Beta blog: development log with a bit of random yak shaving


#21

Enclosure update

We’re partnering with a company that specializes in quick-turn low tooling cost plastic enclosures, because we want to offer up flexibility for when you eventually want to make some changes to the synth and want a pretty way of packaging it up.

Currently, plastic injection is the best per-unit price, but then the design is fixed. We considered going with laser-cut acrylic, but a lot of people complain of this looking amateurish. As if being an amateur was anything besides exciting ;). We’re still going to use laser-cut acrylic for concept synths, low run, or just for making something fun.

This new approach, on the other hand, is a great compromise between wanting to make 10 of something, without needing a $7000+ mold. I’m really excited what this is opening up and I can’t wait to see what people do with the concept.

We’re still in the engineering phase, I’ll let you know how things look after we get a prototype. I’m really eager to start sharing videos, because this synth sounds amazing.

Synth Models

What I’m most excited about is working on the actual synth models. A couple weeks ago, we made a 16xN step sequencer. It featured four simultaneous 16-step sequencers, with the ability to scale to N number of sequences. It needs a lot more work to make jumping pages, copy, paste, edit, and all of the menu-sugar to make it lovable, but at this point you can save and load sequences, and in case I sound like a broken record, these synths sound great.

I’m not even doing anything that exotic. Simple sine oscillators with an ADSR filter, step-volume, step-pitch already give the synth some amazing sounds, especially when combined with polyphony. I can’t wait to hear it with more complex oscillator models.

After getting the basics of the step sequencer ready, we’ve started a beatbox, analog drum machine. Think of live drum pads with knobs to control and tune the instruments (which could be anything, sine, sax, strings, or I guess a snare, if you’re a slave to the rhythm).

We started with four osc polyphony and have been optimizing the code to eight channel simultaneous triggering. I’m not sure if we’ll get up to 16, but we’re going to make this module as small as possible, so you get as much room on the chip to add your own flair. We’ve got basic play and saving/loading working. After it’s optimized, we’ll be adding menu-sugar, but also a pattern mode where you can turn on and off notes for triggering. Live record / playback might require a hardware change, but we might be able to squeeze that much functionality into the system. We should at least be able to shoehorn in a rigid 16 step recording function.

PCB1003 Piano update

The new PCB is in, parts are soldered on and we will be testing shortly. We had to break bit order (Pad0-Bit0) in order to get the signals far enough away from each other, so we shall see if physics has prevailed!

Pricing / Ordering / Shipdate

We are postponing pricing to April 15th, and ordering until May 15th. We’re likely to be able to start shipping out before then, but


#22

Piano / Touchpad update

Great news! When we started picking replacements to the MPR121R captouch chip, we were worried that the alternative was going to have a huge trade-off. More code, more firmware, more programmers, more expensive, less robust, less responsive.

While not the smoothest of platform on-boardings, the Silicon Labs “Simplicity Studio” does a moderately okay job of getting software on your computer that can program the CPT112S. It’s leaps and bounds ahead of anything using Eclipse, because you just never know if it’s you, or one of the bajillion dependencies. I still had to have started doing tech dev in the 90s to even remotely know where to look when the software to figure out where the install disconnect was. I give them credit though, they’re doing a better job at corralling your eyeballs and attention in one place. It’s a big difference from, oh, let’s say TI, who has thousands of little islands of software and documentation. Their other benefit was there was virtually no firmware to be written. Everything was configured in a configurator, that once running was only moderately kludgey. If you ever needed to hand program hundreds of boards, it’s as buggy as any other SW-programmer toolchain. You have to restart the software a lot. It won’t recognize the tool. You have to reflash. I’m going command line on a raspberry pi as soon as I can ;). We will definitely post any code for tools.

The result, however, was worth the anguish. At first the new chips felt slightly floaty compared to the previous, but within a minute of playing, my plastic brain had already adapted to the new timing loop and it feels as tight as the last. I think the only people who are disappointed would be percussionists, but I guarantee all these knobs make up for that :smile:.

As a bonus, these pads work through plastic, so you can laser cut acrylic overlays with cool designs and patterns. I can’t wait to see what you come up with.

Bonus: This was what I was listening to when I got up and danced in celebration: https://www.youtube.com/watch?v=8YijrDu2PLU


#23

Enclosure Prototypes Arrive!

We’ve got the first prototype enclosure back from the CM, and it exceeds my expectations. It’s sturdy, roomy, well built. You can tell that the plastic has been folded instead of formed, and that the bosses and tabs are glued in place, but the gluing is inside the case, and the fold stress gives the corners a sheen I kind of like.

What you can’t see is something amazing. The cost difference between the first 100 units of a CNC + folding enclosure versus a plastic injection molded case is over an order of magnitude different. So much of open engineering and design is big ideas that start small, they need this ramp to get started cheaply. The other hidden effect is that small projects don’t ever have to get big. With a much smaller pool of users, you can just keep iterating and the changes in cost to the enclosure are drops in the bucket compared to re-tooling an injection mold. The real savings, though, is time. This method re-tools with a file. The injection mold would take weeks ~ months depending on iterations and prototyping.

Here are a few grainy photographs. We’ve already sent away for the next iteration, centering the console vertically, and widening the board (and the screws) by 2U (10.16 mm). The big gaping hole on the rear will be replaced with a bushing.

The top is going to be centered, and there will be a diffusing lens for the LEDs (They’re waaaay too bright).

Next Steps

Now that the new enclosures are here, we’re going to be ordering PCBs and parts for the next Beta round. We’ve been very cautious to release too many systems in the wild until the designs were more stable. Supporting a fragmented hardware base is something we’d like to minimize. I thank everyone for their patience. With these new enclosures, I’m sure it will be worth the wait.

We’ve got more surprises in store, so keep checking back!


#24

#We’re moving!
Our neighborhood is in hot demand, so we had to ditch the lease. Luckily we found a place with even more space! We’ve had to push everything back by about 2 weeks. Our assembly / test / debug equipment will be the last to go, and we’re putting together prototypes, but the video will be after we get into the new space.

#PCB1003 IRQ Trace bug
We found a trace that passes Design Rule Check (DRC), but shorts to neighboring pads. I think the finish of these PCBs is coming off during the reflow and allowing the traces to short. We have a simple fix. Luckily the translator board that makes wiring the system together has pullup resistors for I2C, so we get to keep using these boards, and don’t have to do a slow and costly white-wire solution.

#Test Fixture
Our ultimate goal is to create and release test PCBs with pogo pins, so we can automate the testing of the boards. For now, we’re testing everything by hand, but we upgraded the test station. We had previously been hot-gluing jumper wires to headers and swapping the cables, but our new translator board makes everything super easy. We cannibalized a bare PCB and wired up a quick fixture. This easily saves about 20 minutes per build. We’re really focused on all of our test jigs and manufacturing processes so we’re sure of a smooth pilot run.

Our next iteration of the test fixture will have pogo pins and the ability to tell if the pins on the chips are shorted or if the chips aren’t soldered correctly. We’ll be shipping before then, but we’re pretty excited to be sharing any techniques and scripts that we make.

It’s going to be a busy couple of weeks, but when the dust settles, I’m going to be happy to show you a great demo of the synth. See you guys in May!


#25

#We’ve moved!
It was a rough 30 days, but our R&D station is back online. The new space should be able to accommodate doing manufacturing in-house. We’re really excited and can’t wait for the dust to settle to take some pictures.

#Beta-2 Feedback
We’ve gotten a lot of great feedback on the Beta-2. All of the positive stuff is about the sound and how easy it is to build synth models. The negative points are mostly things like intermittent connection and glitchy systems. We’re going to revamp the power system, and upgrade the wiring harnesses. The system works really well the first time it’s put together, but as you open and close the system, things are getting loose. The cabling we’re working with seems particularly brittle, so we’re going to source stranded cable that can be moved around without breaking connections.

We’re also having ground plane issues with the capacitive touch pads. It’s not something that can’t be solved, but it’s back into science project territory.

#Announcing Beta-3 "Believotron Core"
We’re keeping all the hardware that is rock solid, but we’re moving the capacitive touch and the LED board to the experimental branch. We’re going to keep playing around with many different features, but we want to get more feedback on the software models.

Check back on June 15th we’ll have some specs and photos and beta-3 pre-ordering. Enjoy the spring / fall!


#26

#Free Laser Cutter Repair Technician University

It’s funny, the 80th time you’ve smelled a transformer boil off its laminate and short-out is a lot easier to debug.

It’s okay, I’ve always wanted to know how to repair a laser cutter :confused:.

When it’s back online, we’ll show you some of the cool DIY lasercut designs for the Believotron Core.


#27

Quick update:
We got the laser cutter up and running and have been rapidly iterating on the form and function of the boards.

We swapped out the original joystick for the same models used on the x-box. This caused a cascading change through the system because the pots are taller.

Luckily it means that we’re replacing the original potentiometer. We’ve found one that has 1M cycles, instead of 5k. Let’s hear it for 200 times improvement.

Everything we’ve put together, we’ve taken apart almost right away to keep making changes. We’re close to this latest shakeup being camera ready. We can’t wait to show you our progress!

Stay tuned


#28

The new beta hardware is up at the store.


#29

We have a video of the synth in action: https://www.youtube.com/watch?v=FuBrryA14pA


#30

We created a drumpad keytar and took it to our local full moon jam.

Before the demo unit was destroyed by an overzealous dancer, we got a lot of great feedback on the setup. My favorite part was when a 70 year old woman played the drums for 20 minutes. I was dancing to her drumming. She was drumming to my dancing. It was a hoot and a half.

Lessons learned:

  • Louder speakers.
    We brought a portable speaker that is usually plenty loud on its own. There was no competition at all against the cacophonous sea of a drum circle.
  • More drum pads.
    The kit we brought had 8 drum pads. Pretty much everyone wanted to put together more complex and varied rhythms
  • Better accent control.
    The velocity sensing on the pads is decent, but we found it was better to map out the amount of expression to a few pads and vary between them. This allows the rhythm of the player to affect the accenting instead of using the velocity sense. I’m thinking this is going to be a matter of personal preference, but we’re going to keep playing and see what we can do.
  • Synth noises sound great at a drum circle.
    The drums that mimic actual drums sound great, but the synth noises that generate sounds you can’t make on a drum sound incredibly good. The first time you throw in a space cowbell, it sounds out of place, but after about four measures, you’ll hear someone else joining in, and someone else still playing a counter rhythm.

All in all, it felt more like play than product testing. Thank you all for coming out and playing with the instrument.


#31

We’re currently doing documentation / video / photo production to demonstrate the features of the synth.

Until the units start getting out there, we’re building feature patches for demonstrations. If you have any suggestions, we’d love to craft some patches and record some sounds.

Because of the excitement at the Full Moon Jam, the next patch we’re doing is a drum kit.

We’re going to try out a cheap drum pad with a lot of buttons but no pressure sensitivity:

There is also a pressure sensitive drumpad at around the same cost. It’s got fewer buttons, but we’d like to see if the pressure sensitivity is enough to counteract having so many buttons.

We’ll post it here when we have the results.


#32

#Drum Pad Success

We were able to get full functionality out of the Novation Launchpad Mini. We’ve made a patch in our repo:

Here is a crude sketch of the CC#s for the buttons

We’re in the process of rewriting the drum objects to map to 64+ buttons. We’ll check back with a video demo.


#33

#Beta Hardware Update
We now have all of the hardware in-house to start assembling the beta orders. We took in more orders than we anticipated, so we might have to wait a couple of days if something is shorted, or we have poor yields.

The software is functional, but we would like to clean it up and make updates as we start to understand how people are using the synths.

Once we finish assembling and shipping, we’ll go back to creating cool videos of the Wanderlust in action. We’ve already completed a couple of demos using the Novation Launchpad, as well as a Demo for a generic MIDI controller.


#34

#Beta Hardware Update

I wanted to keep you up to date on the beta production. We received the bulk of the synthesizer core hardware and are assembling the daughterboards with the joysticks, knobs, and OLEDs. So far, our test fixtures are working, and finding production defects.
We are waiting on another shipment of PCBs that should show up today to balance out the orders.

We are, however, delayed because our laser cutter is failing and ruining enclosures. I apologize because we could have ordered these from a contract manufacturer weeks ago, but we wanted to cut and build the enclosures one-at-a-time to see if we could make any further design improvements. We’ve been troubleshooting it, but even with all the improvements we’ve made, it’s still spotty enough and is keeping the team from finishing the boards.

I’m headed over to a few shops in Chicago to see if I can expedite and get this material cut in the next few days and shipped by early next week. The worst case scenario is we put in an order with a contract manufacturer. I don’t expect it to clear before the holidays, but definitely no later than the second week of January. I don’t think it will come to that, but it makes me happy to have a back up plan that will work, albeit slowly.

I thank you for your patience while we breathe life into this musical instrument.


#35

Terse update:

All the beta hardware has shipped besides two customers in Chicago who have asked to pick up their synths in person.

If you have an issue, please contact us here. We’re closed for the rest of the day in celebration, but will be back tomorrow to work on software and push out all the updates and tutorials.

Thanks again to everyone who participating!


#36

Devlog: Finished Believotron Wanderlust Assembly Instructions.

Current: Refactor software repo.
Next: Getting started guide.

We’re going to be spending the next couple of days cleaning up the software repo and will be pushing a stable release this Friday afternoon (Chicago time). Feel free to comment or ask questions in the meantime.

Unfortunately we have to clean up the repo before finishing the getting started guide. I hope the stock patches we shipped on the synthesizers keep you all busy until this weekend!


#37

We’re midway through refactoring the code-base. Check out the commits on https://github.com/Believotron/Axoloti-Patches to see what’s being changed. We’re making unit tests for as many of our objects as possible, to help with the SQA regression testing.


#38

We knocked out refactoring and creating tests for /logic today.


#39

#Code repo update
We’ve done our first pass cleaning of all of our code repo. The files are not 100% tested, but the spaghetti has been turned into cous cous.


#40

Code update

OLEDs

We currently have a dev branch with choppy updating on 8 simultaneous bipolar values. The next step is to only update when the value changes to increase the update rate.

There’s a disconnect stability problem, so I suspect that there’s either a memory leak or a buffer overrun.

There is more work to do in OLED text formatting and making the GUI elements usable. It would be nice to separate the OLED GUI objects from the other parts, but it’s incredibly convenient to have one control loop to manage all of the peripheral operations. It might take 2 days to clean every thing up or 14. We’ll get a short video uploaded as soon as we think it’s stable.

If you have issues with the system, please post anywhere in the forums.