Ever since I started messing with computer music and gear I've been looking for the perfect MIDI controller. 15 years later, it still doesn't exist. My idea of the perfect MIDI controller is as follows:

Controls

• 64 high-resolution smooth (i.e. not detented or notched) push-encoders, like those of the BCR2000 but with metal shafts and panel nuts to ensure encoders are firmly secured in place.
• Tapered encoder caps (turn by the bottom for more precise movements).
• Generously spaced encoders to prevent accidentally knocking adjacent encoder caps.
• 16 backlit Launchpad S style buttons on top face (fixed velocity).

LEDs

• Encoder LED rings supporting a number of modes (level, pan, dot, two-dot, etc), just like on BCR2000. At least 15 LEDs per ring.
• Large LED under each encoder associated with the push action.
• All LEDs RGB except encoder rings which are white. Encoder ring centre LED is also RGB to function as visual centre detent, like in DJ Tech Tools MFT.
• All LEDs accept feedback from host e.g. Studio One, 0Live, Traktor, etc (or any software that supports sending parameter values).
• LED double-buffering like Launchpad S (for fast bulk LED state updating)
• Per-LED special modes e.g. various blinking rates, glow, rainbow, etc
• (Not included in my 3D model) A small numerical LED 2-digit 7-segment display to display active preset number, but this could of course be added.

Software/Firmware

• A decent software editor.
• Programmability of the Behringer BCR2000 (this thing packs a lot of functionality under the hood).
• At least 32 banks that can be switched using the aforementioned back-lit buttons, as programmed by the user.
• Ideally open-source so users with programming skills can customize and contribute to the product.

Mechanical:

• Made from tough & durable injection-moulded plastic as seen in older IBM Thinkpads, not the Mickey Mouse crap Novation use in their gear which scratches at the touch of a feather.
• Fillets on all edges to make the controller harder to scratch (unlike Novation's latest product range which is nothing bug sharp edges)
• No paint, all surfaces are raw plastic (unlike Novation products) to make it more durable.
• No rubberised parts as the rubber finish always degrades after a while, except encoder caps (see next point)
• Easy to replace encoder caps.
• Clean design of DJ Tech Tools MIDI Fighter Twister.
• USB connector on rear face (not on the side like Novation Launchpad which is basically stupid).
• Detachable rubber feet.
• No concealed screws to allow users to service their hardware easily.
• Stainless steel Allen screws throughout.
• Main shell height not exceeding 30 mm (i.e. not including encoder caps)

Price

• Somewhere in the 300-400 USD range, otherwise few people will want to buy it.

So... that's a long list right? Since this thing doesn't exist and nobody seems interested in making it, I'm guessing there's an opportunity in the market.

I'm in China so I should be able to find affordable suppliers. Might need to go the Kickstarter route especially for getting the moulds which is usually the most expensive part.

Now, I don't have all the skills required to pull this off, but working with a couple of guys I know I could assemble a team that would encompass all the required skills—software, firmware, electronics, industrial design, documentation and marketing.

Before I start wasting time on this instead of making tunes, I wanted to ask here if anyone would actually be interested in this thing. Below is a mock-up of what it might look like. The encoder caps could be longer. Changing encoder spacing is pretty trivial as the 3D model is fully parametric (I just type in the new value and it's done).

I love the notion of a flexible surface with lots of physical controls to grab. I'm less convinced that rotary encoders are perfect for every task, or that undifferentiated arrays of controls are ideal. To me, a more ideal layout is one that offers some memorable structure – like the way Modal clusters smaller rotary controllers around larger ones in groups. I'd gladly trade fewer inputs for a better layout, and perhaps a small number of sliders as part of the mix both for memorability and suitability for purpose in a handful of cases.

Well, the LEDs are color-coded. This controller concept was mainly geared for DJing. I use a DJ Tech Tools Midi Fighter Twister to control the mixer in Studio One and it works well, with LED feedback and all. I use the push action for Mute. This gives me access to 16 channels, or 64 if I use banks. No parameter jumps, and the thing measures just 15x15 cm.

Unfortunately it has no top-facing buttons for switching banks, that's why I came up with this idea. Maybe a 32-encoder design might be better.

Ableton Twist

I'd consider using pots, then on power-up, the actual pot position can be read and used instantly. I'm not a fan of encoders for real-time MIDI control.

what do you mean by: on power up actual pot position can be read and used instantly?

with encoders you can read the position at all times and use them any time without parameter jumps or unnatural relative curves.

why are you not a fan of encoders, i see no reason in that.
maybe you never tried them or had only bad ones?


btw a new pot controller, just how you describe it, is now available from Faderfox

there is no way to make this happen for 300-400 USD...
well, there is.... but then we end up with a toy....
we already have enough "toys" to choose from

i would pay 1000 or more for a high quality device like this, the controller market really lacks high quality
ppl pay this sum and more for instruments, this would be my main dream controller, well worth it.


it has to support OSC, I'm not even sure if midi can handle all of this in a nice and smooth way.

the led rings have to be rgb as well, that adds nice orientation capabilities

i see no reason for the arcade buttons on the left and top, keep version 1 as simple as possible.

the faceplate and inside structure has to be made out of metal imo.

a nice software editor that works on both mac and pc and is maintained for newer OS over time is a highly desirable feature.

my 2 cents, love your idea

Top facing buttons are needed, otherwise I would just buy a couple of Midi Fighter Twisters and not bother with this. Another idea is to use regular pots instead of encoders with two LEDs above each knob to indicate which way it should be turned in order to catch up with the software. This would rely on the software transmitting midi feedback. The media controller would then keep a record of all possible midi parameters, that is, channels 1 to 16, 128 parameters per channel, because the host only transmits parameter values when the parameter changes, this way the controller can retrieve the last received value of any CC parameter by reading it off it's own memory. There would still be situations where parameter jumps would occur, but these would be significantly fewer than if soft takeover was implemented entirely in the software. Encoders also have drawbacks, and are less tactile, but are more convenient because they are always synced with the host. I'm not sure why no manufacturer uses them, I suppose they are more expensive.

they are way more expensive and a hassle to make them work smoothly.
Akai has the best ones i tried so far, the space between knobs and the space between knobs and led ring is perfect.
the led rings and encoder curve works excellent.


in your rendering, the big leds at the bottom of the encoders could be made as buttons as well.

The ADC on your chip can read the voltage from the pot at startup which provides its actual position.

Encoders just indicate relative movement, not actual position.

I've been making MIDI gear for over 10 years and use encoders for user interfaces, just not realtime control. I like to be able to look at the pot and know where it is, and you have a lot more control with a pot, as most encoders only have 24 detents. With a pot and a 10-bit ADC you get 1024 positions in only 270 degrees of movement. With 12-bit, quadruple that.

Have you ever messed with a Behringer BCR2000? The encoders in it seem cheap—the shafts are made of plastic and there's no panel mount thread or nut—but they appear to have very high resolution. For example, using 'BC Manager' I was able to set to map the full CC value range over approx. 270° and still traverse the range in steps of 1 if I turned the encoder very slowly. Is it possible the BCR encoders have more than 24 steps? TBH the BCR is the best implementation of endless encoders I've ever seen. I've used much more expensive devices e.g. Livid Instruments where encoders would skip or transmit the wrong direction when turned fast, this never happens with the BCR.

Another idea that crossed my mind was to use some kind of haptic feedback to emulate a physical stop. This would only work well when turning only one encoder at a time, for obvious reasons. Still better than no feedback at all. It would be nice if there was an encoder with an electromechanical stop that could be controlled in software; that, combined with higher resolution, would make such an encoder practically indistinguishable from a a pot.

Incidentally, considering most MIDI-capable software and hardware use CC for MIDI control, do you find many situations where you can use more than 7-bit of resolution? (I'm aware of pitch bend etc, but I'm curious). Admittedly, 7 bits is too low by any standards, but that's the way MIDI was defined and now we are stuck with it. I'm not aware of any hardware that supports OSC natively (feel free to educate me), and it's going to take a major player to catalyse widespread adoption, for instance, if Microsoft decided to build a 'Generic OSC Driver' right into Windows, that might encourage third-parties to embrace OSC. I don't even know if that's technically possible.

I really hate programming VST Synth patches with mouse & keyboard. This would be a dream come true.

definitely interested, but should be able to also support NRPN for higher resolution. And while wishing for ponies: OLED scribble strips for each encoder would be plain awesome!
I'm still on the verge of getting a Mackie C4 but unfortunately I mostly use Live 10 which still does not support sysex

How many characters does MCU support per track on its scribble strips? The problem with fancy displays (RGB, etc) is that no host supports it natively, it's either MCU or ugly third-party wrapper for which support could be dropped at any time (think NI Kore).

I would be really interested in this - if it was built better than the BCR2000, I'd consider buying 4 of these (provided using multiple controllers with one computer would be supported)

I'd love to see a small display that shows the value of the currently turned encoder - even better would be an oled display and the possibility of creating some custom text output. So say you have assigned one dial to change oscillator shape, it could say SAW, TRI, SINE etc. depending on what value the encoder was at.

Also, being able to send MIDI to the controller, without it being fed back to the output would be awesome. Meaning that in Reaktor, I could set the output of an LFO to the MIDI CC value of one of the encoders. I could then see the LED bounce back and forth at the frequency of the LFO, wihtout creating a horrible MIDI feedback loop.

I'm suprised there aren't more of these types of controllers out there. If you play live, and change presets in a synth, with regular encoders, looking down on the controller gives you no indication of what the values are. With encoders with LEDs, you could instantly look at the controller and see that for example, LFO1 is set at a high speed.

I have one of these Doepfer controllers from the late 90's. Its also an 8 track midi/Analog sequencer. Its a fantastic unit which I got it mostly for the controller although the sequencer is fun to use and produces quirky results. It's superb for all controlling duties and feels like a real piece of gear.

However, I wouldn't be averse to a new model with slinky smooth faders and a couple of rows of knobs. The faders on the Doepfer are decent but not ideal for slamming in 8 faders during the heat of the moment. A little to much resistance.

I searched a long time for something like this. Didn't even know it existed until I saw it for sale on Ebay. The world wants another one of these. I would pay up to a grand for good quality.

Your proposed project would need a display imho. I cant remember what four pots are doing let alone 64. Ever thought about going modular approach? 8 channel modules, faders and pots and buttons - buy as you need them. Good luck with it, hope it brings you great success





And this one is so so close. I need 24 faders minimum though so would have to buy two.

I'm sorry I forgot to answer your question: imho it's 6 characters. I could probably live with that

DJ Techs rubber knobs and sliders were meant for DJs, but I automate lots of gear and lights from a Physis K4.
Occasionally sweat gets slippery but the Rubber Chroma stuff is a lifesaver in between solos or scenes.

I took a bit of a look at pricing some options out on this as a DIY thing using alibaba. It looks like doing it with led rings is pretty expensive. OLED displays actually end up being cheaper funnily (at like $2ea including shipping). Would love to see what you think a mockup using LED displays instead of rings would look like. The OLED boards around 1in square.

imo oled would be fine!

If the topic is still of interest: I have develpped such a device. Prototype state only, but could be multiplied if people are interested in it. Also others have copied the idea in the meanwhile:
https://www.mikrocontroller.net/topic/284005

The idea behind that was a high precision control device for industrial apps, finally also beeing appropriate for music.
I intended a 8x4 matrix, totally configurable with dynamic data representation by the RGB-LEDs, with a layered display using TWO LEDs to indicate the position like hours and minutes on a clock. With three colours you had a decimal system with up to 30x30x30 positions.

http://96khz.org/htm/midicontroller31.htm

Can you briefly explain what the project is exactly? I'm not sure I understand. Is this some kind of modular kit?

I would love to see some kind of modular MIDI DIY kit including a main MIDI board with loads of I/O and a bunch of modules that connect directly to the main module using i2c or something like that. All the user needs to do is hook the modules up to the main board and make a box with holes to hold the bits create a custom layout.

While there are commercially available MIDI I/O modules, these only support pots, switches and single LEDs—no funky LED rings, LCD displays, endless endless encoders, etc, like the ones shown at that link you shared.

There's also the Special Waves' Mine S but I personally would be very more interested in a modular DIY MIDI kit consisting of a main USB-MIDI and modules that connect to it over i2c to simplify wiring. This way all modules need just four wires (SCL, SDA, Vcc and ground) regardless of complexity. For example, you'd only need four wires to connect an encoder+led ring module. Ideally some kind of jump wires would be used, no soldering needed. The kit would require no programming skills, and MIDI assignments, LED colour, etc would be established using some kind of user-friendly software editor.

I know about the "MINE S" The point of configurable hardware is usually the price. No idea how much is such a piece.

My Controller in intended to have a fixed configuration. Regarding programming: I also thought of I2C (you mentioned that, "SDA" "SLC") but finally went with shift registers directly driven by an FPGA. This is easier for me. In the protype, there is one value from (in my case) -4095 and 4095 and another one as code which describes the way of displaying this value. So to speak the strategy. Both are written in one of 32x2 registers. Typically only the 32 value channels are updated during movement. The FPGA controls the values of the LEDs by PWM then. With e.g. 768kHz (to be in synch with the synth) one has an update rate of 750Hz. This is enough to let the LEDs flash in accordance with the beat if necessary. The FPGA is a isolated extra piece and only receices information from the synth, like the current channel ampltude via S/PDIF. The bill of material was about 500,- (15,- / per Channel). Including a large hammond case.

In a later approach I used a WS2812 LED Ring which could lower the cost to e.g. 12,- per channel.

As said, another guy started to produces something similar in that thread so I did not continue to collect interested persons anymore.

Currently I am designing without feedback LEDs and use a VGA Monitor instead. In the VGA both values are shown: 1) The setting of the knob and 2) the impact oft it, e.g. the resulting amplitude of the wave. I also use digital numbers in some configs.

Anyhow I am still not confident with what I can have and what I can buy. A controller with only encoders and a suitable setup is hard to find.

In fact I need at least 54 Encoders to control all the parameters of a voice / part of my synth, (16+8+4) to control the mixer output and at least 13 to control the generall params of the synth ("globals"). And i need decoders (not potentiometers) with differently programable parameter ranges to by able to cover the desired values. Therefore the fader port is not an option.

The only thing really suffient at the moment is the MIDI figher, but this also would be 15,- Euro per knob. Behringer BC2000 had 24 encoders with 10,- per channel. But is out of sale and also the quality was not sufficient.

Another Idea was to use something like a "knob-only" launch pad using 2 columns for "up/down". An AKAI would give me (8/2=4 channels + 1 fader ) x 8, leading to someting like 2,- / channel. But I found no way so far to chain them directly and convert to standard midi to directly use it in my synth. It requires a MIDI-host to MIDI converter für each of the devices.

If there were enough people I would design a board for 4 of such encoder peaces with an LED-ring option to mount if desired and use a daisy chain to communicate with an FPGA.
This could be upsized up to 16 x 16 channels driving 16 classical midi channel with 16 virtual channels each + my fast MIDI via S/PDIF (1024 channels).

We then only need a VST driver to transfor S/PDIF back into a classical 16 MIDI channel input inside the PC to provide this for the DAWs

This causes issues when swiching the controllers to another target.
Generally one would not like to change any value in a synth just because moving the focus of the controller to another synth channel.

The best would be xy-pot having to rings which can be interpretated the analog way. I have a setup for such a controller and you can get nearly the same feeling as with a real pot but do not have an end position. But this pots are expensive.

The alternative way are magnetic / hall sensor based encoders. These are very smoothy and easy to interpret but even more expensive.

I thinks common encoders with 30/12 or 24/12 are the cheapest solution. I got one lot of 100 of them for just 120,- having pre soldered connectors which make it easy to replace a defect device.

News:

http://pyratone.de/htm/diymidikits.htm

This is to full version with most of my MIDI Functions wich should make it possible to connect all gear including old MIDI Controllers and (via host) also modern USB-Controllers and mix / direct them in realtime. Inlcuded a sequencer and effect.

64 channels as demanded above can be achieved with two devices linked via S/PDIF.

Since this dream of mine I've almost entirely given up on MIDI controllers because it's a constant fight where it feels like painting a house through the keyhole. In the end the mouse is more efficient. If I need to record a couple of parameters here and there I just map them on a one-off basis, do the thing, then move on. No more time wasted creating custom mappings and fighting DAWs that were clearly not designed to work well with MIDI controllers.

Exactly what I think. DAWs only want you to go with 8 knobs, wheither what you control is continuous or discrete stepped values or on/off switches.
I found Reason/Remote with Lektar Panorama controller is the best ever combination but it seems nobody cares about that kind of remote control since there are not other controller in that league.
And it’s not going to change. I don’t remember reading anything about remote control in the CLAP announcement for example.
So mouse, 8 knobs or a hardware synth !