the idea is just to make *something* out of these damaged parts rather than make them perfect, but I must confess that after the success with body patching from a couple of weeks ago I'm a little sad that this didn't work out as well, but that's all part of learning. Clearly whilst you can use the sawdust and wood glue trick successfully with dark woods (as I’ve done successfully on wenge) lighter woods it doesn’t work so well on.

My nemesis, also known as Makespace’s CNC Router, also continued to plague me this week. Whilst I don’t trust this machine, it still works will for many people’s purposes and Makespace, and as Makespace is a community workshop and I know more than average about the machine, I’ve become a recognised owner of the machine, and as such I do occasional maintenance on it. 

This week we needed to replace the sacrificial bed on the machine; this is the bit of MDF that is anchored to the bed of the machine that people then mount their workpieces to. Having a soft bed like this rather than a metal bed makes using it a lot easier in general (you can just screw your material to the bed rather than fight with lots of annoying clamps), but as the name implies, the sacrificial bed needs replacing from time to time. 

Replacing it is fine, but when you do so you also need to level it. The general approach to this is to use a wide router bit (e.g., I have a 1” wide router bit I use for facing off material). But I’d noticed a while ago that doing this was leaving ridges in the material, implying the machine spindle wasn’t truly perpendicular to the machine bed. You can see the effect in this picture I took a couple of weeks ago in the face of this body I was patching - those ridges are because the 1” router bit wasn’t truly flat, but at a slight angle.


Before I levelled a new sacrificial bed I wanted to see if I could sort out this problem, as there’s no point trying to level something just to have it not actually flat. The way to assess whether the spindle on the router is true or not us by a process called tramming, where you take a “tram”, which is just a bit of metal that fits in the spindle, goes out about 20 cm to the side, and then down a bit. The idea is that if everything is good, then you can fit the tram to the CNC Router, get the tip to touch the bed at one point, and then spin it 180˚, and it should still be touching. If there’s a gap or you can’t turn it due to the tram hitting the bed then you have a non true spindle. You can see the setup here.


In the picture I’m using a bit of metal over the foam bed cover to make it easier to tell whether I’m hitting or not (whether I get a nice clink or not). Sure enough, when measured I discovered that the spindle was out by about 1.5˚, which is quite a bit for a machine like this.

Now that I knew it was definitely misaligned, the next question is what to do about it. Helped by other, more experienced, Makespace members, we spent an afternoon working back to find the source of the misalignment, which was that the top arm was slightly off, rather than the spindle itself being off. This we found using a Dial Test Instrument (DTI), which is an extremely sensitive gauge that measures small changes in distance as it’s run along a surface.


Using this we measured all the degrees of freedom and was pleased to find the misalignment was just limited to the one axis. Having found the source of the mis-alignment, and knowing how much it was misaligned by, we then shimmed the mounts for the main arm by an appropriate amount, which involved a bit of disassembly and re-assembly of the machine.


Afterwards we were left with a wonderfully aligned spindle, and at some point this week I can find a little time to make the new sacrificial bed for the router.

Overall I’m still feeling frustrated I’m not making progress with more guitars: whilst I have managed to finish the prototype offset, of which I'm very proud as my best guitar to date, I've not made nearly as much progress, or as many guitars, as I'd hoped. I’m stuck in a kind of limbo right now whereby although I can move things forward by hand, there’s certain jobs where I know it’ll be better done by a working CNC router, and the Makespace CNC router works almost well enough that I end up procrastinating trying to see if I can find ways to move things forward from that perspective. I'd hoped my trip to DoES in Liverpool would provide a more readily available path to completion, but whilst I'd still like to do something with DoES that's not a short term win in the way I'd hoped. 

None of which pontificating and procrastination gets guitars built of course. I need to push more into making myself happy that I have a route to completion on the guitars to a quality level I'm happy with over the next few weeks. I think this is where I'm getting stuck: I have a high quality bar for what I think I can ship as a guitar, and I'm not experimenting enough to try and close that gap based on what I think I can achieve with the tools available to me.

I do also need to start moving on the guitars I have made to cover some of my costs. The prototype offset will go on sale shortly after it completes final setup at full price (£1800), so if that model appeals to your, get in touch, otherwise I imagine it’ll appear on reverb in the next month. I’m also going to sell my P90 based blues deluxe model at some point; my plan is to make it a new neck first as it’s not up to the standard I’d want if charging full price, but given lots of people who have played it have loved it I’m also happy to let it go as is for a discount (50% off, aka £900), so if that guitar feels like your thing then get in touch if you fancy a bargain before I rebuild the neck.

A week in the shop

My week notes are a bit late due to a weekend spent learning how to to hunt, gather, and cook (and make fire) in a forest in Kent and doing contract software work from when I got back, but here’s a belated catchup on the fun in the shop last week! I love this image, which pretty much covers the things I've been working on in the shop during that week:


Let’s start with some non-guitar design and woodwork fun. The week before last I talked about how my friend Jason was visiting, and together we were working on making a case for his nice nixie clock that he’s been building. 


As a quick recap from that week: we’d gone through several iterations of the classic rapid prototyping flow: we had a design in Fusion 360 of the clock and the case, which we’d then 3D print using one of Makespace’s Ultimakers, which we’d test fit/measure, work out what needed changed, updated our design, print again, etc. After a few times around, we’d got to the point where we we were happy with the design and we had 3D printed case we were happy with.


We also used the laser cutter to cut a nice acrylic base for the case. Although the case looks simple (it's just a box with some holes, right?), there’s two points of complexity in the design. Firstly, the v1 clock circuit board wasn’t really designed with the idea it’d be mounted in a case, and the layout is a little awkward for this purpose: the headers supporting some of the child boards (e.g., the left most one in the first picture) are very close to the holes where we'll need to mount the board to the case. This, coupled with the USB port that powers the clock being almost flush with the rear of the board, means the mount posts in the box have to be very narrow and have thin walls around the screw holes, as shown here:

Screen Shot 2018-07-18 at 08.28.31.png

The posts look big in that picture, but are only 5 mm across and have to take a 2 mm screw, so we really don’t have much margin for error here before things won’t fit. Indeed, we’re sufficiently close that we have had to route the channel you can see above on the inside of the rear wall to let us slide the almost but not quite flush USB port into the box. The other complexity comes from the fact that the ventilation and USB port are on the rear wall, not on the top or bottom. Why is that an issue? Well, we'll be milling the final box on the CNC router, which only works in one plane, so whilst we can hollow out the box, cut the nixie holes and the screw holes in the mount posts, we can't use that same process to make the holes on the back, we’ll need to do those separately afterwards. Given the rear wall is only 2 mm thick (going down to 1.2 mm where the USB port channel is) we’ll need to be very careful when we eventually drill those holes in, as we risk splitting the material given how thin it is.

Once happy with the 3D printed designs, it was time to start thinking about making it out of wood. Jason ordered some wonderful wild olive wood from Exotic Hardwoods UK Ltd, which is possibly the nicest bit of wood I’ve ever worked with. It looks beautiful, and whilst we were working on it it made the workshop smell very nice. 


The wild olive wood is as expensive as it is beautiful, and thus following our regular mantra of “measure, measure, cut”, before putting our expensive nice wood into the CNC router we first glued together a few layers of MDF I had laying around to about the correct thickness, and tested cutting out the clock case on this. People have a tendency to assume CNC routing is the easy way out: you just feed your design into a machine and press go, but it’s actually quite a labour intensive process the first time you cut a design. The CNC router is a dumb machine that will only do what you tell it; you have to use Fusion 360 to work out all the tool paths you want it to follow, use experience or trial to work out which cuts will work best in a given scenario etc. The first time I machine anything on the CNC router, although I work out all the tool paths before I enter the shop, I’m always revising the tool paths as I go as I realise assumptions I’d made about how best to approach things were wrong, or I can see a better way to get the machine to craft what I want.

As such it took Jason and I the better part of a morning to get the first MDF version milled out:


We learned a lot in this process about how to approach the milling of the wood, and we then could test fit against the actual board to check nothing had been lost in translation. We also at this point tested our plan to make the holes on the back using a laser cut template which we clamped to the back and used the pillar drill to make the actual holes. This mostly worked fine, but we learned that we needed to put a block of wood into the box whilst we drilled to stop the wood splitting given it was so thin.


Happy with our design, we then set about taking the wild olive wood and repeating the process. Even here we ended up having to tweak the design and react as we went along: we quickly realised that the harder olive wood reacted differently to being cut compared to the MDF, having a tendency to chip at the edges if we took too much off, meaning we had to go much slower with the olive wood than we had with the MDF.

Thus a few hours of tending the CNC router and on my side a lot of nerves given I know it's not been reliable on my more complex guitar designs, we had successfully made the shape of what we hoped would be the final version of the case!


Given it was now late evening and we’d been in the workshop all day, we decided to defer cutting the rear holes until later in the week - there's no point rushing on a job like this given the cost of a mistake will put you back to square one. Still, even if we couldn't yet power the board on, it was nice to see another physical manifestation of the “measure, measure, cut” mantra by comparing the 3D printed version, the MDF version and the final olive wood version of the case:


Working with an organic material like wood is full of surprises, and when two days later we came back to our case to drill the rear holes we were initially quite confused: the acrylic template we’d made for drilling the holes (and had used on the MDF version successfully) didn’t match the size of the case we’d made! More weirdly it did still match the size of the other cases, just not the final wild olive one. What was going on? We broke out the callipers, and discovered the rear of the case, which should have been 33 mm tall, was 1.2 mm shorter than expected!

IMG_0574 2.jpg

(as an aside, just check out the lovely flame texture in that olive wood :)

At first we assumed perhaps we’d messed up on the CNC router, but going back to our CAM design we could see everything was good there. Only then did we check on the properties of olive wood in a wood database: it turns out wild olive wood can shrink by up to 15% in volume once cut out! The nice block of wood we’d started with was sealed in wax to prevent it shrinking, but now that we’d cut this bit of the wood free and hollowed it to a series of thin walls it’d quickly started to shrink. Luckily the board still fitted and our screw holes were still just about aligned with the board, but we decided that it was probably best that we quickly get the holes on the rear wall drilled and get to oiling the wood as soon as possible to see if we could limit further shrinkage.

As mentioned earlier, given how thin the rear wall is, to prevent the wood splitting when we drilled the rear holes we found some scrap wood and quickly machined it (a fancy way of saying I measured it, roughed it on the band saw, and then sanded it) to the right size to make a snug fit.


We then taped on our template, noting that despite the shrinkage the internal mount posts in the case hand’t shrunk notably, so we could keep the top edge of the template flush with the top of the box and still have the USB port hole align with the mounted board. 


For the USB port hole, which is small slot rather than a single circular hole, we actually just drilled two holes, one at either end of the slot, and then Jason hand filed the wood in between to make the correct gap.


After which we were able to finally do our first assembly of the box, and it looked just as lovely as we’d hoped:


Now that the case was fully machined, we sanded it down through all the grits: from 120 grit to get rid of machining marks, all the way up to 2000 grit, at which point the wild olive feels almost like a synthetic material. This done we could start the oiling process, which took up the rest of the week. We used the same process as I use on guitar bodies and necks: applying a coat of finishing oil (I used Crimson Guitar's finishing oil, but you can use things like tru-oil etc. to the same effect) twice a day, leaving it at least six hours to dry, and then sanding it back with 2000 grit again to get a polish on it, and repeat until happy.

Alas we ran out of time in the week to do final assembly and sign off on the clock, but we did get the case completed and Jason's taking it back, along with all the other bits we made in Makespace that week (such as the acrylic base and the 3D printed feet) and will finish it off once he's got the right screws etc. I very much look forward to seeing the final object in pride of place in Jason's home!

It was a great little project to work on, I'm very grateful for Jason letting me take part in his project and proud of what we achieved in just two weeks effectively. We went from concept to a high quality, beautiful, and practical piece very quickly thanks to a combination of rapid prototyping and both Jason and I paying attention to detail throughout. I also learned a bunch about working with another kind of material, and I look forward to applying that to my guitars in the future.


When not helping Jason with the clock case, I did managed to do various things guitar related. At the end of last week I was making another revision of the 3D printed controls for the prototype offset. Although happy with the ergonomics, I just couldn’t get the hole inside the control knobs I’d printed to mesh tightly with the knurled shaft on the control pots and rotary pickup selector switch. So I tweaked my control design and printed a new version with a hole for a grub screw in it, and I had hoped that the resin would be soft enough that I could just screw in the grub screw without tapping the hole; I was wrong. When I tried to just screw the grub screw into the hole in the resin it just split the resin near the opening and then wouldn’t go in any further.

So, it was time to skill up and learn how to tap a hole properly.  For those less familiar with this term (as I was not so long ago): tapping a hole is to make the threads for the screw to follow on the inside of said hole. In a lot of cases in woodwork you don’t need to do this, you can just use self tapping screws that will cut the thread into the wood as you screw them in. But on harder materials like metal and (in this case) resin you’ll find you can’t do that, and you need to make the threads using a tap and die set.

Once again I was reminded of one of the main advantages of a communal workshop like Makespace: when I asked Alaric, one of the regular members, about where we keep the right tools he not only pointed me in the right direction, he also explained how to tap a small M3 hole like this reliably. So 5 minutes after starting, I’d taken one of the failed prints, drilled a hole of the correct size (2.8mm for a 3mm screw) tapped it, and fitted it to a pot where it remained absolutely tight. It was very much a Matrix style "I know kung-fu" moment - going from ignorance to perfectly executed in a few minutes thanks to Alaric's guidance!


I then redid the switch on the offset with a tapped hole, fitted it with a grub screw, and it’s wonderfully secure: no more slipping when it reaches the ends of its range and you try to force it further. Very happy that I’ve learned this new skill, and that my 3D printed controls now are just as secure as they are nice to use.


Jason applied the same knowledge to the clock case, tapping the screw holes in the olive wood. Given we knew it was somewhat brittle from how it reacted when we milled it out, we felt gently tapping the holes would be easier on the wood than trying to use self tapping screws.

I’m still a little unhappy with the tuning stability on the prototype offset’s dynamic vibrato. I posted on instagram asking for advice and got a few suggestions from the kind folk there: you need at least 11 gauge strings for it to have enough tension; ensure the strings aren’t sticking in the nut by lubricating it; lubricate the posts on the vibrato; tighten the springs in the vibrato. I’m working my way through this list going from easiest to do through to more destructive (adjusting the springs will not be a nice job given how they’re bent over when fitted). But shortage of time meant I only got as far as fitting some gauge 11 strings, so I hope to revisit this in the coming days.

A lot of people tell me that the Mustang style dynamic vibrato just isn’t that stable, and that may indeed be the case, but if I’m going to sell this guitar to someone I want to know it’s setup as best it can be, so definitely want to keep at this until I know I’ve got it as good as possible. If you have any advice on setting up this style of guitar please do drop me a line!

The recovered offset body got close to finished this week when I completed the shaping that the CNC router didn’t do: I used a hand router to curve over the edges, and then using a rasp I got my monthly exercise allowance by hand carving the comfort carves on the body: the arm carve on the top and the belly carve on the bottom.


It’s a lot of effort to do this by hand, but it’s actually much easier in terms of accuracy than one might think when using such a brutal tool. I use regularly spaced pencil markings to guide me, which I refresh through out the process to ensure I almost always have some visual reference as to what I'm doing, but beyond that it’s just going slowly. This body is now done bar final sanding to remove machine marks and getting it ready to stain and oil. Having got this far, I’m now confident to cut out the delayed commission by hand rather than using the CNC router at all, which will be my first body made without using the CNC router at all. A bit of a backward step, but I’m eager to get these guitars moving and not hanging over me.

A week in the shop(s)

This week was topped and tailed by trying to better understand two CNC routers in an attempt to get back to where I want to be production wise, and the middle bit was learning to cope without a reliable CNC machine.

I started the week in Liverpool. Following on from MakeFest last week, I spent a couple of days visiting DoES Liverpool. DoES is a maker space/co-working space/event space set in Liverpool’s Fabric District, and in part I was visiting as they’ve taken delivery of a new CNC router, and I wanted to investigate if it’d be a viable alternative for me to use whilst the one in Cambridge’s Makespace is giving me grief. 

Thanks to the help of some DoES regulars, notably “Nuclear” Sean and Adrian McEwan, I was able to have a look at their machine and try it out. The machine itself is a CNCStep High-Z router with a 720x420mm bed. Despite being smaller than the one we have in Cambridge, it’s actually rated for metal work which ours is not, which is pretty cool. However, because the machine was new, it was still in the process of being set up; notably they didn’t have a proper sacrificial bed set up, and they had a limited range of collets, which meant I couldn’t use the router bits and drill bits I’d need to make guitar parts there and then. However, I did my best to help document what they had and work out what kind of collets etc. they needed to get things up and running properly, so whilst the trip didn’t lead to any guitars being made, I did do my best to try help move them forward given my time there.

My plan is to go back in a month or so (modulo the madness that will be August travel wise) to both try do some work on the router and continue to try contribute to the DoES community by running my Fusion 360 training course for their members. It will also just nice to be back to hang out with the DoES crowd - they have some great makers up there, so if you're in Liverpool I can recommend going along to some of their events and meeting the crew.

Back in Cambridge we had an unsuccessful attempt to breath confidence back into the CNC Router at Makespace by replacing it’s controller with a new one. We’re fairly convinced that all the errors we’ve seen with it are software bugs in the controller, and we found a roughly equivalent pendant controller for the router similar to the one we have. Whilst there are many wonderful other ways you can control a CNC Router, we had hoped to replace the existing controller with something roughly the same to avoid having to re-train everyone at Makespace.


So a group of us on Saturday tried to swap the controllers, and despite what we though was initial success when we managed to get the new controller and it’s breakout board installed and controlling the motors and responding to switches etc., the moment we started to try loading actual CNC g-code into the new controller it started crashing and losing its settings, which is quite dangerous on a machine like this. Thus we had to roll back to the original controller and replan.

Despite the failure, I learned a lot about how a CNC router is put together this week, so overall I feel better positioned to try help move things forward now.

Back in the workshop I’ve continued to try and get around all this CNC router nonsense by doing more by hand, so I can get these commissions done. The body that I talked about patching last week needed two more stages done to it: it needed to lose some depth, and it needed the holes put into it.

When I thickness the planks to make a guitar, I always leave a couple of mm more than I need for the CNC router to face off. Unfortunately now I’m not using the CNC router to finish the recovered body, I needed to by hand take those 2mm off by hand. Whilst it may not seem like much, this body is made from American Ash, which is quite dense, so anything I can do to keep the weight manageable is worth it. So, it was out with my trusty number 7 hand plane, a bit of time sharpening the blade, and then an hour or so of hard graft in this week’s high temperatures.


The results are great. I don’t use the plane enough to be an expert, so it took me a little while to get back into it, but in the end I had a nice flat back to the guitar and it was down to the size I wanted.

Next up was to use the templates to add the holes to the body for the neck, bridge, and through body string holes. I was a bit wary of doing this, as when I did the first guitar I did all these by hand, and whilst I did an okay job, it wasn’t nearly as good as the CNC router can do with it’s mechanical precision. However, using the MDF templates I made along with a lot of careful alignment on the pillar drill, I was able to get everything lined up and I’m really pleased with the results.


I’m now fairly confident that with the templates I have I can make a regular solid body shape by hand. The only bit that I still need to figure out is the headstock transition on the neck, which I’d just got happy with using the CNC router, as I wasn’t happy with the results of using a drum sander on a pillar drill. 

I’ve taken another pass at trying to tweak my pickup selector switch on the prototype offset. I’m super happy with the ergonomics of the switch, but The 3D printed part just isn’t tough enough to rotate the selector switch without slipping. So, after chatting with Matt at Fidelity for a bit, he suggested I should try adding a grub screw to it. Thus I broke out Fusion and tweaked my design to accommodate that:

Screen Shot 2018-07-10 at 10.36.53.png

I had to add a rear section to the selector to give me enough depth to hide a grub screw, as before there was nothing on the back and it was too shallow. But I think actually the switch looks nicer as a result, making the selector look less like Beaker from the Muppets :). I’ve printed one out, and I’m just waiting for some pointed grub screws to arrive in the post for me to test it.

A week in the shop and in Liverpool

This week was the run-up to Liverpool MakeFest, and so the week was in somewhat dominated by preparation for that. Liverpool MakeFest is a chance for the public to come and see lots of maker projects and hopefully this will inspire them to try their hands at making themselves. As such, as I did last year, I used my stand to try and teach people how an electric guitar is made and de-mystify the process somewhat. To do this I had a table that tells the story left to right: on the left we have wood blanks for the body, neck and fretboard, in the middle we have an in progress build, and on the right we have the finished product. I then use this to guide people through the process so that hopefully they understand how an electric guitar comes to be.


I also had stashed under the table my trusty Blackstar Fly BT workbench amp, so I could let anyone who wants to have a play. Originally I planned to take the newly finished prototype offset, but I realised at the last minute that the Blues Deluxe guitar shown above was actually one of the in progress pieces I had on display last year, and that made for a nicer story for the few repeat visitors we had.

As with the year before, the day of MakeFest was nicely busy. Doors opened at 9am, and whilst the first hour was slow,  I was talking to people all day until doors closed at 5pm without much of a break. I had lots of great conversations with people about guitar building, playing, types of wood, how fretting is done, and I took fun trying to get children to see the guitar in the wood blanks (which I’d laid out in the rough shape), then talking them through the process (comparing routing to tracing on paper etc.) to the end, and then going back and seeing if they could see the guitar in the blanks afterwards, and typically they could. I really was a fun day, and I got to crank up the amp as loud as I could after close and belt out some White Stripes in a public library, just because you don’t often get the opportunity to :)

Many thanks to everyone who stopped by to chat, those who played the guitar, to Laura for helping get me set up, Jason for coffee, and obviously a huge thanks to the organisers and volunteers running Liverpool Makefest - hope to see you all again in 2019!

To get things to show for MakeFest I decided to get the failed body and neck I have hanging around into some presentable state, so they could be examples of in progress work. Last week I talked about how I’d routed out the body that was half cut then damaged by the CNC router using templates and a hand router. This week I patched up that body, and I also went through the same process of finishing the CNC router’s job by hand on the neck it damaged.

With the guitar body being fully cut out the next job was to patch the hole that the CNC router had made in the body as best I could:


A traditional approach to this would be to cut out more material to make it a regular shape and then fill it with a plug made of the same wood in that shape. However, inspiration (or insanity) struck me, and I thought I might be able to plug it without taking any more material out if I took a rubbing of the hole on paper, imported that into Fusion 360, and then CNC cut a plug that matched exactly. This crazy plan actually worked, much to my mild surprise. I’ve written up the process in full over here, but the end result is like this:

IMG_0382 2.jpg

The colour match isn’t perfect, but it’s certainly a perfect fit, so with a dark stain or painted over you’d not know that this was patched, so a great success as far as I’m concerned. I could possibly have got a better colour match had I managed to hold onto a bit of offcut that was closer to where the damage occurred, but I didn’t realise the colour of the wood changed from side to side until too late. Still, this is all important learning. 

I’m very grateful to the encouragement I’ve been given by Ray at Black Swamp Guitars and Matt at Fidelity Guitars on coming to understand that mistakes will happen as they do in any line of work, and just because you can’t hit undo doesn’t mean you give up and start again. Similarly theres just going to be defects in wood that you only uncover part way into a build, so you really do have to be prepared to react to what’s happening as you go along, and not (as I’ve been doing) despairing the moment things aren’t perfect. 

Having sorted out the body ready for show at MakeFest, the next step was to get the neck where I had my first incident of the CNC router going rouge cut out. On this one the neck was almost complete and the CNC router doing the final contour cutout stage when it decided to plunge the bit into the wood as it cut, shattering both the wood in that area and my expensive router bit. Ultimately though this is another time where the plan is to just make the hole more of a regular shape and then plug it. The trick I did for the body won’t work here as the hole really is irregular on all axis, not just a 2D irregularity. 

First step then was to use the bandsaw to cut off the bulk of the wood that wasn’t the neck, a task made easier by the fact that the router had done 3/4 of a contour pass giving me nice guidelines to use. For the rest I just pencilled around the MDF neck template I’d made previously and used that as a guide, keeping 10mm or so clear.


The neck now roughly freed I screwed on the template to the top side that would be hidden by the fretboard, and used the router to follow around the main part of the neck. This didn’t work for the headstock though, as I’d already cut away the face of the headstock, so I then had to flip things around and use clamps to hold the template on the back of the neck and clear the headstock. This done, I could now see the damage more clearly.


Thankfully it actually isn't as bad as it looks: the damage is around the 12th fret area, which means the neck will be curved on the back still at this point, so actually only a tiny amount of damage will need to be patched, not the entire hole. My plan now is to to the initial roughing pass on the back of the neck just to get a 45˚ face on the back of the neck, then patch it. However, I stopped at this stage due to lack of time this week, and the neck in the template cut state was good enough to show people at Makefest.

On the topic of necks, given my new found distrust of CNC Routers, I wondered how I’d do the truss rod channel without a CNC router; it’s easy to do the outside lines of the neck as I’d just done, but getting the truss rod channel is much harder. For this I’d need a jig, so I decided to make one.


For this jig I’d need basically two parts to it: the first part is a bit at the bottom to hold the neck snug and in a known position, and then a second part above that where I can insert my hand router and run it along with stops on all sides to ensure the only bits I cut are those needed for the channel. I decided to knock this design up in Fusion 360 and make it out of MDF. Note I don’t need it for the above neck, it already has a truss rod channel, but I just wanted to understand what was required to help me become more confident that I can make guitars without the crutch of a CNC router that will never fail.

You can see the jig above. It’s designed to be made out of 3 bits of 12mm MDF. The bottom layer has a pocket the exact shape for the necks I’m currently making, which will hold the newly templated neck in place. The second layer is a spacer, and provides a bed for the hand router to site on. The top layer is sized so that the based of the hand router can’t move side to side, ensuring the channel is straight down the middle, and it can’t move to far at either end, so the channel is just the right length. Having designed it, I then cut it out in MDF and I have it ready for action.

Screen Shot 2018-07-02 at 10.22.20.png

I’ve always been unreasonable scared of making jigs; it seemed something quite complicated to get right, so building this was quite confidence building. Thanks to my neck design being in Fusion already, generating the jig design that exactly matched too less than half an hour, and cutting it out on the CNC router didn’t take that long either (again, although the CNC Router will occasionally fail and damage bits, I’m still happy to cut MDF on it as the cost of failure is low: I just don’t trust it to cut guitar parts as the cost of failure there is high).

The other fun project this week was working with my friend Jason, who’s visiting over from the US. Jason’s recently been working on some amazing electronics projects, including his Anytime Nixie tube based clock that shows the time of where his friends are in the world. Whilst Jason was in Cambridge we thought it’d be fun to work on something together, and so I’m helping him make a wooden case for the clock in Makespace.


The clock currently looks like the above picture: a set of boards with the 4 nixie tubes on top, and it’s powered by a USB micro port on the back. Obviously I want to design and build the base using Fusion 360, as that’s the tool I’m familiar with (as is Jason). Thankfully Jason did his PCB layouts using Autodesk Eagle, which will export the PCBs into Fusion 360 with the correct board sizes and surface silkscreens, making it really easy to get started with. I just had to model the headers connecting them and the nixie tube holders, and I’d very quickly got a mock up of Jason’s latest physical version of the clock.

Screen Shot 2018-07-02 at 22.18.46.png

From here I then did a quick rough model of the case before Jason arrived, and then together we refined it and did a series of test 3D prints until we got a design we were happy with. It’s really quite powerful having access to a 3D printer for this kind of thing. Although I want to machine the final case in wood, that’s a slow process that will need constant watching, so quite expensive for throw away prototyping. But a 3D printer will make a case design in a few hours and I don’t need to baby sit it for that time. So we were able to go from this:

Screen Shot 2018-07-02 at 22.20.48.png

To this:


with next to no effort. Similarly we wanted an acrylic base sheet, and designing that and getting a test cut using the laser cutters in the workshop took literally ten minutes. We’re truly living the rapid prototyping dream :)

Having got his far, the next step once back from Liverpool will be to mill out a wooden model in scrap wood, before finalising the design and making it out of some nice textured wood so that Jason has a lovely looking clock he lave running in his home without it looking like a lab experiment. 

In last week’s notes I celebrated finishing the prototype offset build, this week I sat down to try do the setup, and immediately learned why people have a love/hate relationship with vibrato units (or as they’re more commonly and incorrectly known, tremolo units) on guitars: the sound is nice, but getting them to behave is a real pain. At first I just assumed either I was an idiot or it was broken, but eventually (buried in some youtube comments somewhere) I found some advice about making the Mustang style dynamic vibrato bridge work.

The general advice is ensure the springs are as far down the posts under the bridge as possible, and that the cigar tube bit the strings are into is as low as possible too, but not with the strings touching, you want space for them to move a little. Having done this you then just tune up, use the vibrato (not gently either, but also not dive bombing it), at which point it’ll go out of tune, and then you tune again and repeat until things are roughly stable. Even at this point, I still find it goes out of tune more than I’m used to, but at least I can get through a song using the vibrato without being out of key.


How to plug a mostly-irregular shaped hole in wood using Fusion 360

Regular readers will know I’ve had a bit of bad luck with the CNC router this last year, the latest of which was where the router unexpectedly decided to cut into the body of a guitar when it was supposed to be cutting the contours, ruining the body like so:


Whilst this body is now not fit for its original purpose any more, as that guitar will have a natural finish, I’m keep to try patch this error and make something from this body given the time and expense that has already gone into it. Having finished the CNC routers job of cutting out the rest of the body by hand using a template and a hand router, I wanted to patch up the hole using a bit of left over wood I had from when I cut the body out.

The typical approach here would be to cut a more regular shape out of the body and then plug that, so (as suggested by Ray over at Black Swamp Guitars) I could have used a wide spade bit to drill out a circular area and plan plugged that with a circular plug. Basically you’re trying to make the problem simple with this cut then patch approach, as patching irregular shapes is very difficult. However, for fun, I wondered if I could somehow model the area cut out here, as whilst it’s not as simple a shape as a circle, it is a series of clean curves made with a router bit of a known size, and it’s also nicely uniform across its depth due to how it was made: thus if I could get the top 2D profile then I could just extrude that to make the plug.

To make the plug, I first did something I’ve not done since I was a child: I got some paper and a pencil and took a rubbing of the top of the guitar in the area where it went wrong.


Having done this, I then scanned it and imported it as a canvas into fusion:

Screen Shot 2018-07-01 at 16.18.36.png

When you import a canvas Fusion doesn’t know how big the image is (it appears to ignore the DPI information in the image header alas), so the next thing I did was make a couple of reference construction circles in the sketch with the canvas, the diameter of which was the diameter of the bit the router was using when it went rogue (in this case 1/4”). Having got those two circles I positioned one so that it was roughly where the inner sweep was as the router cut in, and one in it’s final resting place, and then I could use that to start resizing the image canvas until I was happy I had the image to scale. Once that was done I used the Fusion sketch tools to draw the outline of the damaged area, with a little extra sticking out the side:

Screen Shot 2018-07-01 at 16.00.02.png

This done I could then extrude the entire shape, and now I have a design for a plug that in theory should perfectly match.

Screen Shot 2018-07-01 at 16.01.16.png

Having got my design, I then made some CNC router toolpaths, and I did an initial test run using a bit of scrap ply I found in the workshop. The result wasn’t bad, a lot better than I expected in fact give I was just basing this off a rubbing I'd made!


The spacing wasn’t perfect though, so I went back to my design and added an offset outline of a couple of tenths of a millimetre, and the next plywood test plug I cut, with a little sanding on the side, fit snug. Satisfied with that, I then went to my ash offcuts and cut a taller test plug, which still worked, and finally I did a 4th and final plug where I matched the grain of the wood as best I could to the surrounding area where it was going to sit. Again, after a little sanding, it fit perfect:


Obviously when I say fits perfect you can see I have overhangs on the top and side, but that’s deliberate: it’s much easier to level those bits with a hand router and sanding than it would be to cut it perfect at this stage. Having got a plug that matched as best I could, I glued it in, left it clamped over night, and the next day trimmed the excess off, the result of which is a nicely plugged hole:

IMG_0382 2.jpg

Whilst the end result is still noticeable in terms of colour at this stage, if I paint this body or even use a dark stain then you now won’t be aware that any damage was done and it had to be patched. I’m looking forward now to finishing this body off and turning it into a guitar!