MUFF WIGGLER Forum Index
 FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 
WIGGLING 'LITE' IN GUEST MODE

Aether Machine : Oscilloplasm
MUFF WIGGLER Forum Index -> Wiard Goto page 1, 2, 3, 4, 5, 6  Next [all]
Author Aether Machine : Oscilloplasm
slow_riot
6U compatible all analogue variable harmonic FM oscillator pair


As with all sound, it starts with a sine, in this instance generated through a state variable feedback oscillator yielding sine and cosine. This method of generating sinusoids is inherently accurate and this particular design uses mathematical symmetry to stabilise the loop without any distortion products. This is apt as if applied as an FM modulator any spurious signals will create unwanted sidebands. The sine and cosine both feature auxilliary second harmonic (i.e. 2 * frequency) outputs with same waveform purity.

This modulation oscillator and it's second harmonic are hardwired as the FM sources for the through zero primary oscillator via independent VCAs to allow control of FM index and harmonic balance. Voltage controlled amplitude changes are in decibels (i.e. not linear) to best correspond to the way the ear perceives changes in volume.

The primary oscillator implements the Associated Legendre differential equation, which describes the native oscillations of a celestial or planetary mass (so called spherical harmonics). 2 controls, Zonal and Tesseral allow the output waveform to move smoothly through harmonic series sinusoids. This is effectively an analogue implementation of a digital wavetable oscillator. It is formed through a state variable feeedback equation using 4 quadrant linear multipliers, with this equation also being controlled by a saw wave and the 2 VC parameters.

Every quantum particle in this module has been infinitely scrutinised to bring this idea into reality. Low noise, low interference design and fully balanced circuitry throughout. Initial tests reveal the exponential generator varies max 0.02% over a 40 degree change in temperature, meaning the exponential curve does not distort as is the case with tempco based designs.

Front panel design commissioned especially from the inimitable Zeke Clough.

I will be open for orders in approx 3-4 months once ready for production. More updates and information to follow as the design comes to fruition
davidh
following this Guinness ftw!
analogsteve
Exciting stuff. More manufacturers should infinitely scrutinize every quantum particle of their modules...
I hope it will be available with TT jacks.
slow_riot
analogsteve wrote:
Exciting stuff. More manufacturers should infinitely scrutinize every quantum particle of their modules...
I hope it will be available with TT jacks.


Thankyou! (and to Davidh)

In truth I think I should have said that every quantum probability has been analysed, as particle behaviour actually becomes wavelike in the quantum domain, but the sentiment remains.

Quantum physics is actually quite relevant to analogue design. Particles behaving like waves demonstrates that the laws of time break down, and in an oscillator feedback loop the whole circuit can "see" every other part.

TT only in fact! Pretty much every connection in the module is balanced for noise and interference rejection, so it does not make sense to use an inferior connection for interfacing with the outside world.
analogsteve
slow_riot wrote:
TT only in fact! Pretty much every connection in the module is balanced for noise and interference rejection, so it does not make sense to use an inferior connection for interfacing with the outside world.


I applaud your commitment to quality. Making the module TT only is an audacious (and possibly polarizing) move.
slow_riot
Having studied how signal and power interconnections are properly achieved, regarding noise as well as safety, I couldn't bring myself to do anything other than what I consider the best. I don't know whether that is autism or engineering, maybe both! Anyway TT and balanced connections costs the same as minijack, even cheaper over the lifetime of the connectors, so I do think it's the right choice. And if the "old way" still works for you adapter cables work perfectly.
wyrtti
This sounds so interesting. Hoping to hear more as it progresses.

I will get try out a Wiard 300 system next week. smile
drumsofd00m
Glad I bought a handful TT to 3.5mm adapter cables from Cary last year. I like the TT format and of course know its advantages, but converting old Wiard modules means considerable hassle and risk due to the need for shipping. FWIW, none of my other gear with analog I/O (FX or synths I tend to insert in the Wiard's signal path frequently for filtering, phasing, bitcrushing etc) have balanced connections either.

So, no polarizing intended, I'd just be surprised if many existing Wiard users changed their format.

OTOH, probably many of these have maxed out systems already and this module might mainly cater to new customers anyway.

Best of luck!
slow_riot
As you move towards larger system installations in pro studios, or live environments the need for balancing becomes critical.

Too much equipment, including DJ mixers, some modular PSUs (ahem), rely on a terrible hack that involves disconnecting the safety earth in everything except the output mixer to prevent ground currents flowing and generating hum. My friend who once got 240VAC from his mixer still swears by it.

When Moses came down from Mount Sinai he left an 11th commandment that was omitted from most religious texts and only now has it's meaning become known:

"Pin 1 is connected to chassis earth"

Looks like I might have to tool up for making adapter cables then! (thanks for the good wishes btw)
drumsofd00m
slow_riot wrote:

Looks like I might have to tool up for making adapter cables then! (thanks for the good wishes btw)


I thought Cary still makes those?

Of course in an ideal world all manufacturers would heed that 11th command (plus a 12th one for TT, as there's also balanced 6.3mm) and we could do away with D/I boxes and adapter cables... But there's a lot of other things that would change as well then - digital synths would have digital outputs and/ or high quality D/A converters, all synth presets would be dry without shitty internal reverb, MIDI would be reinvented by Don Buchla in liaison with Graham Hinton... ;-)
slow_riot
Yes, if only more people would decide to do it properly and strive towards the Ideal. smile
slow_riot
So now there is a working physical prototype for the sine modulation oscillator. I've been working on the design in a software simulator up until this point, a method that has it's critics, but I have to say that it is really an invaluable tool that covers a lot of ground. Quite pleased that the design works in the real world though and lots to learn from it.

My spectrum analyser shows the harmonic content of the sine to be really quite pure. The sine wave of a typical analogue VCO waveshaped from a triangle has quite a few peaks at harmonic intervals, this design is down generally minimum 20dB on each with some gone completely. A typical triangle core VCO has a bell shaped noise floor around the center frequency peaking at as much as -40dB below the primary, which is completely gone in this oscillator. I think this will be seen as a real musical improvement when used as a frequency modulation source.
slow_riot
Today I looked again at the Sine circuit, after some tweaks to power decoupling, especially on the pins of the AD633 multipliers, which are like magnets for stray current. As well as adjusting the "compression" circuit which makes sure the oscillator feedback loop stays in check (I backed off on the amount of damping force it exerts on the oscillator), the noise floor of the Sine is -100dB below the fundamental across the frequency spectrum.

The drawback of this design is that this analogue computing approach gives a much smaller frequency response, I've tweaked for sub sonic to 1.5khz , which is obviously still some ways off a classic oscillator, but given the scope in the module for timbral modulation, I think there is a lot to be found. (e.g. the primary oscillator will give harmonics way in the many khz, and the second harmonic of the Sine modulator will extend the 1.5khz to 3khz), and then there are the resulting sidebands from FM.
slow_riot
I am interested in people's views regarding panel layout, legending and final feature set. Obviously there are compromises but I want to make sure I settle on the right ones. Here is a very rough mockup.



First up, the hardwired FM functionality. Hardwiring it has just seemed to be the right choice. Sometimes FM is a pain in the arse and you just want a single non FMed oscillator, and be damned with finding the right ratio. But the way FM is implemented in this oscillator is not something that I can wire to the front panel easily, as you need a 2 phase oscillator, and the onboard sine is symmetric enough to not require AC coupling, apart from the second harmonic which has a large offset. This is removed with a DC servo circuit rather than the traditional AC coupling cap.

The LEDs, I was kindly given a set of super pretty RGB 5mm LEDs, and the way I am wiring them is to use all 4 as a multicoloured oscilloscope type display for the primary oscillator output. I think this positioning is about right, they need to be in a square as transposing the oscillator onto this map is easily possible with onboard circuitry, not so with 4 in a line.

The legending on the panel. LIN controls are the base FM indices of each osc. Is this obvious? FM1 ENV and FM2 ENV, is it obvious what they are for?

The 2 switches below Zonal and Tesseral switch the VC input between positive polarity and negative polarity.

Each osc has 2 1v/oct inputs, where one 1v/oct has a switch between pos and neg (you can thank me for that later).

The sine / cosine and 2nd harm aux are outputs, I will have them drawn properly, along with the primary outputs to their right. I think this graphical flow is appropriate.

Edit.... so the terms Zonal and Tesseral aren't being used properly, they are related but not in the way I was thinking. I need to have a think to describe these inputs in a musical way.
wyrtti
Hi Mark!

And thanks for the picture. To my eyes it seems pretty clear. The one thing that sticks out, when compared with Wiard 300 stuff, is the square positioning of the LEDs. But you explained the reason for that, so it's not a biggie.

Cheers,
Topi
slow_riot
Thanks for the feedback Topi. Hopefully it will be worth it when you see the RGB LEDs in action, they have 3 LEDs in one, so by changing the levels of each you can morph through the colour spectrum. I will have to think about the exact positioning so that it won't stand out so much when racked but it will also sit properly within itself.

Hopefully the final layout will follow the great tradition established by Wiard, whose modules are really a masterclass in ergonomics and art.
plord
I am trying to picture the distribution of switches and jacks; compared to most classic Wiard, where the switches are generally in isolated spots with no cable interference, this seems like it may pose some challenges in a dense patch.

Have you tried a layout where the grouped functions are in rows but the left/right side functions are grouped vertically instead of horizontally? So, the switches would be grouped in columns above all of the jack inputs, with the related inputs immediately below; all of the shared outputs and the inputs stay in rows but move down. Make sense?

Simpler description: take the FM inputs and their switches and move them above the graphics and outputs, then rearrange the FM, V/Oct, Zonal/Tesseral into left/right groupings that isolate the switch field outside of the cable plant.
slow_riot
Thankyou Paul. I got a bit carried away with the switching functionality without considering that, especially with my love of bantams that switch placement is suicidal!

From a signal flow point of view I really want to keep the current vertical and horizontal arrangement. This helps simplify the internals of the module a great deal to be able to display the exact flow across the connectors.

So what I've done is to swap out the polarity switches for FM and the harmonic control inputs, in the Wiard lexicon the Envelator has +/- outputs anyway. The accessible switches are dedicated to one 1v/oct on each side.

I will leave the 4 row blank, and wire it according to users specs, either a 4 way multiple or 2 inverters.

thesnow
you couldn't implement 4 more simple led's and line them up to match the other wiard modules?
slow_riot
check this vid of what RGB LEDs look like smile I think it's worth the slight offset.

https://www.youtube.com/watch?v=jvcHy4gvkxM

I could put 4 of these in a row but then I need to design circuitry to cycle between the 4, at the moment there is a square wave used in the primary that I can use to switch between 2 sets of 2... I'm already really pushing the limits of what will fit in the enclosure, cost and weight wise.
thesnow
looks and sounds awesome
slow_riot
Thanks, when you are back in the fold I'll be here smile

I made a change to the last image... the 2 harmonic controls are now named DIVIDE and MUTATE !!

Here's a fun fact, the primary oscillator is made of 7 multiplication operations, 8 additions (opamps), 3 memory elements (capacitors), and two 1-bit ADCs (comparators). Analogue is really quite computationally efficient (until you go start counting the individual transistors, that is)...
slow_riot
Sine wave plus 4 VCAs board, take 43!

slow_riot


Just about ready to fire up the Primary oscillator prototype, the final piece of the puzzle was some circuitry to prevent certain combinations of Divide or Mutate from occurring that would have caused the oscillator to lockup.

The circuit is not that far away from a standard Sine/Cosine. On the top left there is a Saw oscillator that is used to generate the term (t squared -1) that is applied to the feedback loop. Zener diode limiting on one side and a reset switch on the other to ensure the feedback loop stays locked to the saw completes the picture.

The Modulator sine is absolutely distortion free, this one will have some slight distortions due to the added complexity, but overall I think it will be very pure and clean. If the Waveform City could be considered to be like a male formant structure, this one is probably more female.

The LEDs are not for audio reasons, on the final PCB there will be a set of LDRs next to them which are used to cycle the oscillator signal between the LEDs on the front panel. Rather than use complex switches etc I thought I would just use what is already there. Waste not, want not, as they say! smile
slow_riot
front panel first draft...

MUFF WIGGLER Forum Index -> Wiard Goto page 1, 2, 3, 4, 5, 6  Next [all]
Page 1 of 6
Powered by phpBB © phpBB Group