onsdag 31. desember 2014

Xonik VCO

I have spent a lot of time lately understanding the exponential converter and temperature correction built into it. I have even written a long document about it that I will publish as soon as I find a place to put the files.

Now, after understanding most of it, I have built a tiny VCO. The core is the well known Terry Michaels core from the electronotes articles, the same as the ASM-1, Ken Stone Catgirl synth and the PicoVCO (which in turn is presumably the same circuit as the Hearn Morley HMVCO1c). The waveshaping is a somewhat modified version of the picoVCO for the triangle part combined with the pulse circuit from the Yusynth VCO. All wave outputs are +/-5V.

The PCB layout is heavily inspired by the PicoVCO v2.0, but I have kept the linear FM input and moved the triangle symmetry potmeter off board so that all pots are on the motherboard. The symmerty circuit is also somewhat changed. Physically it is about 21 x 33 mm and it has 13 pins, so it's slightly larger than the PicoVCO.

lørdag 23. august 2014

The new super-mini-dual-VCA

I had a little time to tinker with Eagle again yesterday, and the result is a 40% smaller dual VCA. I've now put parts on both sides. I had to remove the separate CV input and trimmer outputs - instead of the trimmers you can put four 510 Ohm resistors on the board. The physical size (excluding pins) is now 27.3 * 16.8 mm.

I must admin that I like the original a little better as it had labeled pins and a nice big Xonik logo on it, but this one is very cool for being so small.

torsdag 21. august 2014

VCA works as expected

I tested the dual VCA today. Except for a mixup of output pin 1 and 2, it works as expected. At 10V the amplification is just slightly higher than 1. The VCA is linear, so at 5V it looks roughly like 50% amplification

VCA hooked up to the bread board. The two 1k trim pots are for adjusting the symmetry around 0V. 

Response at CV=10V. The outermost triangle wave is the output, and it's slightly larger than the input.

Response at CV=5V, the output is about half the size of the input.

torsdag 14. august 2014

7 saw controller board populated

I finished populating the 7 saw controller board tonight. It has not been tested yet but it looks good :-D

I didn't have any 22k 0204-size resistors, and I can't find any 3M3 or 4M7 resistors anywhere in that form factor, so I used normal 1% metal film ones instead. It looks a bit wonky but will work fine.

I have yet to test both the dual VCA and this board so not sure it works yet...

7 saw controller board

tiny LM13700 based dual VCA

Silk screen looks great :-)

lørdag 3. mai 2014

Working on analog multiplexing

I am trying to do analog multiplexing. It's a bit harder than key/digital multiplexing, and I have made this initial design. Let's see what the nice people at electro-music think of it.

lørdag 22. mars 2014

First multi-sub-oscillator board populated

The first 4 x sub oscillator board has been populated. This is the top board, a second pcb with four more sub oscillators will be mounted below it. This time I've made sure connections from the lower board actually go through the top board, so no cables have to be attached to the lower board (as opposed to the oscillator boards where you have to connect cables to both boards.

This is the rev 1 board which has the same bug as the single sub oscillator, so I have soldered 100nF caps to the flip flop feedback lines to make them flip and flop like it should.

Quad sub oscillator
Close up, note the tiny 100nF caps soldered directly onto the smd components' legs.

New panels

This week I've finished five new panels and prepared prints for three more. I've done two more oscillator panels, two sub oscillators and one 7-saw controller. I've also printed two more sub oscillators and one controller. Since the edges of the first oscillator panel didn't get as sharp as I wanted when folding the plastic around the edges, I've decided to cut the foil along the edges this time.
Plexi glass panels before attaching the adhesive film

Three oscillator modules - the first one to the left has plastic folded around the edges while the new ones are square cut for a sharper edge. I've also realized that I will run out of orange potmeter caps. Unfortunately, Re'an has stopped producing these so it's hard to get more. I've found some similar looking ones on Ebay, they seem to be 1mm wider but it may work out ok.
Finished sum oscillator and 7-saw control panels

Close-up of the sub oscillator. Note the input polarity switch, setting this to the opposite of the input wave will give that cool saw-within-saw wave shown in an earlier post

tirsdag 11. mars 2014

Tiny but crucial bug fix on the suboscillator board

A couple of weeks back I populated the first single suboscillator board, but much to my dismay it didn't work. I didn't have time to inspect it further, but I suspected it was related to the CD4013 flip flop as I had some trouble getting it to work originally. I ordered some more from a different source and they arrived yesterday.

I swapped the old chip for the new one, but it still didn't work. I then rebuilt the first part of the circuit on a breadboard, and it sort-of worked, but both flip flops in the package seemed to flip and flop at the same frequency, which they definitely shouldn't.

Today, I tried adding in a 2k2 resistor between the D and not-Q ports, as well as a 100nF cap from D to ground, a trick I found somewhere online when I originally designed the circuit. This made the circuit work again.

Doing this on the PCB would require me to cut some traces, which would be a lot of trouble, especially on the SMD-based quad suboscillator.

Luckily, when i replaced the 2k2 resistor with a wire, while leaving the cap in place, it still worked :) I soldered two caps onto the pcb as well, and it too worked like it should. I have not tried the board at various frequencies, but for the time being it seems to work fine.

While searching for solutions to my problem, I came across this post named "cd4013 weird shit" - that really sums up my thoughts about this whole problem too...

The fully populated sub oscillator board. It looks a bit like a bug of sorts, don't you think?

A closeup of the 100nF capacitor. Also note the beautiful curved cutouts, I just love the boards OSHPark are able to produce!

tirsdag 18. februar 2014

Final (?) Saw mix and detune circuit tested

This week I've spent at least 20 hours recalculating and building the saw mix and detune circuits on a solderless breadboard. I changed the mix and detune inputs to accept 0 to 10V instead of 0 to 5V, and also added an offset input to be able to use -5 to 5V inputs. The detune output still ranges from 0 to 5V, but the mixer slopes are now 0 to 10V as well, as this is what the VCAs I use expects.

I did a lot of up front calculations, but the final circuits were adjusted by ear. I am pretty pleased with the result, enough to go ahead and redraw the PCBs. I have decided to place the two VCAs on a separate, tiny PCB, which can be reused for other projects. I will also try to fit all parts on a single PCB instead of the two that the initial design uses, but this probably requires me to use SMD components again - the prototype uses almost 30 op amps and more than 50 resistors and takes up one and a half proto board :-)

The result is quite good, although not as good as I initially hoped and dreamed about. I am still impressed with the result though, especially at lower frequencies - it will make one hell of a bass sound. I may try to make a "computune"-style automatic tuner, to keep the oscillators better synced. I will also add a chorus, filter and perhaps distortion module, but for now I will focus on making the controller PCB.

The prototype fills two breadboards and includes all the necessary electronics for generating separate detune slopes for each oscillator, mixers to mix the single input pitch CV with the detune CV, mix CV slope generators, two VCAs and two audio mixers. To the right, behind the breadboard, is the 8 oscillator module.

Before protoboarding this circuit, i finished the 8 oscillator module, complete with separate fine tune pots and 8 CV in / 8 saw out connectors. It looks very cool!.

I shot a few videos for a friend of mine - the quality is crap but it illustrates how things work:

søndag 5. januar 2014

Quad saw mixing

Before new years I tried hooking up a simple mixer to mix four saw waves. The mixer consisted of half of a TL072 and five 47k resisors.

While I DID get a mixed output, it sounded worse for every new oscillator I connected. Today I hooked up an oscilloscope to try to see if I could figure out what was going on, and in fact I quickly realized the problem.

When mixing two +/- 5V signals with unity gain, they will occasionally reach +/10V. Mixing three oscillators will result in +/-15V, and four results in +/-20V. The problem is of course that above 14.8V (or thereabouts), the op amp starts clipping, chopping of the tops. This does not sound very nice at all.

By reducing the gain resistor to 12k, the sum of the four oscillators will be within the original +/-5V. And indeed, the output sounds so much better! Now I'm closing in on the buzzing, swarming multi-saw sound I am looking for :-)

I have to rethink the oscillator mixer for the 7-saw though, to prevent clipping at all times. Next up is breadboarding the mixer again :-D

onsdag 1. januar 2014

How to make cheap front panels

Due to our recent car purchase, I expect money to be a bit tighter than usual in the next months. I still need some cool front panels for my upcoming polyphonic/super saw synth. I have therefore been looking for ways to create decent panels at a low price instead of using Schaeffer/front panel express or similar services.

Today I completed the first prototype, a panel for the 8-saw oscillator, and this post explains the process in detail.

I have the luxury of co-owning a Silhouette Cameo cutter with my wife. This machine, while meant for scrapbooking and cutting vinyl for arts & crafts, works perfectly for cutting all sorts of soft materials. While you can get away with just an x-acto knife or similar, the machine makes the job so much easier and the end result much better.


For the panels, you need the following:
- 2 mm thick perspex/plexiglass or similar hard plastic glass
- Adhesive-backed glossy photo paper
- Adhesive-backed transparent plastic

I got my perspex from Biltema here in Norway, but you can get similar stuff elsewhere. Make sure it is 2 mm, any more and you may not be able to attach the mini jack sockets, any less and the panel will bend too easily.

The photo paper I used is called Skyhorse Premium Quality Glossy Photo Paper, 130g, self-adhesive cast-coated. I got it off ebay in the UK. It is rather thick, and the ink I use (Canon CLI-55xx) is not absorbed well enough to protect it from wear and tear - it smudges easily even after drying for some time. This is however no problem as long as the paper is covered with the adhesive transparent plastic

The adhesive plastic I use is the kind used as a dust cover on books. It comes in glossy and matte finish, I've chosen the glossy one.

The process


First I designed the panel in Illustrator. When doing this, I added cross hairs to all circular components, to tell me where to drill holes. When satisfied with the design, I made a copy of it and removed all but the outline and cross hairs. I then printed this design onto a piece of the adhesive photo paper, trimmed the edges using a paper cutter and attached it to the perspex. To make it easy to remove after drilling, I put the paper on top of the protective film that covered the perspex.

I then drilled all holes using a 1 mm drill bit in a drill press. It's much easier to get an accurately positioned hole using such a small bit first. Remember to also drill holes for any key tabs on the potentiometers.

Some of my holes are very close (1 mm) to the edge of the perspex. I drilled these first, in case they crack the perspex, so that as little work as possible is lost. Be very gentle and do not put a lot of pressure on the drill.

I then drilled 3 mm holes, even in those places where the end hole is supposed to be larger than that - still to get a better accuracy.

Then I used a conical drill bit to drill any holes larger than 10 mm. Finally, I drilled any remaining 6 and 7 mm holes.

After drilling all the holes I used a scroll saw to cut the edges and straightened them by sliding the panel along a metal file. This is the end result:

Cut and drilled perspex
Note that I tried to drill the holes for the potentiometer key tab only partially through the perspex. This was a bad decision as the plastic extrudes slightly, making the hole visible on the front even after attaching the photo paper. Had I drilled it all the way through, it would probably have been invisible.

Photo paper

After creating the raw perspex panel, I printed the panel design onto the adhesive photo paper.

The Silhouette cameo comes with a plugin for Illustrator that lets you add what is called registration marks to your drawing. Later, when you want to cut out parts of the design (holes for pots etc), the Cameo recognizes these registration marks and knows where the design is located on the paper. The plugin costs $40 but is well worth the money.

Panel design with registration marks in the corners
Next up is to cover the panel with the transparent plastic to increase its durability. I decided I wanted the edges to be well protected as well, so I came up with a way that let me put the plastic around the edges and onto the back. If you do not need this, just put plastic all over the panel and cut along the edges of the front panel (after creating holes for the pots etc).

In my case, I had to cut away parts of the photo paper around the edges of the panel so that the plastic film would not stick there. As the photo paper already has a non-stick backing paper, you only need to remove the photo paper itself. However, it is hard to cut through the top layer only, so I did instead cut all the way through:
Cut two grooves about 10 mm apart.
After cutting two grooves, I turned the paper over and put some Scotch magic tape over it to keep it in place. I then did the next side the same way etc:

Finally, I removed the photo paper from the parts I had cut out:

I could now cover the whole paper with plastic film:

 After covering the paper, I attached it to the cutting mat of the Silhouette Cameo:

Based on the same Illustrator design, I created outlines for the potmeter holes etc in the same position as the previously created cross hairs. I made sure to make the holes about 1 mm larger that the holes I drilled, as the silhouette is not accurate enough to get a perfect hit. The design is then sent to the machine and "printed" onto the panel. Due to the thickness of the plastic and paper combined, I had to set the cutting knife to 8, and even this proved to be a bit too little in some cases. Ideally you want to cut all the way through the photo paper but not through the backing paper.

I now carefully removed all the cut out parts using a sharp hook

Then I cut off the edges using a paper cutter. The corners are cut diagonally to be able to wrap them around the edges

Finally, I removed the panel from the backing paper and attached it to the perspex

After attaching the front panel components (knobs, jacks etc), all edges are covered and the result is quite pleasing. The plastic is slipping slightly from the photo paper close to the edges, this will probably not happen if you do not fold the plastic around to the back. Also, when tightening the DIN-jack sockets the plastic bulges slightly, this can possibly be prevented by cutting the plastic closer to the screw holes.

All in all I think the result is very good, especially considering the price and the time it took to create it.