I was just curious why the Shruthi design calls for pots vs. encoders (or does it?) for the 4 knobs below the display. It seems that endless rotary encoders would make more sense given the interface.
Here is a use case that helps explain my reasoning: When creating patches on the Shruthi I will frequently go back and adjust a parameter value up or down as I hear the interaction with some modulator. But whenever I go back to the parameter in question, there is a very good chance that the position of the knob controlling that parameter is no longer aligned with the actual parameter value. So I will have to move the knob further than needed when all I want to do is just nudge the value up or down. Encoders would solve this problem since the encoder would adjust the value relative to its current position.
Of course, on synths having more of a one knob-per-function design, pots usually make more sense. Although my Microwave XT uses encoders for every parameter on the interface and that aspect of the UI is really elegant. (Other parts of the interface are not quite as slick.)
Anyway, I’m just wondering why pots were chosen over encoders for the Shruthi.
You can use the encoder for editing parameters as well.
I prefer knobs because it takes a long time to go from 0 to 127 with encoders. And speed is important when trying different values for a parameter.
Pondering over the schematic, a pot takes just one port on the MCU, and encoder takes two (or three if it’s clickable).
Range: Cutoff goes from 0 to 127, so assuming we use an encoder of the common type (24 steps per rotation), it would take more than 10 turns to do a cutoff sweep. I’ve seen people do this kind of tweaking on elektron machines and it looks horrible. Encoders with more steps per rotation are very expensive. Acceleration curves / coarse setting schemes somehow solve the problem at the cost of more stuff done in software.
Feel: Even without detents, encoders are not as smooth as pots.
Lack of direct access: if I know I want a parameter to be at a particular position (say I want cutoff to be somehow in the middle), I can do it with a pot, not with an encoder.
Reliability: encoders have a shorter life than pots.
CPU power: encoders need to be constantly monitored by the CPU, at a very high rate, so that motion is detected. A bit of CPU overload, and boom, we miss a pulse and the information we get tells us it has been clockwise while it has turned counter-clockwise. Pots are much cheaper to read/scan in terms of computational costs.
@pichenettes. Interesting. As you are saying, I guess it all comes down to the quality/cost of the encoders and the associated software routines. For instance, the encoders on my Microwave XT and Nord Rack 3 are both very responsive (i.e., don’t require multiple rotations to span the parameter range) and very precise. But those synths were comparatively expensive. So it makes sense that they have high quality encoders.
The few encoders that are on my potentiometer Prophet '08 rack, on the other hand, are of the sort you describe - it takes forever to span the range of values when using those encoders. Encoders like that would suck on the Shruthi.
Anyway, thanks for the information.
There are also some interesting problems when the user fiddles with more than one encoder at a time - or tweaks it really fast.
On my supernova keyboard I experience anything from artifacts on the VFD screen, to backwards moving parameters if I tweak something really fast using a encoder… Not exactly the best user experience, but I must admit that I haven’t updated it to 2.0 yet…
In my personal opinion, when doing tweaking, id rather have pots and not endless enconders.
The Nord 3 has the best interface ever for a synth with patch memories and is definitely how to do encoders right but in general encoders annoy me.
Mechanical encoders wear out. Optical encoders get full of dirt and become very glitchy.
>The Nord 3 has the best interface ever for a synth with patch memories and is definitely how to do encoders right but in general encoders annoy me.
Couldn’t agree more. The Nord 3 interface is a work of art. Even though the sound is perhaps not the fattest or warmest, the Nord 3 is my favorite synth because the interface is such a joy to use.
The Microwave XT interface is nowhere near as elegant as the one on the Nord 3. But at least the XT encoders are super smooth and responsive. To keep them that way, I should probably cover up the front panel when I am not using the synth so as to prevent dust from getting in the encoders.
Problem about the XT is that its a lot of a hell more complex synthesis as the Nord Lead…
While there are parts that could be implemented better (albeit more expensive) i think Waldorf did a pretty much great job with the UI of the XT. Not as smooth as the Wave but way better as the Q.
I got a Kenton Killamix Mini the other day and I now thoroughly understand the difference between cheap and expensive encoders…
I’m using a home-made interface to program my Shruthi, a keyboard to chose the parameter and one encoder to adjust it.
I’m using an Arduino card with a midi interface to send to the synthetiser the NRPN INC and DEC commands matching the different parameters.
The encoder is running with interupt so losses of informations is prevented.
To handle the encoder rotation speed I tried two ways:
The First method was to use a timer to manage an acceleration curve.
The seconde one consist in an incrementation of the parameter by 1 if I only rotate the encoder without pushing it and by 10 if I rotate and push it. (close the switch)
Theses two methods are great and work well, making the interface really pleasant to use.
Finally, to display the parameter value I use a Sparkfun LED Ring like Nord lead 3.
@phm78. Wow, cool! Where there is a will there is a way.