Chysn's Gear Rant Thread...

chysn · « **Reply #100 on:** August 08, 2016, 08:13:10 PM »

I successfully completed my first firmware modification for my Mutable Instruments Peaks. I wanted the sequencer to have a variable number of steps (the factory firmware sets it to four steps). I added a line of code that resets the sequencer to the first step when it hits a knob that's turned all the way down. That is, if I turn the fourth knob all the way to the left, it becomes a three-step sequencer.

I wanted to start with something simple, because I spent enough time installing and learning the ARM toolchain and getting familiar with the original source code. Now that I have some confidence in it, there are various projects that I want to do, most of which involve totally beefing up the sequencer. For example, I'm pretty sure I can turn it into an 8- or 16-step sequencer once I design a reasonable UI workflow. I'm starting by taking inspiration from the ModSeq, which has four knobs, but gets eight steps by inverting voltages on the back four steps.

Now that I can compile and install my own firmware, the sky's kind of the limit.

chysn · « **Reply #101 on:** August 14, 2016, 08:24:24 PM »

Peaks, Part II:

I basically spent today writing my own sequencer for Peaks. I wanted a sequencer that could (1) skip steps and (2) quantize notes chromatically (accurately).

I've made good progress, but I learned (the hard way) two important things about Peaks.

First, its processor does not support floating point types. C++ code using float compiles, but Peaks crashes during runtime when you try to use a float variable for anything. So that immediately put a damper on raising the twelfth root of 2 to the anythingth power. Fortunately, the thing supports 32-bit integers, so I could make a lookup table and do the math with integers. Since I have no way to debug other than "compile, install, see what happens," figuring this out took up my entire afternoon.

Second, the positive and negative potentiometer values are asymmetrical. 1 volt is at around 3800, -1 volt is at around -4500, with 0 in the center. I've got the values tweaked pretty well (using the chromatically-quantized Octone for reference), but I'll have to grit my teeth, dig in, and get the tuning just right to make the sequencer more usable.

My next trick is going to be to retool the Split function to make a 2:1 clock divider, so that my Peaks sequencer advances one step for every two gate signals from the eight-step Octone.

BobTheDog · « **Reply #102 on:** August 15, 2016, 12:26:25 AM »

The Cortex M3 doesn't have an FPU, the standard libs do support software floating point though. The compiler/linker should work this out if you are telling it that you are using an M3?

Better still use fixed point as it will be much faster, you could try http://www.quinapalus.com/qfplib.html it is tiny in size and does the job.

chysn · « **Reply #103 on:** August 15, 2016, 04:12:14 AM »

Quote from: BobTheDog on August 15, 2016, 12:26:25 AM

The Cortex M3 doesn't have an FPU, the standard libs do support software floating point though. The compiler/linker should work this out if you are telling it that you are using an M3?

Better still use fixed point as it will be much faster, you could try http://www.quinapalus.com/qfplib.html it is tiny in size and does the job.

Yeah, the target is specified as M3 (and I'm just using Mutable's makefile). It's hard to know what's going on, as I don't have a way to collect any info about failures.

32-bit int is big enough for the fixed-point precision that I need for this particular project. A blunt instrument, but not uselessly blunt.

Thank you for the Qfplib link. After I get my other features working, I'll circle back to my quantize function and try it out.

BobTheDog · « **Reply #104 on:** August 15, 2016, 11:34:29 PM »

Are you using the JTAG to program it? Doesn't that also support debugging?

chysn · « **Reply #105 on:** August 16, 2016, 03:42:35 AM »

Quote from: BobTheDog on August 15, 2016, 11:34:29 PM

Are you using the JTAG to program it? Doesn't that also support debugging?

Right now I'm just using the WAV file update procedure. If I get a JTAG or FTDI, it would be much faster and easier, and I'd feel like I have more freedom to try weird things without it taking forever. It's a great idea. Thanks again, Mr. The Dog!

chysn · « **Reply #106 on:** August 17, 2016, 09:41:18 PM »

I wrote my second Peaks sequencer today. This one is a modulation sequencer called Rizm, and its focus is on rhythmic sequences rather than melodic ones.

Rather than assign each of the four knobs to one step, Rizm is an eight-step sequencer that uses the knobs for levels of beat divisions for whole, half, and quarter notes. Then the fourth knob selects between 25 eighth-note accent patterns with a set level, plus a bonus level based on the other knobs. I keep shifting the algorithm around, but so far it's pretty musical.

chysn · « **Reply #107 on:** August 21, 2016, 09:04:04 AM »

I asked about the aforementioned zero-volt asymmetry on the Mutable Instruments forum. A few hours later, Olivier Gillet committed a DAC output calibration process to the Peaks firmware repository. That's a ridiculous level of customer support. I mean, who does that?

BobTheDog · « **Reply #108 on:** August 21, 2016, 12:02:47 PM »

I might get myself one as well to mess around with, it all sounds fun.

A few years ago I looked at getting an ADAC001, 002 and 004 but the cost was around 700 euros and I balked.

I then started creating my own programmable module using an M3 as the MCU and two Freescale DSPB56371s for the DSP. It had a touch screen, 8 CV ins, 8 gate ins, 8 CVs out, 8 gate outs and 16 endless encoders. The idea was that you could use the touch screen to connect the internal software modules to each other and to the ins and outs and encoders and then save this setup as a preset, a bit like a simplified Nord G2 in a CV module.

It also supported Open Sound Control for control from a computer. I got the basics going with some dev boards and a mess of breadboards and then realised I just didn't have the time. I would love a flexible module like this though, maybe if I ever retire I can get on it again!

chysn · « **Reply #109 on:** August 21, 2016, 12:43:01 PM »

What are the ADAC001, 002 and 004?

BobTheDog · « **Reply #110 on:** August 21, 2016, 01:12:21 PM »

Sorry, should be ADDAC:

http://www.addacsystem.com/products/ADDAC000%20Series

chysn · « **Reply #111 on:** August 21, 2016, 03:52:36 PM »

Oh, okay, kind of a programmable CV box. I think I prefer the Mutable Instruments approach to user-programmable modules. You can look at the panel, and get a good idea about what they do, but the firmware is open-source, so you can either tweak it, or stretch it. Peaks threads that needle really well. The ADDAC001 just makes me feel tired.

Incidentally, something like Disting doesn't even appeal to me that much, either, because of its opacity. I can't really tell you exactly where my "multi-function" dividing line is, but I like there to be some functional coherence. That is, I can say "Peaks converts triggers to CV levels," whereas Disting does everything. You could build a whole synth out of a half dozen or so Distings, but completely lack clear labeling.

BobTheDog · « **Reply #112 on:** August 22, 2016, 03:58:17 AM »

Never seen the Disting before, looks very useful!

I stopped looking at modules as the next one will require a new case and madness lies there!

chysn · « **Reply #113 on:** August 22, 2016, 04:15:29 AM »

Do you have a link to your current system on Modular Grid? I'd love to see what you're using.

BobTheDog · « **Reply #114 on:** August 22, 2016, 06:11:27 AM »

Never used modular grid, my setup is pretty boring, no fancy stuff.

There is a pic somewhere on the forum, let me see if I can find it...

BobTheDog · « **Reply #115 on:** August 22, 2016, 06:13:27 AM »

here it is:

http://forum.davesmithinstruments.com/index.php/topic,47.msg144.html#msg144

chysn · « **Reply #116 on:** September 23, 2016, 07:21:33 PM »

I wanted to do this as part of a different thread, but I was concerned that it would take that thread off-topic.

Here's a short demo of a lag processor as a waveshaping tool. https://soundcloud.com/beige-maze/lag-sweep-sine

In this demo, I'm playing a self-oscillating filter (DSM01), and running that directly into a Make Noise Function, and just playing with the rise and fall times while playing something random with my left hand.

As the rise time increases, the sine becomes saw-ish, and then tips over into greater complexity.

Note that the amplitude falls off pretty quickly, probably as the rise time exceeds the wavelength.

This sort of thing makes me want to get a third Function, because this module not only flexible, but it's just beautifully elegant in its controls.

The tracking on the DSM01 is decent, too. Below 0V, it loses its way completely, but above 0V it's pretty solid. I'd say that the 4075 takes 1V/8ve more seriously (e.g., calibration and a dedicated 1V input), but this is the DSI forum, after all, so I'll use that in these kinds of demos.

chysn · « **Reply #117 on:** November 25, 2016, 09:55:52 AM »

Well, my eurorack synth is done! Meaning, I'm out of space. There are actually a few small things left to do, like upgrade the power supply, and upgrade the low pass gate, but its capabilities are more-or-less finalized according to what I set out to do.

It's basically arranged into four quadrants: sound source, filtering and effects, composition tools, and modulation, with each quadrant divided by a mixer/attenuator. The final piece, which arrived today, was an Intellijel µScale.

The µScale is great. Most quantizers have the ability to choose a scale (major, minor, chromatic, etc.), but the µScale is the only one I know of that allows you to create your own scale in real time. The buttons correspond to notes on a piano keyboard, and when you push them, they toggle whether or not that note is quantized. There are presets for recalling various scales in various keys, but if you paid attention in music theory, it only takes a few seconds to punch in whatever you want without using presets. Then you just feed it a modulation source or two and see what you get.

BobTheDog · « **Reply #118 on:** November 25, 2016, 11:07:16 AM »

I think you will find you now need a bigger box, Christmas is coming...

chysn · « **Reply #119 on:** November 25, 2016, 12:34:15 PM »

Quote from: BobTheDog on November 25, 2016, 11:07:16 AM

I think you will find you now need a bigger box, Christmas is coming...

Sometimes it's okay to be done, even with a eurorack system. I'll never be one of those wall-of-modules guys. I expect to make modest tweaks over time (like, I can see swapping out the sequencer, filters, etc.), but the breakneck larval stage is over.

For one thing, it's hard to build a box, for an inexperienced carpenter like me. I'm sure it'll be easier the next time I decide to do it, but it's still not something I wish to undertake this soon. So now it's time to pause and compose for the marvelous instrument I put together.

My next significant music purchase, actually, must be a decent trumpet for my son.