rotary encoders

[ian-s]

I have been thinking about synthesiser design lately, specifically the use of rotary encoders.
In the dawn of the age of “digital patch memory” synthesisers, the usual strategy was to use conventional pots which were read by an ADC and microcontroller. The values were saved in digital memory so that they could be sent to a DAC and Demux when the patch was recalled. Some designs offered a live mode where the pot could be switched into a direct connection to the circuit it controlled, like filter frequency. The Korg PS3200 used push switches on the pots. Others simply detected the change in value as the knob was moved to enter into ‘live adjust’ mode. Both these systems would result in a likely rapid jump as the value quickly changed from its memory, to the actual physical knob position.
I think one early attempt at a solution to this, was to require a knob to be moved to the stored position before switching to live mode, less than ideal.
With the advent of continuous rotary encoders, the problem of a fixed 330 degree rotation was overcome. It was now possible to make an adjustment relative to whatever position had been recalled. A new problem however is that the knob cannot have a conventional pointer. The LED collar found on some equipment is designed to solve this last issue.
The LED collar does look impressive, at least till you move a knob and notice the jerky, coarse steps. Perhaps the next generation will use a higher number of smaller LEDs in combination with visual anti-aliasing to improve this. Maybe it is just me being too picky.
Anyway, the point to all this is that in my new synthesiser design, I would like to use rotary encoders without LED (or other pointers). One knob dedicated to each function.
I think part of the panel playability of early analog synthesisers, stemmed from the fact that each function had a dedicated control. As the player became familiar with the panel, rapid and expressive tweaking became a matter of muscle memory.
Under these circumstances, what is the point to a (knob) pointer? If you want to increase the rate of LFO2, you grab the knob and turn it clockwise, right?
As I recall, in the pre patch memory days, some people would rely on the knob pointer positions, for laboriously recording a patch to a blank patch sheet, so they could hopefully ‘recall’ it later. But with instant save and recall to memory, I can’t think of a situation that would require visual feedback of the knob positions.
Ditching the LED collar would lower costs (because I’m thinking a lot of knobs) and I could be more selective, choosing a control that has that nice smooth, liquid quality that I like.
What do you think?

[jksuperstar]

I like encoders, but I think I still like some feedback to as to what the value is, besides waiting for the sound to come and give me audible feedback. Sometimes timing an envelope step to an exact value, or an LFO rate, is very useful.

I suppose if it came down to cost, you could have all the knobs in the world, then have a display (text or graphical) to show more exact editing parameters.

What about surfaces and sensors that are beyond knobs? Or Knobs that ksense that you touch them (to bring up that info on the display, perhaps?). just thinking out loud...

[ian-s]

[DrJustice]

More thoughts:

You get rotaries with a momentary push button function. This could be used this to bring up the value in a central display.

Depending on the rest of the design, marker lines around a rotary can be redundant visual clutter (or very stylish...).

DJ
--

[blue hell]

Or have a follow/steady toggle next to the display,when steady it keeps its current value when in follow mode it displays the param name and value for the last knob move. I think good rotaries a hard enough to find without the push action ?

[mosc]

I like the rotary encoder. Sliders are cool but they must be motorized if you are going to get the same preset changing control as the rotary encoder. I would like a rotary encoder that sits on a motorized linear fader. A momemtary on/off button on the top of the rotary would be great. Better yet would be a rotary encoder (with sitch) on a motorized XY plane.

[Tim Kleinert]

I also think that encoders are where it's at, conceptually. Personally, I wouldn't absolutely need LED collars, although on the NordLead3 they give excellent feedback of any kind of morphing that is going on, which nice. A big central 7-segment digital value display, well readable even from a distance or in the dark, would do the job too.

My gripe with encoders is that I have never encountered one that really lives to my expectations of quality. If we are talking of synthesisers as musical instruments, then the longterm durability of the controls is paramount. And here is where all encoders fail. I've worked with equipment from different manufacturers -Akai, EMU and Yamaha samplers, FX units from Lexicon and TC, digital desks from Yamaha or Mackie. All of their encoders became dodgy within a few years of normal studio use. Same for the NordLead3, and I guess for my G2s within one or two more years . That simply isn't good enough in my book. I want encoders that give me a minimum of 10 years trouble-free daily use!

I know too little about the subject to know if such encoders simply do not exist, or if they do, but are too expensive. I'd be interested to know more.

[DrJustice]

Tim, I agree with you re. the quality of encoders as used in musical gear. Very often you'll find a cheap (around $5 or less) mechanical encoder which typically have a specified minimum lifetime of around 20000 rotations. Some examples from the ALPS selection at ELFA (available with push buttons BTW), here and here. There are datasheet links besides the prices.

If you pay a good bit more (in the range of $20 or more) you can get an industrial quality item with a lifetime of several million rotations. Here's an example of a Grayhill optical encoder with a million rotations. Digikey sell some Grayhill models.

So it's the same old "you get what you pay for". Buy a CAT scanner and it may come equipped with Grayhills...

DJ
--

[Tim Kleinert]

[blue hell]

[Antimon]

I like the expressiveness of the mod wheels, like on the Yamaha DX7 and KX88. It would be cool to have a line of 10, 20 or more wheels... And you could make them endless! Perhaps with simple led rows on the side showing the position. It would be a bit easier to to move several controls at once

You can kind of throw mod wheels, so their inertia keeps them rolling. If you had an endless mod wheel with a huge range and resolution (say, 65536 steps), you could keep rolling, rolling and make really smooth transitions.

Or you could use trackballs! A trackball isn't much more than a two-dimensional endless modwheel. A synthesizer with 20 trackballs above the keyboard would look really cool. Hard to keep clean and stable, though.

OK, I'll stop now.

/Stefan

[mosc]

[Tim Kleinert]

Yes -trackballs are really an original idea, and might work very well, although one would have to get used to them. But since they are XY-controllers, one would also have to find meaningful couplings of two parameters.

[Antimon]

Thanks. On my computer, I use optic trackballs instead of a mice for ergonomic reasons (it helped my arm stop aching), and it works great for most purposes, especially in the G2 editor. When you press a knob to tweak the value, the mouse pointer won't move when you move the mouse to change the value, which means you will have to eventually lift the mouse back in center to continue work in the editor.

With a trackball that's as easy as lifting your finger back on top of the ball. And you can throw it (it keeps rolling after you release it) to maximum or minimum easily. Trackballs take a bit getting used to, but they're worth the effort, I think.

I imagine the price tag for a 20-trackball controller to be a bit too staggering for me though.

/Stefan

[mosc]

I use the Logitech Cordless Track Man Wheel.



It is great, but it is made only for the right hand. When the USB is connected to the computer, OS interprets the mouse and key clicks. I guess it would be pretty simple to write an application to generate MIDI. There are two switches and a scroll wheel too.

It would be great to have one one the right and one of the left of the keyboard on my G2X.

[jksuperstar]

I picked up a few of these a couple weeks ago, been playing with mice, trackballs, and other ps2 devices lately as controllers. I'd like to add something like this to a wireless USB game controller, pumped into a game->midi converter.

http://www.goldmine-elec-products.com/prodinfo.asp?number=G5514

They're small, made for laptops & such (back in the day), but work like a charm. Though, for musical work, it'd be nice to have something like this with a mouse-like ball (with inertia/weight and a rubberized feel), so the small ball doesn't slip.

[xbeemer]

For starters, here is rotary controllers done right, with all (well most) options for various modes built into the firmware. It suffers from a resolution of 128 steps, but with rotary controllers even that can be fixed in software, and has.

http://www.behringer.com/BCR2000/index.cfm?lang=ENG

What is perhaps most amazing about this controller is its price, about $125. Is it a cheap plastic box that will fall apart with a year or two of hard use? Perhaps. I got one and I'm pretty happy with its construction and feel, and delighted with it electronically. And if it fails in a year or two, just apply the Timex concept...toss it out and buy another.

Now, synthesizers.

The problem with making yet another minimoog is that there are so many of them already out there, and Moog is still doing it anyway. Another one, unless there is a bazillion dollar launch PR budget, is unlikely to get very far, especially if it is more expensive than most (as rotary knobs, trackball controllers, etc. would surely make it).

Suggest you think out of the box - literally. Since one of the largest expense components of synths is not the electronics but the connectors and knobs, consider making a black box synth that has no controls at all other than a MIDI Plug, then specify it so it can be controlled by MIDI in real time, and have it work immediately with one of the standard MIDI Controller devices, such as the Behringer, above.

Ideally a device such as this could be a single synth performance instrument, or (because you cleverly made them so cheap), they could be ganged together to make a modular polysynth. I don't buy hardware synths anymore, but I'd buy one (or several) of these if you got the sound right and the price right.

[DrJustice]

[xbeemer]

[mosc]

[jksuperstar]

One big problem I have is in live situations: when you're doing 2 things at once, you reach to grab a knob, but on encoders you can't feel where the value currently is. Sometimes that's ok, but other times it's not. I like motorized sliders for that exact reason-- even when you aren't looking, you know "where" they are just by their position.

[mi_dach]

The Machinedrum has an interesting use of encoders, they have an LCD which shows a picture of the eight encoders and their position. When changing presets, all of the knob positions are updated on the display at once, it works quite well, and you can see at a glance the state of the controls. Too bad the screen is a horribly blinding red color though.

They also have a push-action thing going on with the encoders, and its really cool. You push the knob in and theres a gentle click, and when you rotate it now incrememnts the values much more quickly. I think this both useful, and an extremely clever work-around for encoders with low resolution. DIY'ers can add similiar style with a 'shift' button nearby, rather than expensive fancyass encoders.

I've just bought 8 encoders and am looking forward to doing some experiments I'll certainly be trying to avoid collars of LEDS anyway

[bantri]

I really hate resistive potentiometers.



Their construction is usually made not to last, because the way they work involves physical contact over a thin carbon resistive layer that worns out with use.

I can bet everybody here has seen at least one scratching/jaggy volume control, because i´ve seen several of those nasty annoying resistive pots in equipment ranging from 10 to 10k dollars, dimmers, mixing tables, amplifiers, synths... you name it. (i´ve had several headaches and money loss with all of these)

The more knobs you have, the bigger the chance you´re going to have worn potentiometers, and unfortunately, that´s the case with expensive equipment like synths or mixing consoles, and only a few of them were made in such a way that is easy to replace the potentiometer with a new, cheap, ready to buy, spare part that can be bought in advance to have a backup.

Also if you accidentally get too excited and put excess force near the begininng or end limit of the analog knob, it can be permanently damaged in a single move.

This is why i love rotary endless optical encoders, they were made to eliminate all those problems and endure 5 lifetimes.



No physical contact, the reading is done by a ray of light passing through a slotted disk, like was done on the old ball mouses in which the ball was connected to two slotted disks, and lasted several years surfing faithfully several miles per day on the internet.

No limit to one turn, as they are relative they don´t send the absolute value like resistive pots, but increment/decrement clicks.

Even though, there are problems with encoders, obviously, the mechanical ones, these wear out with time too.

Luckily, mechanical and optical encoders operate the same way, so chances are that is possible to find an optical equivalent as a replacement to eliminate the problem.

The optional led collar comes with a bonus beyond simply displaying a graphical feedback the value of the knob, it can be automatically reajusted, which means that as soon as you change the patch on a synth, the leds automatic move to the correct position, and this in conjuntion with relative rotary encoders, avoids "parameter jumps" when you start turning the knob on a new patch, this would be a digital equivalent of a motorized fader which is IMHO completely obsolete (heavy, big, expensive, resistive potentiometer based)

Using this led collar in lots of knobs can be expensive though, if you´re using a computer screen that can supply the position of the knobs.

For the ones that use only a synth in live without a computer, the led collars are useful.

For both tastes there are cheap solutions, (using rotary encoders of course, like i said, i hate resistive potentiomenters)

i think that rather than sacrificing your expensive synth knobs, it´s better to buy a cheap controller and sacrifice it´s knobs.

or maybe in some cases like the nord modular, to just buy the engine (that has no knobs at all) and an external controller which when both combined, save a lot of money, and the cheap part that worns out can be readily replaced.

After doing some searching i found these:

(Note the quickly recognizable characteristic of NOT having a pointer or marker painted on the knobs themselves which i like to use for spotting *genuine* endless rotary encoder knobs.)

(i suspect that all use electro-mechanical encoders,

but a little circuit bending can be done to solve it, like adding external optical encoders using flat cables)


Doepfer midi encoder board


Doepfer Pocket Dial midi encoder controller (full of knobs, midi, but no USB)


Killamix (knobs with led collars, USB, but no midi)


Faderfox (full of knobs, midi with merge, but no USB)


have fun encoding!

[mi_dach]

Do you know where to find knobs without a marker, something like those on the doepfer or the faderfox?

Having used the machinedrum a little more since my previous post, I'd like to add that it's knobs are totally smooth, and get very slippy when you are sweaty. Two of the knobs developed a small fault where a quick twist would cause the value to actually drift in the opposite direction a little bit. Very annoying.

[bantri]

i´ve found some beautiful metallic ones...

with skidproof finishing around them.

i think they are not marked...

BestBassGear