Your most interesting abandoned project

[Theo Verelst]

Abandoned means let go without completion, not never started or ended without any success, but actively or passively abandoned. My abandoned electronics project, directly related to synthesis, is a digitally controlled analog mixer, of the highest quality normally possible (parts with very high quality electronics specs).

 

It started humbly, with this (at the time new) tiny Arm RISC processor board, from analog devices, a demo board. I made a supply unit for it, connected a slider to the built in AD converter, wrote a program to average slider movements, translate the setting (when changed) to control commands for the analog volume chip with digital control input, and so I had a small board to control volume from a long wire without passing the stereo signal through the slider.

 

http://www.theover.org/Diary/Ldi22/HPIM1209b1.JPG

 

The idea was to get a more powerful processor, at the time a little linux board intended for security cams, connect up more than one stereo volume control chip, and some more esoteric things (e.g. analog/digital ground separation, automated controls, impedance adaptations, tricks for more accurate volume settings/calibration, etc), and feed that array to a quality zero input impedance mixing stage of the best audio chips on earth, pretty much.

 

Like this mix board I used for years, but broke because of supply line capacitors blew up the inputs of these chips (cost 25 euros each, so not replaced yet):

[video:youtube]

 

In this video I show (before my analog simulation digital synth is shown passing tthrough it) replacing the small microcontroller board with the linux board, connecting up with a PC to control the volume "live""

 

[video:youtube]https://youtu.be/AlG3OLNh6sE

 

This is the Tcl script

I use on the (any OS) PC, connecting over ethernet, and this is the enclosure I prepared:

 

http://www.theover.org/Diary/Ldi42/mix1.gif

 

So what part caused the abandonment ? A variety of reasons, mainly the cost of the system when fully made, like with many mix channels and sub/aux buses, both the audio chips I like for having sufficient wide range accuracy and noise poorness (OPA 627s) and the A grade stereo volume control chips it would be more than an experiment prototype. And some time ago I started using a Yamaha mixer, signals from audio and electronic sources (companies, trade shows, AES) weren't great about this target, and I started to pay attention to other things than perfect linear mixers, even though I'd still want one! Then there's the amount of work, especially when certain parts (like the digital drive part) are under constant siege to remain, sort of elusive. There was also the huge LED display driver CPLD that couldn't drive the required current.

 

So I started the "fake faders" array for a 4 channel stereo version, in a modular version:

 

http://www.theover.org/Diary/Ldi42/dsc00418bm.jpg

 

but didn't feel that much like designing electronics and wires like I did in high school already!

 

Theo V

[232006]

Please tell me you don't work as a "Wireman" for a defence electronics manufacturer, avionics or on nuclear power stations.....

 

Edit 1 - I guess you could do the control of the Volume chips with a $10 Raspberry Pi Zero W (Wi-Fi and Bluetooth) which have GPIO - inc SPI - perhaps you have already done this.

 

Edit 2 - As others have said Theo that is a great project - I do realise how much work must have gone into that, and please don't be offended by my silly joke about the "Wireman" - I know how prototyping is a hell of a lot of (seemingly) wild wiring!

[Markyboard]

That"s really cool Theo. I can proudly claim I"ve never abandoned anything close to that complexity.

[drawback]

That"s really cool Theo. I can proudly claim I"ve never abandoned anything close to that complexity.

 

+1

 

When I saw the thread title, I was thinking I might post 'learning how to play like Art Tatum.'

[stoken6]

I've started and abandoned projects for "versatile" looms/snakes, which could be used with a variety of rigs, but nothing approaching the complexity of Theo's project. I reckon 10 years ago you could have got the OFC/solid silver/directionally optimised hifi crowd interested in a remote volume control like the one Theo suggests. I assume now that power amps have a remote-control app...

 

Cheers, Mike.

[Theo Verelst]

Please tell me you don't work as a "Wireman" for a defence electronics manufacturer, avionics or on nuclear power stations.....

...

I do not.

 

That"s really cool Theo. I can proudly claim I"ve never abandoned anything close to that complexity.

Thanks!

 

+1

 

When I saw the thread title, I was thinking I might post 'learning how to play like Art Tatum.'

You'd need a good sound, or it wouldn't make much sense. Only recently I've been able to create sounds that at least make such an idea interesting, unless you'd use a real piano!

 

 

... I reckon 10 years ago you could have got the OFC/solid silver/directionally optimised hifi crowd interested in a remote volume control like the one Theo suggests. ...

 

It was about making a mixer, so the shown rack unit would have enough power to use all those balanced Jack ports at the back to automatically mix all your keyboards and effect equipment. In fact I also created a board where I soldered one of the smallest footprint DIL for a digitally controlled gain mic preamp of very high quality, and the pre-amplifier board I showed in the YT contained a test circuit for the Phantom power protection. Another reason for abandonment, because the ypical application circuit in the BurrBrown/TexasInstruments datasheet for that chip caused way to much audible distortion, albeit at non-mic levels.

 

So actually, it would still be the perfect rack mixer with lots of Open Source possibilities to control the mix over Midi or Ethernet, regardless of the processor used. At the time when I started this over 15 years ago, Linux boards were not common, and the use of FPGA chips (for the translation of some logic signals) neither. Now a little shorter ago, I wanted to use a extensive Zynq FPGA chip with dual core ARM 600 MHz to drive as many of the voolume chips I managed to receive as free samples (more than a few). *THAT* got abandoned because the chip was going to create like a hundred control signals all at a rate above the audio rate, but the makers of the particular "parallella" didn't really pursue things like adapter boards and programming the right pin numbers as something easily usable, and the multitasking Linux constructs require more work than I was willing to put in. Dozens of inputs and various mix rails would still be cool in that 1 unit height rack unit, with almost complete signal transparency, even for high Q equipment.

 

T

[232006]

Please tell me you don't work as a "Wireman" for a defence electronics manufacturer, avionics or on nuclear power stations.....

...

I do not.

 

 

 

Hi Theo,

 

I don't know if you saw "Edit 2" in that reply I made, but I genuinely didn't mean to offend you, it was just a silly joke and I hope you didn't think it was a "dig" at you.

 

It certainly wasn't intended as such.

 

I am a tech guy myself and do appreciate all the thought and work that goes into prototyping.

 

Kind Regards

[David Emm]

Its more "on ice" than abandoned, but I keep being mildly annoyed by grinning over the idea of rendering Talking Heads' "Psycho-Killer" for solo pipe organ. The idea seems easier because it would be a Full Swell thing with very few chapel organ flute stops in evidence. Hulk smash puny flute stops! Its been growing in my thoughts because I feel as if I already have the bass pedal part worked out. Its too esoteric, potentially a neo-prog eye-roller around the edges and I've never had a pipe organ *lesson* in my life. I've played one and loved getting the feel of it, but a candidate for Hauptwerk I am not. It would/will be an interesting COVID project, though....

[allan_evett]

I've been planning to record Chopin's Prelude in Bm as a re-imagined synth arrangement, for a couple of years now; but it hasn't happened yet.

[Mighty Motif Max]

A large heavily-orchestrated keyboard-heavy Christmas album. Every time I worked on a piece, something would come up, and I'd come back and listen to it later and decide it wasn't good enough. Never happened and my earliest song that was supposed to go on the album is from 2015. Such as it is with artistic growth I guess, but I couldn't ever see those going on the album if I ever do finish it, as they're not up to the current standards. Which is exactly what happens every time I think I've finished a song that's high-quality enough to be released on that Christmas album lol.

[Groove On]

I've been hacking away at a 61-key midi controller with weighted or hammer-action keys. Turns out the MIDI controller electronics + case are relatively simple, but coming up with a good key action is the challenge and also affects how I implement the electronics. That's why this project is stalled. I could just copy some standard digital key actions, but my gut tells me to keep searching for an elegant "travel keyboard" solution. Something that's simple, DIY, and easy to troubleshoot / fix.

 

In the meantime, I ordered an inexpensive 50 watt laser engraver so I could play around with different case designs. But that key action design is always floating around in my subconscious - really just want to build my own keyboard that I can maintain myself.

[Theo Verelst]

Unless you use an off the shelve solution, there is some difficulty in making and programming a keyboard scan solution with proper Midi implementation.

 

T

[Threadslayer]

Unless you use an off the shelve solution, there is some difficulty in making and programming a keyboard scan solution with proper Midi implementation.

 

T

When you say "some difficulty" are you referring to such things as velocity sensing and after touch, or just the (relatively straightforward) MIDI note on/note off scanning?

[Theo Verelst]

Bringing up the micro controller, implementing some scan logic, providing accurate time measurement for velocity sensing, getting a serial connection to properly form Midi messages, multitask between key sensing and Midi message buffering, deciding on overrun and ordering priorities, and making the whole, including wiring, work reliably.

[Threadslayer]

As for working reliably I'd think the mechanical challenges in building a playable and reliable keyboard would be a lot more exacting than the electronics and firmware, especially today with free online resources like This: to get you started.

[rogs]

Back in the day - around 1980 - new polysynths like the Prophet 5 cost around half a year's money here in the UK.

 

I was working as a keyboard tech at the time, so I thought I'd have a go at making one.... I bought some Curtis 3340 oscillator and some SSM 2040 filter chips (easy to find and cheap in those days!) - located a 61 note keyboard from an old home organ, and had a go at at building a 5 voice polysynth.

 

I did get it all to work - although it was a devil to tune - and I never finished the case construction ...All long since relegated to the great synth yard in the sky...

 

Sadly, I don't have any photos of the project -- never thought about taking any at the time - All I have is a few snippets of cassette recordings from the time.

 

The recordings were made with home made sequencers and analogue drum and 'hand clap' machines as well....

 

All sounds pretty awful now, but at the time I was quite pleased (making these types of sounds with commercial kit was for the wealthy few in those days! )

 

Short clip here: www.jp137.com/las/polysynth80.m4a I warn you, it's not good ...but it does bring back some happy memories of that time for me....

[elif]

Abandoned/shelved projects. I have had a few. Here is one:

Leslie Simulator

 

Some time in the 90's, being dissatified with any simulator I had heard, I decided to build one using a time-varying FIR filter using the measured impulse reponses of a real Leslie for coefficients. I recorded a set of impulse responses of my Leslie 122 (in my living room). The set for each rotor were measured at fixed azimuth angles of 22.5 degrees, giving a set of sixteen for each rotor.

 

I came up with an algorithm that would work, but that would be too computationally intensive for any single DSP available at the time, and probably even now. That project and those coefficients were shelved until 2017 when I did an Octave simulation of a Hammond chorus/vibrato that sounded pretty decent to me. Could the same algorithmic approach work for the Leslie sim?

 

I retired from the day job in Jan 2018 and worked for three months on it. I modeled it in Octave, then I ported the algorithm to a TI TMS320C1648 low cost development kit. It worked well, but the audio output of the LCDK had some buzzing artifacts. To fix that, I started laying out a DAC board with some peripheral inputs to attach to the LCDK.

 

Then I bought a Ventilator II.

 

Hey! The Vent sounds like my sim! But the Vent has nice hardware and lights and multiple I/O and lots of options and customer support. In general, I don't build things I can buy. So far, I haven't thrown the sim out but it has been shelved indefinitely. Though, maybe an FPGA solution with the original algorithm...

[Theo Verelst]

FPGA's are interesting because they can give you a lot of power and if you know how also incredibly low latency. And nowadays the Silicon Compilation is in principle making it possible to take a function in C source code, and automatically implement it in the FPGA, up to quite high efficiency, depending.

 

T

[RudyS]

Project 'Wendy' during high school. Just bumped into her the other day. Wish I finished what I started...ah well, story of my life:)

[Theo Verelst]

I almost abandoned my Amazon Cloud Services FPGA (F1 node) experiments, but I just changed to a cheap dev node and couldn't resist connecting to the remote desktop to start the latest Xilinx example Vitis project, to see if maybe by now versions are compatible, after having (re-)established FPGA running permissions and such. I should be cool to get my mathematical formula sound rendering software out of the fridge and run it on super fast programmable logic...

 

TV

[stoken6]

Project 'Wendy' during high school. Just bumped into her the other day. Wish I finished what I started...ah well, story of my life:)

Ah, well, we've all got projects like that.

 

Cheers, Mike.

[Groove On]

Unless you use an off the shelve solution, there is some difficulty in making and programming a keyboard scan solution with proper Midi implementation.

 

On the electronics side, I"m using an Arduino. Their MIDI library converts serial data to MIDI, so I just needed to build a MIDI Out port. And for MIDI over USB, I selected a micro-controller that is USB class compliant. Once those requirements are satisfied, implementing MIDI on the Arduino is trivial. Not to say there aren"t still serious challenges but that part is neatly resolved by the Arduino and running nicely on a breadboard prototype. I also have a Bluetooth Module to implement BLE-MIDI - but I haven"t got around to it yet.

 

If I was just building the typical push button / potentiometer type of MIDI controller, I would be done already, but as you pointed out there are challenges with keyboard scanning. And there"s the even bigger challenge of thinking up an elegant design for the mechanical piano keyboard that is easy to build, maintain and fix.

 

I decided to skip the traditional switch & matrix design found on most digital keyboards and instead will use DIY-friendly infrared optical sensors. It"s the same type of design as Cybergene"s Grand Piano project over on the Pianoworld forum. With an optical sensor + comparator to calculate velocity - also running nicely on the breadboard prototype (but not without its own issues).

 

Floating in the back of my head is a crazy idea to power the keys with a push-pull solenoid for both resistance and sensing velocity. I keep saying it"s completely nuts, but another part of me keeps saying - 'but damn it, it would be so cool'.

 

Back to the mechanical piano keyboard, which has stopped the project, I have figured out a nice assembly line method in my workshop for making wooden keys, which can also be used to cut hard PVC sheet, so I can pump those out pretty quickly. But I"m unsatisfied with the traditional key stabilizer pins into a wood bed. It requires felt / Teflon which means maintenance similar to an acoustic piano - argh. So I"m hoping to find a better low maintenance way to stabilize the keys that"s more friendly for a 61-key MIDI 'travel' controller.

[Groove On]

So what part caused the abandonment ? A variety of reasons, mainly the cost of the system when fully made, like with many mix channels and sub/aux buses, both the audio chips I like for having sufficient wide range accuracy and noise poorness (OPA 627s) and the A grade stereo volume control chips it would be more than an experiment prototype.

Theo V

 

How long ago did you abandon this project?

[Theo Verelst]

Well, it's been of and on until I guess a decade ago, parts worked, quality had been assessed, I used a mix setup intended for it until recently it's ESD resistence was superceded by hot switching an input cable... I got into heavy Linux Jack based Dsp using long buffers so good I7s can actually perform some work with their supercomputer architecture parts, which lets me into high grade studio signal ideas. I use a (analog) Yamaha mixer and don't really need the rack unit, so sometimes I use parts for expeiments, but I moved on.

 

T

[RichieP_MechE]

Several years ago I set out to improve the design of the K&M 18880 stand, making the expansion mechanism all rotary joints instead of rotary + sliding, and all adjustments could be made without tools. Never made it beyond the 3D model. The power strips in the model are a placeholder for a 2U rack.

[EscapeRocks]

While not abandoned, I haven't turned a wrench on this in 5 years.

 

1948 Chevy 3100 p/u

 

 

Eventually it will be great for hauling gear. Eventually....

[SteveCoscia]

Using the piano to convey customer service concepts seemed like a worthwhile project. I created three videos and they went nowhere. Project aborted. Oh well.

 

[video:youtube]

[mate stubb]

Probably for me it would be the SuperMoe 16x4 step sequencer project. It was an homage to the original E-MU modular sequencer, with more control added. Panels were designed and manufactured and circuit boards were laid out. Then I abandoned it and sold my big modular to fund building my Hammond midi controller.

 

The only photo I have is of the main pot matrix modules, seen in the left cabinet here.

 

http://hotrodmotm.com/images/case/sod_biggest.jpg

 

Here are renderings of a couple of the larger modules.

 

http://hotrodmotm.com/images/addrgen2.jpg

 

http://hotrodmotm.com/images/CV_Generator_8x4new.jpg

[Theo Verelst]

Then I made stuff that could probably be considered the PhD version of the "Big Data" movement, and... largely put that aside too!

 

From others' examples and my own a bit as well it could be maintained abandonment isn't a bad idea sometimes.

 

T

[Groove On]

Probably for me it would be the SuperMoe 16x4 step sequencer project. It was an homage to the original E-MU modular sequencer, with more control added. Panels were designed and manufactured and circuit boards were laid out. Then I abandoned it and sold my big modular to fund building my Hammond midi controller.]

Nowadays, some people use kickstarter/patreon for funds. This guy built a 1000 oscillator synth using donations from people who just wanted to see it built. All the people who donated have their names written on the knobs.

 

[video:youtube]