More Guitars!

Just a quick note that there’s now three guitars on my workbench. Shortly after Julia I also got contacted by Jörg Hiller again, as we never fully got the robo-tuners on his Fender working.

If you look closely you can see the modified Tronical system on the back of the headstock.

At that rate I might be able to switch to full-time guitar projects soon!!

Fretlets of a different kind

Who would have thought that fretlets would become a “theme” so fast. While still crazy busy installing the fret scanning system in my new guitar (see previous post), I got approached by local guitar player Julia Reidy about the possibility of modding a guitar for unusual tunings, e.g. just intonation or Asian tuning systems like slendro.

This is not a new idea of course – there’s many examples of such guitars and one can even buy necks with certain tunings. The challenge though was to build this in a way that would allow for experimentation, i.e. with flexible fret positions.

After some research we came across the Youtube channel of Tolgahan Çoğulu and decided to try something similar. The way he approached the problem was to cut long channels into the fretboard and then put fretlets on small pieces of wood that can slide inside these channels.

Our first experiments were done on an old beater guitar of Julia’s. I decided to use little pieces of brass instead of wood to put the fretlets on:

The trick is to get the size of the channel just right, so the friction keeps the brass piece in, but allows for it to be moved around.

Here’s the neck with 10 pieces:

After this proof of concept, we moved on to the actual guitar. I had an old fretless guitar that I had stopped playing decades ago, so it was the perfect instrument to bring back to life for this.

Of course, cutting the channels with the required precision (<0.1mm) was another job for the CNC. Sadly the longest cut I can make with my current machine is 400mm, so I couldn’t do the whole length of the neck in one pass…

…but with some extra effort I eventually got it done:

And here’s a picture by Julia with the majority of the fretlets installed:

Looking forward to the music!


As mentioned at the end of my last post it was now time to finally get some frets into the fretboard. While usually not such a big deal, the plan was to put a fret-scanner in, which made this task “slightly” more complicated.

There’s two main ways to accomplish fret-scanning. The first one views the strings and the frets as a 2D-matrix, where you send time-multiplexed pulses down the strings and pick them up at the frets. The advantage of this is, that you have to do less wiring (one wire per fret), but you do need to isolate the strings from ground, which I found a bit problematic to achieve.

The second approach is to divide the frets into 6 isolated segments. This means a lot of mechanical work and a lot of wiring, but the detection circuit is much simpler. Each MCU input is programmed with an internal pull-up resistor, and then you just have to watch for the voltage to drop to zero as the string touches the fret.

Here’s the first few fretlets, as I decided to call these fret segments:

Each has a wire soldered to it that pokes out at the back of the neck

Bringing all these wires (6 x 19 = 114!) to a centralized location wouldn’t work of course, so instead I installed little MCUs right into the back of the neck, one per fret. This, again, was cut with the CNC, using a very small endmill:

There was a limit to how far I could the CNC though. Taking the curvature of the neck into account would have required a precise scan, and preferably also an additional axis to tilt either the guitar or the toolhead. I wasn’t quite willing to run an extra loop of figuring all this out, so I decided to carve the channels from the MCU to the fretlets with a manual tool.

And here with a chip wired up:

Needless to say that all this carving took a long time…

…and got complicated at the neck-body joint:

Also had to install some spring-loaded pin-headers to connect to the body electronics.

The individual MCUs are all connected by a daisy-chained SPI bus. The overall update frequency is about 1kHz.

The chips and wire-channels will eventually be covered with clear shellac, but I’ll wait with this until I’m absolutely sure that I won’t have to re-program the MCUs again. So far the detection seems to work quite well though

New Guitar Case Amp

Ever since I started making music again in ~2013, my main guitar amplifier has been the very case that the guitar travels in. I use this for all solo performances, as well as my duos with Werner Cee, Chris Abrahams, and most notably IMD with Axel Dörner.

The idea to use transducers to turn its shell into a speaker had intrigued me for a long time, however previous attempts had failed simply because the first cases I tried this with were made of ABS plastic which kinda sounds like a wet sack. So only after I came across a guitar case made of epoxy did this whole idea come to fruition.

I put in a total of 6 transducers to go with my 6-channel per-string processing that I’ve been employing for all my setups during the last 30 years. Their different locations also means various filtering effects as sound is panned across the outputs. I do compensate for this in software to a certain extend, however for the most part I embraced this natural behaviour.

One issue you run into when setting the whole shell into vibrations is that you have to make sure to dampen any mechanical buzz that can emanate from basically anything that’s connected to it, most notably the latches. So in the end I removed them all, but then kinda struggled to find a replacement to keep things shut tight.

A rubber band:

Sail-boat like posts

And this is what I use today – an ordinary suitcase belt…

Here’s some more pictures of the inside:

8-channel amplifier (50W Class-D)

6-in, 8-out sound card (XMOS dev. board)

Despite this mess, the guitar still fits in and can be safely transported.

In fact I installed a pannier handle and can lug my whole setup with my bike (at least to local gigs)

So why “new guitar case amp”? Well, lately the original build has been cracking at the seams – I had to replace broken transducers, fix cabling, and various mechanical issues. I also wanted to upgrade to a beefier power supply and type of transducer as well as a better soundcard. Mostly though, the whole effect processing should be embedded as originally intended, rather than always connecting a laptop externally. Moreover, as the new guitar will have a digital output, this needs to be reflected by the overall system design. So, as often, too many things to modify at once…

Sometimes then, it’s helpful when outside events force your hand a bit and you gotta move forward with a partial build. In this case that’s a concert I’m playing next Wednesday (Aug 26th) with Axel at Au Topsi Pohl, where I thought it would be a good idea to have a new blog post along with the newsletter invitation. If you read this in time and wanna come, make sure to rsvp on their website (because, you know, COVID-19 and limited seats and all)

Anyway, here’s some pictures of the current state:

New transducers

New soundcard

Cleaner wiring

But no embedding yet and no digital guitar link either. Stay tuned for updates on that!

Ratschen Reloaded

This project dates back to 1998, right after I had relocated to Berlin, when Jens Brand approached me about building an interface to control a series of geared motors. These motors were then connected to the axes of a number of ratchets, the general concept being to create very loud, computer-controlled sound, but without employing a speaker system.

© Jens Brand

Ratchets are used in Germany during the carnival season and even a single one can be brutally loud, even outdoors from a distance. So imagine 8 of these in a small room! If I remember correctly he once did a duo concert with a guitar player using a big Marshall stack, and that the latter was not audible anymore when the ratchets where running at full speed.

Back in 1998 there wasn’t a strong maker scene and no distributors like Adafruit or Sparkfun to provide all the control and driver modules that we can choose from now . So the interface was build from scratch, using a Microchip PIC16F84 and discrete MOSFET drivers. The controller didn’t even have sufficient PWM outputs, so it was programmed in Assembler with what I called “synchronous code”. What this meant was that I took note of the execution times of each block of code and added extra NOPs to match them to the PWM frequency. All while polling the serial input for MIDI messages.

Sadly there’s no picture of this initial build and in fact the reason why Jens approached me about making a new version was that the original device had been stolen :(

My go-to microcontroller in 2020 is the Teensy which is build around an ARM processor and programmed thru the Arduino IDE. I like its combination of small form factor, sufficient pin count, and processing power. There’s even a cool Audio Library.

To drive the motors I used some Pololu driver modules, and then all I had to do was to mount all these modules on perfboard and connect them with wires. Not a single discrete component.

The ratchets themselves were also worn down from decades of use, so they needed to be rebuild as well, which was done by our joint friend Paper Blattmacher.

Sadly, the premiere at Phill Niblock’s Experimental Intermedia had to be cancelled (*) due to COVID-19, but I hope we’ll soon have a chance to hear this instrument again.

* actually it was streamed, but that doesn’t convey the physical experience at all

Lots of sanding and polishing

Not much to write about the last 6 weeks – it’s been mostly sanding and polishing a lot. But wanted to share some pictures as it always feels quite rewarding when you’re done with the finest sandpaper and apply some wax.

The bottom and the sides still need to be done, though..

Next I plan to do a 3D scan of the top, so I can further deepen the back pockets without accidentally breaking thru the front. Also need to get started on the frets!

Welcome 2020

I’m making yet another attempt at trying to bring this website back to life. Will start by posting some pictures of the new guitar I’ve been building since early 2018.

I’ve set the publication dates to when I actually did the work, so if you prefer to read this chronologically you might wanna start down at the bottom of this page, or use the individual post links on the left.

More shaping

In 2019 I didn’t make much progress with the guitar at all, mostly because my day job kept me busy for much longer than planned. On top of that I moved both our home and my shop (into one location actually).

So it took until April 2020 that I sprung the CNC machine back to life and continued giving the guitar further shape. Here you can see it working on the neck-body joint:

To properly cut the neck contour I had to go beyond the 2.5D approach and go full 3D. In a move to use more open-source software I opted for FreeCAD (also switch to kicad for PCB design recently).

3D tools have a much steeper learning curve of course, so it took me a while to come up with this first design:

Here’s that section after some manual sanding:

While I started applying this 3D approach to the body as well, I realized that it was way too much work just to give all the edges a 5mm radius, so from there on out I went back to good old files and sand paper.

Here’s how that turned out for a section of the back:

I also applied the manual approach to shape the head-stock. Here’s a picture of its back:

With strings!

body shape

As you can see from my previous posts, the CNC machine was used for much more than originally intended. So as a next step it came naturally to also apply it to fine-tuning the body shape.

As CamBam isn’t really the best tool to draw Bezier curves, I resorted to good old Adobe Illustrator to draw the outline:

Although not trivial, this could then be exported as DXF and then imported into CamBam to base the toolpath on.

I had to order an extra long end mill tool to cut all the way down from 40mm height to ground (took a while too…)

This slope actually done manually:

The next design decision was not what I had planned (although one guitar maker had warned me when I bought the ziricote). The problem was that this slab of wood was darn heavy:

More than 50% above a standard Strat body:

So as I needed a lot of room for electronics anyway, I decided to cut pockets everywhere, turning this axe into a semi-acoustic (or maybe quarter-)

Much better now :)


As you might have guessed from the previous posts and pictures, the amount of wood I left for the headstock would not allow for the traditional placements of the tuners. In other words, it’s a so-called head-less design (technically there’s actually still a head – it’s just much shorter)

So this approach calls for two components:
First, you need tuning machines down at the bridge. This is relatively easy to source – in this case I went for ABM bridges with Graphtec pickups. Thanks to Peter Borowski from ABM, Berlin for helping with this.

The second part are the clamps to hold down the strings at the headstock. While there are purchase options for these too, most do not come as single pieces but as a block for all 6 strings. Which, again, doesn’t work with the wide string spacing I prefer.

So I decided to come up with my own design, where the plan was to work as much as possible with the wood itself, rather than attaching a full-metal solution.

Of course metal is needed in some parts, otherwise the wood would wear down over the years. So I designed an inlay where the string runs over a tiny metal bridge, with slots at each side for the metal clamp.

I milled this inlay out of a piece of brass that I had still lying around. This was the first time I used the CNC for cutting metal, so I was a bit nervous if it could do the job, but things turned out nicely without breaking anything:

And the pocket to match:

This was then extended with deeper holes for the clamps:

And holes from the back for screws to hold them down:

Tada, first string clamped down:

30 years of custom technology for musicians and artists