Building a Triangle-Core VCO in ACE

[utdgrant]

I recently watched a video entitled "This Circuitry Is AMAZING - 1970s Music Synth Restoration". This was on YouTube channel THIS MUSEUM IS (NOT) OBSOLETE, which is a spin-off of LOOK MUM, NO COMPUTER.

The synth being restored was in fact an ETI 4600, constructed at home from a Maplin Electronics kit (presumably in the UK). The majority of the video concentrated on bringing all four VCOs back to life. The presenter (Mitch from Hack Modular, rather than Sam from LMNC) described in detail how a triangle-core VCO operated in order to explain the diagnostics he was running against the faulty VCOs. I found the concept fascinating, as I'd always assumed that analogue VCOs were based on a sawtooth-core.

Anyhow, it got me thinking about the principles of operation behind a triangle-core VCO, and I decided to try and build one using the modules available in the Audio Computing Engine bundle (or ACE).

The first circuit was capturing the basics of a triangle oscillator, using one ACE Integrator and an ACE S-R Latch.

SimpleTriangleOscillator.jpg (220.15 KiB) Viewed 14707 times

The +ve and -ve Overflow outputs switch the S-R Latch to the 'Reset' state and 'Set' state respectively. This means that when the Integrator's Vout reaches +5V, the S-R Latch will enable its "Inv Out", sending -5V to the Integrator's Vin. Similarly, when Vout on the Integrator reaches -5V, it will enable the S-R Latch "Output", which sends +5V to the Integrator's Vin. This sequence of events repeats itself at a constant frequency, giving you a stable triangle-wave output voltage.

YouTube Demo of the above patch.

[utdgrant]

The next stage of my experimentation was to add a sawtooth waveform output and provide a 1V/oct frequency response.

The sawtooth wave uses a second ACE Integrator. It has a constant Vin of -2.5V, giving it a falling slope at half the rate of each triangle slope. The Vout is reinitialised to +5V on the peak or trough of the triangle wave, depending on which Overflow pulse you connect to the Initialise trigger input. I added a Monkey Business AB Switch to let the user easily switch between them.

1V/oct frequency control is provided by feeding the keyboard Pitch CV through ACE Powers (y1 = 2^x1), then using the output to control the 'Slope CV' input on both Integrators.

1VperOctaveAndSawtoothOutput.jpg (359.93 KiB) Viewed 14706 times

YouTube demo of the circuit.

[utdgrant]

The final circuit provides all four 'standard analogue' waveforms: Triangle, Sawtooth, Pulse and Sine.

First, I used an ACE Comparator module to derive a pulse wave from either the triangle output or the sawtooth output. I found that using the triangle wave as the source led to a more pleasing, more 'musical' effect than when using the sawtooth. I placed another Monkey Business AB Switch in the circuit to allow the user to select between either source. When you apply PWM, there is a subtle, but detectable sonic difference between the two comparator sources. Even on a fixed pulse width, there are different phase responses when you mix the pulse wave with the other waveform outputs.

The sine wave output is derived from the triangle wave by feeding it through a Blue Velvet Soft Clip Mixer. This gives a close approximation of the diode ladder wave shaper within the 2400 VCO. Because the 'distorted' output of Blue Velvet is close to a pure sine wave, I didn't bother switching on the oversampling option.

TriangleCoreWithSawPulseAndSine.jpg (397.33 KiB) Viewed 14706 times

YouTube demo of the circuit.

[utdgrant]

Here are the three patches shown above, if you fancy experimenting for yourself.

Basic Triangle Oscillator from first principles:

Triangle-Core VCO Basic.voltagepreset

(10.82 KiB) Downloaded 328 times

Sawtooth Wave and 1V/Oct CV:

Triangle-Core VCO 1VOct and Sawtooth Output.voltagepreset

(14.23 KiB) Downloaded 306 times

Full-Fat Triangle-Core VCO with Triangle, Sawtooth, Pulse and Sine wave outputs:

Triangle-Core VCO.voltagepreset

(30.82 KiB) Downloaded 304 times

[utdgrant]

I should mention one caveat - the pitch tracking gets wildly inaccurate once you go above about C4. This is due to two factors:

1. The Overflow logic signals from the main triangle-core Integrator take an extra ample period to reach the S-R Latch inputs, a further sample period for the S-R Latch outputs to activate, then a further sample period where those outputs are multiplied by the Constants and Multipliers module. This means it takes 3 sample periods to reverse the direction of the waveform, and this reversal happens twice during one oscillation cycle. This adds an extra 125 microseconds to each period of the oscillator. It doesn't matter too much for lower pitches, but starts to become very significant as the pitch rises, causing a progressive lengthening of the period and flattening of the target pitch. There's not much you can do to mitigate this problem, as signals which flow between modules within VM will always 'cost' one sample period.

2. When calculating logic within one sample period, it's difficult to avoid the period of an oscillator falliing into integer multiples of 1/48,000th of a second. As a consequence, the higher the required frequency of the oscillator, the less pitch resolution you have to work with.

When you use the Big RAT module as a VCO, the first factor is mitigated. This is because the switching logic / waveform reversal is performed internally, so you don't have to communicate anything between modules. Unfortunately, the second factor is still present, and you start to hear Big RAT 'locking in' to specific oscillation periods which are integer multiples of 1/48,000 sec.

However, I enjoyed creating this proof-of-concept oscillator from first principles and hope you get some enjoyment and insight from it, too.

[ColinP]

Thanks Grant. That's taken me right back to when I spent all my income from doing a paper round and working Saturdays in an electronics store called TeleServicentre in Leeds on magazines like Electronics Today International, Practical Electronics, Everyday Electronics and Wireless World and the synthesizer projects they featured.

Maplin were a great company but it was really difficult to afford the technology. I couldn't always afford the kits so had to do a lot of PCB production using nail varnish as an etch resist. Ah, the evocative scents of acetate and ferric chloride followed a day or so later with the smell and pain of burning flesh after the inevitable soldering mishaps...

[martb]

However, I enjoyed creating this proof-of-concept oscillator from first principles and hope you get some enjoyment and insight from it, too.

Very useful and enjoyable too!
Many thanks for all your excellent contributions here

[utdgrant]

Thanks Grant. That's taken me right back to when I spent all my income from doing a paper round and working Saturdays in an electronics store called TeleServicentre in Leeds on magazines like Electronics Today International, Practical Electronics, Everyday Electronics and Wireless World and the synthesizer projects they featured.

I think we must have been separated at birth! I too had a paper round, and used all my money to save up for electronic components to build audio circuits, as seen in magazines. I regularly bought Hobby Electronics, but would occasionally buy one of the others (PE, ETI, Elektor, etc) when an interesting audio project was featured. My ultimate dream was to own a 'proper' synth, and I was eventually able to afford a second-hand EDP Wasp after saving about 20 weeks' worth of paper round earnings.

There were a couple of decent electronics shops in Glasgow where I could buy discrete components, ICs and enclosures, etc. I didn't have to rely on mail-order suppliers, fortunately.

The trouble with the 'generic' electronics magazines was that they had to create projects to cater for all sorts of readers. Car projects, TV and radio projects, metal detectors, light dimmers, etc. It was a revelation when "Electronics & Music Maker" magazine appeared in 1981, which catered exclusively to all things audio and synth-related. I switched allegiances instantly and permanently.

Maplin were a great company but it was really difficult to afford the technology. I couldn't always afford the kits so had to do a lot of PCB production using nail varnish as an etch resist. Ah, the evocative scents of acetate and ferric chloride followed a day or so later with the smell and pain of burning flesh after the inevitable soldering mishaps...

All my home projects were tested initially on 'breadboards'. If I wanted to create a permanent design, I would then transfer the components over to 'Veroboard' (stripboard). I only got into 'proper' PCB design and etching when I went to Strathclyde Uni. The number of third parties who will manufacture short-run bespoke PCBs for you these days is fantastic!

[ColinP]

I think we must have been separated at birth!

Chuckles. I'm confident we share some ancestors.

It sounds like you were lucky enough to be born slightly later than me (1960) so you probably enjoyed a few things that didn't come along until my late teens. I got into electronics at the age of 11 so had many years in the "dark ages".

I did a few projects using veroboard but PCBs were a bit easier to design with once one got past the initial hurdles.

Yeah, mail order was a bit hit and miss with orders often coming back with one or two critical items "out of stock" but my Saturday job gave me staff discount on basic components and once or twice I hitch-hiked to London to buy ICs.

I didn't get a breadboard until about 1978 and within a few years of that I had moved over to software.

Recently hardware/firmware development has become much easier than it was and I've toyed with the idea of switching to Eurorack module development as the software environment is becoming increasingly frustrating to work in.

But there are many problems with hardware modular synthesis too, such as not being able to save and restore patches and the obvious financial and spatial limitations.

Plus when even someone as talented as Emilie Gillet retires one has to wonder about how much stress lies down that path.

[utdgrant]

I think we must have been separated at birth!

Chuckles. I'm confident we share some ancestors.

It sounds like you were lucky enough to be born slightly later than me (1960) so you probably enjoyed a few things that didn't come along until my late teens. I got into electronics at the age of 11 so had many years in the "dark ages".

Yes indeed. I consider myself to be incredibly fortunate to have been born in 1967. Not only was it an iconic year ("The Summer Of Love"), but it meant that I came of age around about 1980.

Musical technology which had been prohibitively expensive even a few years before, was becoming truly affordable - even to a schoolboy on a meagre budget. I'm talking about analogue monosynths, Casio 'home keyboards', weird second-hand paraphonic poly keyboards, guitar pedals and cheap analogue beat boxes.

As well as having access to all the great music from the '60s and '70s, the British Synth-Pop scene was getting underway around about '79 / '80, too. I was a big fan of these acts like (Vince Clarke era) Depeche Mode, Gary Numan, OMD, Human League, John Foxx, Ultravox, etc.

In a parallel thread, the 8-bit home computer revolution was happening in the UK. I started out with a 1K RAM Sinclair ZX81. Even that primitive machine opened up an entirely new world to me. Here was something that you could actually PROGRAM in a high-level language at home. It's difficult to explain just how much of an effect these early, limited machines had on me. In order to get better performance than the sluggish Sincalir BASIC, I taught myself Z80 machine-code from a book. I can't see me having gone through that process if I'd been born into the internet-connected PC generation. That intimate knowledge of how computers worked, right down at the nuts-and-bolts level, has defined my entire career to this day.

So, yes. All-in-all, I'm eternally grateful that my parents decided to have a third child at precisely that point in time!