Oscilloplasm - complex VCO pair

Earthside communication channel for 6U format devices

Moderators: Kent, slow_riot

Post Reply
User avatar
davidh
Ultra Wiggler
Posts: 776
Joined: Mon Dec 06, 2010 3:58 am
Location: Europe/France/Alsace

Post by davidh » Fri Jun 24, 2016 5:05 am

If it's the same depth than other modules, that won't fit into my case :waah:

Image

slow_riot
Super Deluxe Wiggler
Posts: 4002
Joined: Tue Jun 30, 2009 8:58 pm
Location: uk

Post by slow_riot » Fri Jun 24, 2016 10:13 am

Oh no, as if there wasn't already enough bad news today!

I was thinking about reducing the mounting profile to make the modules better for gigging.

It's 210mm deep. with the lower part of the chassis 13.5mm up from the lowest part of the front panel. but it does look like you will have problems with that case, at the moment, I'm afraid.

We will find a way to make it happen, and trust me, you do want this VCO.

slow_riot
Super Deluxe Wiggler
Posts: 4002
Joined: Tue Jun 30, 2009 8:58 pm
Location: uk

Post by slow_riot » Fri Aug 12, 2016 7:13 am

Sorry for the delays people. The manufacturers for the chassis I spoke to simply would not deal with me with the numbers I was talking about. But thankfully I have found someone who will.

The rev B PCB prototype is coming back from the plant next week which is going to be the final rev before production. Will assemble and put up some demos of the sine on variable harmonic FM, which is where this VCO gets super interesting!

BTW anyone in UK I'm doing a short presention for Sines and Squares festival in Manchester under the topic "Analogue electronics and the quantum domain”. I think analysis of some analogue circuits with respect to our understanding of sub-atomic particle behaviour is scientifically and artistically relevant.

http://sines-squares.org/

KoryB
Common Wiggler
Posts: 246
Joined: Fri Jun 17, 2016 12:07 am
Location: Bay Area

Post by KoryB » Sat Aug 13, 2016 3:10 am

Glad you found a manufacturer that might work out.

I think it's going to be really cool to have more modules in this format.

slow_riot
Super Deluxe Wiggler
Posts: 4002
Joined: Tue Jun 30, 2009 8:58 pm
Location: uk

Post by slow_riot » Sat Aug 13, 2016 3:40 am

KoryB wrote:Glad you found a manufacturer that might work out.

I think it's going to be really cool to have more modules in this format.
It's my cousin Mike, who is an industrial engineer!

http://jvdautomation.co.uk/

Thanks for the support.

slow_riot
Super Deluxe Wiggler
Posts: 4002
Joined: Tue Jun 30, 2009 8:58 pm
Location: uk

Post by slow_riot » Thu Sep 15, 2016 9:48 am

Have been testing the Rev B board. Cut outs in the ground plane has meant that I can get the parasitic resistance across the integrators in the main oscillator down to very low amounts (probably around 1 G ohm), but this still causes the oscillator to damp out at higher frequencies. But, the regenerative feedback circuit, which excites small ( <0.7V ) signals works perfectly to ensure the oscillator works at all settings (at the cost of a little distortion).

I had to tweak the negative feedback/damping to prevent the oscillator from overloading. The primary oscillator is too complex to use an elaborate control mechanism such as in the sine wave osc. What we have instead is a 2 stages of subtle compression/waveshaping and 1 of hard limiting to ensure the oscillation takes the form we desire.

Aside from the sequencer/quantizer, I have started the work on a very similar module to this. Instead of solving Legendre, the next will compute the Bessel equation, which describes the vibrations of a circular membrane (e.g. kick drum). It will otherwise be very similar to this one, including through-zero linear FM and on board sine/cosine modulation osc.

slow_riot
Super Deluxe Wiggler
Posts: 4002
Joined: Tue Jun 30, 2009 8:58 pm
Location: uk

Post by slow_riot » Thu Nov 17, 2016 4:43 am

OK synth fans. I've got the prototype ready to show off for the Sines and Squares Festival. I will admit to being a fan of the thing. It's very interesting how a circuit takes on properties. It sounds like the music that I like myself. I personally wouldn't have expected that given that it's just maths, but maybe the way it comes together as an instrument where the focus is on FM lends itself towards certain things. Of course I would hope that other people can make it sound like things that they like with it, too.

I fixed the distortion problems, and the problem with the oscillations decaying. Tracking on both VCOs is spot on. The calibrated control for harmonics is causing me headaches ATM and I will have to look deeper into that before I order up the final PCB.

When I come back from Manchester I'll record some proper sound files.

We changed the manufacturing process for the panels. Compared to the Wiard panels I really didn't like how dark the text and graphics were on my panels in "mood" lighting (which I prefer for studios and live is pretty much always dark). In stopping to think about that Luigi suggested oxidizing paint as an alternative way to colour the panels. I look forward to seeing how that will turn out.

User avatar
davidh
Ultra Wiggler
Posts: 776
Joined: Mon Dec 06, 2010 3:58 am
Location: Europe/France/Alsace

Post by davidh » Thu Nov 17, 2016 5:04 am

sounds great
can't wait to see some demo :guinness:

slow_riot
Super Deluxe Wiggler
Posts: 4002
Joined: Tue Jun 30, 2009 8:58 pm
Location: uk

Post by slow_riot » Thu Nov 17, 2016 6:10 am

Sine and twice harmonic Sine FMing the main VCO section:

http://aethermachine.files.wordpress.co ... 7_1051.wav

KoryB
Common Wiggler
Posts: 246
Joined: Fri Jun 17, 2016 12:07 am
Location: Bay Area

Post by KoryB » Fri Nov 18, 2016 12:34 am

slow_riot wrote:Sine and twice harmonic Sine FMing the main VCO section:

http://aethermachine.files.wordpress.co ... 7_1051.wav

Sounds great!

When will it be ready for purchase?

slow_riot
Super Deluxe Wiggler
Posts: 4002
Joined: Tue Jun 30, 2009 8:58 pm
Location: uk

Post by slow_riot » Wed Nov 30, 2016 2:51 pm

KoryB wrote:

Sounds great!

When will it be ready for purchase?
Hi Kory, I'm trying to avoid making estimates any more, learned the hard way about that! But it's close, and with the infrastructure in place it opens the channels for more designs.





One thing with the oscillators, is the Mod(ulation) input on the Sine, which has the same response as the VCO itself which uses exponential responding VCAs. Just by way of explanation I chose Mod for the panel as Exp might get confused as another input for the master osc. Anyway I had a last minute freakout as the input should really be linear for the purpose of vibrato, but there just isn't space on the board to linearize the modulation input (lucky it wasn't possible to change because we have already made the panels).

To compensate for this, in a planned future module there will be 2 utility logging circuits. Run your modulation signal into them (which will have their own level controls, as the Mod pot needs to be fully open to correctly linearize the exponential response of the VCO), and then into the Mod input and it will be possible to have proper linear modulation of the Sine. The Logging utility circuits will come in handy for other things too which I will reveal more in due course.

slow_riot
Super Deluxe Wiggler
Posts: 4002
Joined: Tue Jun 30, 2009 8:58 pm
Location: uk

Post by slow_riot » Fri Jan 27, 2017 11:36 am

Soz guys, still not there yet! Further digging into calibrated exponential control for harmonic led me to a major mistake right at the core of the oscillator.

So now that's been sorted, and I'm improving the amplitude stabilization to use an RMS level detection instead of crude limiting.

So, it's looking like another couple of months wait, but it will be worth it!

Initial sale price will be £700, it's high but I am confident that it will be a safe investment. EU customers are advised to get it before we go through with all this Brexit nonsense. It looks like I will need to go VAT registered when that happens.

Thanks for the support/patience/etc.

User avatar
davidh
Ultra Wiggler
Posts: 776
Joined: Mon Dec 06, 2010 3:58 am
Location: Europe/France/Alsace

Post by davidh » Fri Jan 27, 2017 11:41 am

:tu:

slow_riot
Super Deluxe Wiggler
Posts: 4002
Joined: Tue Jun 30, 2009 8:58 pm
Location: uk

Post by slow_riot » Fri Jan 27, 2017 11:55 am

It's too impractical to change the module chassis size BTW to fit angled cases like yours. Mounting the PCB in vertically gives much more usable space and structural strength, and I'm doing circuits which are crazy high density which wouldn't fit behind the panel any other way. It really is the only option. Sorry about that!

User avatar
neil.johnson
Super Deluxe Wiggler
Posts: 1292
Joined: Thu Apr 08, 2010 3:51 pm
Location: Cambridge, UK
Contact:

Post by neil.johnson » Wed Feb 22, 2017 9:15 am

I was re-watching this vid the other day and noticed the slide at 39:29:
[video][/video]

Is this similar to your scheme of using the sin-cos-squared trig identity to control the amplitude?

Neil
Modular stuff :: http://www.njohnson.co.uk

slow_riot
Super Deluxe Wiggler
Posts: 4002
Joined: Tue Jun 30, 2009 8:58 pm
Location: uk

Post by slow_riot » Wed Feb 22, 2017 2:37 pm

Hi Neil

Yes, it's the same idea to use a trig identity to generate a real time error term for the state variable equation. I'm not sure 100% what is entailed by using the error factor to control damping diodes, maybe he's using the DC error term to bias them in or out of conduction. Using a multiplier to apply positive or negative feedback to one integrator proportional to the real time error is how I've done it, which works perfectly.

The Sine/Cosine is the easy part of this dual VCO. The primary oscillator is so endlessly complex that I can be sure that no-one will be able to reproduce it, even if I shared the schematic. LOL.

slow_riot
Super Deluxe Wiggler
Posts: 4002
Joined: Tue Jun 30, 2009 8:58 pm
Location: uk

Post by slow_riot » Sat Apr 08, 2017 4:39 pm

Oof! Still not there yet. Trying to figure out what was going on with the exponential control for harmonic led me to try doing the Chebyshev equation instead of Legendre, but whilst that one is easier to control the harmonics (using an N^2 term as opposed to N^2+N), the core equation is much more complicated (in analogue circuits, complexity = bad). The simplification of the Legendre equation by Levine in his Analogue Computing bible is one of the keys to this VCO design, which gets around a potential division by zero at the singularity of the oscillation by sleight of hand, a trick that only really seems to work when the equation is in a simplified form.

So, as it stands now I've gone back to the earlier revision with a few lessons learned long the way (including distortion free amplitude control using the *excellent* THAT4301 RMS to DC engine). I'm hoping my final attempt to crack the proper scaling for exponential control of harmonic will work, if it doesn't I might have to call it a day on that feature :( , but we will see. :deadbanana: :deadbanana: :deadbanana:

slow_riot
Super Deluxe Wiggler
Posts: 4002
Joined: Tue Jun 30, 2009 8:58 pm
Location: uk

Post by slow_riot » Sun Apr 30, 2017 11:13 am

Ordering final PCB proto board this week. Ready for first orders hopefully 6-8 weeks after.

I am very happy with how the new RMS level detection circuit works. The previous technique I was using forced the master oscillator to be stable under all settings, even the highly dissonant ones. It also corrupted the waveform at the peaks and near the middle.

An RMS to DC converter is a little analogue calculator that measures the average power in an AC waveform, which gives a smart way to make sure the oscillation stays at a constant amplitude (studio compressors use this exact technique). Aside from this technique giving minimal distortion for automatic gain compensation, it does one other thing I wasn't expecting. The atonal "in between" settings are now allowed to naturally decay rather than being forced to be stable and oscillate. This makes it seem like the oscillator has a mind of it's own, as the oscillations do damp out when the oscillator is in settings that are not naturally stable. It also makes the atonal sounds much more interesting, as they are lower level and you can hear the compressive action trying to inject some life into them, meaning you get low level reverberant artifacts in the "bad" settings. In general, I personally don't like the atonal settings, but it's nice to go past the odd settings when moving between the consonant ones.

The main problematic aspect of RMS is that you get a "breathy" effect as it takes several cycles to get a reading, so it does tend to overcompensate then undercompesate with compression causing a slow pulsing amplitude modulation, but for drone music (of which I am unashamed fan) this sounds great. In fact, I decided to bring the RMS level detector output onto a header on the PCB (which gives a 0-10V output that follows the master oscillator amplitude). So, for the curious, you could connect this to one of the positions on the 4 way multiple, or onto your own jackfield. This would be *fantastic* to control a VCA that is controlling FM depth, so the natural amplitude changes of the oscillator are re-inforced by the FM depth.

Back to work now to make sure the 1V/oct control for harmonic tuning works as expected! Not going to take no for an answer this time, mischevious electrons!

slow_riot
Super Deluxe Wiggler
Posts: 4002
Joined: Tue Jun 30, 2009 8:58 pm
Location: uk

Post by slow_riot » Tue May 02, 2017 11:36 pm

Messing around with basic FM patch. Some caveats, number 1 is that I don't have any supporting modules or other gear like sequencers. Number 2 is that this is not the final rev PCB, so some of the behaviour isn't 100%. In some instances the amplitude modulation I spoke about is obvious, and it is easy to see how an envelope follower type output would be useful in a patch.

http://aethermachine.files.wordpress.co ... 3_0430.wav

User avatar
J3RK
Super Deluxe Wiggler
Posts: 6551
Joined: Tue Dec 22, 2009 12:46 pm
Location: Seattle

Post by J3RK » Thu May 04, 2017 5:30 pm

slow_riot wrote:Messing around with basic FM patch. Some caveats, number 1 is that I don't have any supporting modules or other gear like sequencers. Number 2 is that this is not the final rev PCB, so some of the behaviour isn't 100%. In some instances the amplitude modulation I spoke about is obvious, and it is easy to see how an envelope follower type output would be useful in a patch.

http://aethermachine.files.wordpress.co ... 3_0430.wav
That sounds nice all on its own. Can't wait to hear it with some supporting modules/circuits as well. Seems like this in conjunction with a couple of "complex" envelopes (something with a lot of stages) would be very cool. Get some ringing with multiple levels of rise and fall. Anyway, it's sounding very good.
Are you lemon? Does your head come to a nub?

slow_riot
Super Deluxe Wiggler
Posts: 4002
Joined: Tue Jun 30, 2009 8:58 pm
Location: uk

Post by slow_riot » Thu May 04, 2017 5:59 pm

J3RK wrote:
That sounds nice all on its own. Can't wait to hear it with some supporting modules/circuits as well. Seems like this in conjunction with a couple of "complex" envelopes (something with a lot of stages) would be very cool. Get some ringing with multiple levels of rise and fall. Anyway, it's sounding very good.
Thanks Dustin, I'm much happier with the sound now. The RMS to DC conversion meant I could get the "bad" distortion out of the main oscillator, then I used an OP275 in voltage follower mode as the output amplifier (probably any JFET input opamp would do though) to add some "good" distortion onto the cleaner signal.

It will certainly be interesting to see how it gets used by other people with some supporting modules.

slow_riot
Super Deluxe Wiggler
Posts: 4002
Joined: Tue Jun 30, 2009 8:58 pm
Location: uk

Post by slow_riot » Sat May 13, 2017 4:49 pm

Sorry to anyone who is bored of these updates, but I sold many of my possessions to fund it rather than combine long winded perfectionist analogue design with time sensitive crowdfunding so I think I've earned the right, mwuahaha.

Final PCB prototype in production now. I decided to move to a multilayer PCB with dedicated planes for +15V,-15V and 0V. This technology was invented for digital techniques but gives huge gains in performance for audio frequency processes too. Signal layer is on the bottom so rework is no problem.

Calibrated 1v/oct control of waveshape/harmonic now works properly.

I switched to a different technique for shielding at the last minute too. Best practice in the audio field as pioneered by Neil Muncy and enshrined by AES48 is to try to maintain a separate chassis ground and audio ground, to force noisey shield currents away from sensitive signal ground. An alternative strategy that is commonly found in industrial locations is to directly bond signal and chassis grounds. For common impedance coupling this bonding addresses the mechanism that induces noise voltages directly by reducing the common impedance, rather than to try to manage the effects of it. Equipotential bonding gives large gains for EMC as as the module chassis makes a very effective RF ground. (Using a metal body as a ground is how planes don't generally fall out of the sky). (Word of warning though, the module absolutely will not work on power supplies which are not earthed and therefore have no DC ground).

Prettiness of RGB LEDs doesn't show too well, unfortunately. For the production panels we are moving to using patinas which induce metal to corrode and give natural rusting metallic colours.

Image

slow_riot
Super Deluxe Wiggler
Posts: 4002
Joined: Tue Jun 30, 2009 8:58 pm
Location: uk

Post by slow_riot » Tue Jun 20, 2017 11:22 am

The circuit design is finalised. I'm now waiting for my chassis and faceplate contractors who I am hoping will keep to schedule.

I needed to raise the module price to £800 (price includes VAT for EU customers, £600 for non EU) to cover the cost of manufacturer and give me a high enough margin to stay sane in the longer term, sorry about that. VAT registration was something I had hoped to avoid but it covers my bases with the uncertainty of future borders between UK and EU.

In order to cover some of the final costs of production I am looking for several customers who are prepared to make a slight leap of faith, and order at a reduced price of £700, with a 4-6 week lead time. Once my chassis and metalwork guys are ready to ship (which I hope to be within the next month) I will post again that I am ready for these first orders, which will cover PCB production/high quality pots/ICs.

After those initial orders (and no problem if anyone doesn't want to take the risk), the module will be available at the final price with a 1-2 week lead time, max.

Thanks to everyone who has followed the design from concept to manufacture. The module will absolutely not please everyone, but I am hoping it will please a selected few a great deal.

slow_riot
Super Deluxe Wiggler
Posts: 4002
Joined: Tue Jun 30, 2009 8:58 pm
Location: uk

Post by slow_riot » Mon Jul 10, 2017 8:50 pm

no forward movement since last time, but I have a new demo. Still don't have a sequencer I'm afraid!

http://aethermachine.files.wordpress.co ... 0_2004.wav

slow_riot
Super Deluxe Wiggler
Posts: 4002
Joined: Tue Jun 30, 2009 8:58 pm
Location: uk

Post by slow_riot » Wed Jul 19, 2017 9:24 pm


Post Reply

Return to “Aether Machine”