[sdiy] Alternatives to Faders

[Richard Wentk]

On 7 Mar 2007, at 21:12, Edward King wrote:

> It would indeed. I wonder if theres a way to do this under a dark  
> plastic panel?

If it's clear acrylic, it will conduct the light, so all you'd need  
is a four LED assembly immediately under each wheel.

I'm not convinced this would be any more complicated or expensive  
than your bargraph and digits. Of course you wouldn't get the  
precision, but if you're not calibrating faders in dB, you might not  
need it.

> I suppose that depends on the individual needs of the operator  
> and / or where and how the generated values are going to be used.  
> Given that there is likely to be several stages of processing  
> between the rotary fader and the audio generation system, it doesnt  
> hurt to start out with as much accuracy as possible. Moreover, as  
> this a rate thing (rather than an analogue position thing), the  
> difference between data width is processor time, not hardware cost.
> I dont agree with above statement because rate - over position -  
> essentially means that the limitations are down to calculations  
> between the slowest you can possibly move the rotary fader versus  
> the fastest you can move it. If you want really fine accuracy, move  
> the thing very slowly. If you want to jump to max or get somewhere  
> quickly, flick the thing or rotate it quickly and get ready to hit  
> the "hold" button when you see the level meter getting close to  
> where you want to be, moving it slower as you get near there.

I'm not sure how intuitive this would be. The obvious advantage of  
faders is that there's a simple linear relationship between 'output'  
and position. If you attempt to add a fly by wire (as it were) system  
that tries to interpret dr/dt, but needs a 'hold' button to bring it  
screaming to a halt, you have a much more complex system that will  
probably need a lot of fine tuning before it feels right.

The problem isn't actually physical accuracy, so much as operator  
precision. And the fact that different parameters require different  
responses. (Faders should be log/dB, filter cutoff should be log fc,  
ADSR settings can be all-sorts, depending.)

Look at a professional console and you'll see that most of the action  
is in -15dB to 0dB range, with a smaller over area of 0dB to +10dB.  
Below -15db you don't need nearly as much resolution.

A 10-bit resolution is more than ample to cover these details,  
because even if you assume a 100dB range, you're working with 0.1dB  
steps, which is really all you need.

But for maximum *tactile* resolution the range won't be linear, and  
you can easily stretch out the most important top 20dB and squash the  
rest without losing tactile feel.

> I havent decided yet whether to use 0-255 or 0-100. Ive designed  
> the thing with 3 digits each for the coarse and fine values to  
> cater for both.
> Hex is only used in the assembly language I wrote for the PIC  
> controller but I had no plans to use hex for the actual displays....
> I dont know where the idea of hex has come from?

It was Tom's comment.

I have a couple of professional MIDI controllers that seem to use  
hex, and it doesn't seem to be an issue.

But I think for a mixer-type console, you really need the display  
calibrated in dB.

For synthesizer control, that would vary.

I think my ultimate design would probably be to have the colour-coded  
approach under each wheel, and then a single LCD display somewhere  
central showing numeric values - possibly only of the last few faders  
that have been changed.

The advantage is that you can program the LCD more easily to display  
different curves/units, ad lib. If you use bar graphs and LEDs you're  
more limited. With an LCD you can potentially add text, which makes  
the design clearer and more flexible.

Richard