[sdiy] scanned pots "jumping"

[Tom Wiltshire]

The simplest way I know of doing this is a running average of x  
values. If x is a power of two, the division that is implied by an  
average becomes a simple bitshift.
This makes a fairly basic LPF, but it works well enough for this  
application. More sophisticated filters would probably demand more  
processing (FIR typically) or more accuracy (IIR typically).

Even with the LPF (or especially without it) it is usual to set  
limits either side of the current knob position that the knob must  
move past in order for its value to be updated. A LPF will slow down  
the rate of ADC jitter, but won't remove it entirely -  for example,  
any slow drift due to temp change will get past the LPF.
On a 10-bit value, I've found that having limits of current_value-4  
and current_value+4 is usually enough, but it'd depend on how noisy  
the environment is and hence how much ADC jitter you're seeing.

I disagree with D on one point - I think there's a fair bit going on  
here, and it is easy to do badly and hard to do well. That's not very  
simple. He's right that it's a must, though - you need to deal with  
these issues somehow.

T.


On 3 Feb 2010, at 14:40, cheater cheater wrote:

> Dave,
> no, it is very simple and an absolute must.
>
> D.
>
> On Wed, Feb 3, 2010 at 15:35, Dave Kendall  
> <davekendall at ntlworld.com> wrote:
>> Thanks.
>>
>> I could have written that a bit clearer I guess. I leave the pots  
>> at certain
>> settings to prevent them jumping. The actual knobs are fine - they  
>> don't
>> move, only the parameters jump.
>> Interesting that it's a known technique to apply a LPF to the  
>> scanning - the
>> sounds and general build quality are good, I was surprised that to  
>> see the
>> jumping behaviour. Come to think of it, I've twice had to  
>> disassemble it
>> because a key stopped working. In each case a bit of crud had got  
>> under the
>> plastic membrane housing the 2 key contacts for each key - this  
>> was enough
>> to kill that key. Removing it fixed the problem. I guess that's a  
>> mechanical
>> issue - the membrane is not entirely flush with the pcb - it has  
>> some small
>> holes where I guess the dust got in.
>>
>> Shame that the software issue is there - is that sort of thing  
>> hard to
>> implement?
>>
>> cheers,
>> Dave
>>
>> On Feb 3, 2010, at 14:18, cheater cheater wrote:
>>
>>> Dave,
>>> parallax mode is a way of processing the position of the knob. It
>>> requires the potentiometer to be scanned too.
>>>
>>> If the knobs jump around without you touching the synth at all, then
>>> the knobs are most probably jumping because they have no lag
>>> processor/low pass filter applied to them. If the knobs jump around
>>> when you hit the keyboard hard, then it's a mechanical issue that
>>> cannot be addressed in software.
>>>
>>> Either way this is not something that should be happening except in
>>> the cheapest equipment.
>>>
>>> D.
>>>
>>> On Wed, Feb 3, 2010 at 15:04, Dave Kendall  
>>> <davekendall at ntlworld.com>
>>> wrote:
>>>>
>>>> Hi all.
>>>>
>>>>
>>>> Following on from the pots/vs. encoders thing, when playing a Nord
>>>> Electro
>>>> (with scanned pots, not in parallax mode) occasionally a parameter
>>>> setting
>>>> will jump, presuambly from the vibration of the instrument being  
>>>> played
>>>> hard. I've got used to leaving the knobs in a "safe' position  
>>>> before a
>>>> gig
>>>> in case something drastic happens, which on occasion it has..
>>>>
>>>> Is this a common fault with such systems? FWIW, the machine did  
>>>> this more
>>>> often when it was brand new...
>>>>
>>>> cheers,
>>>> Dave
>>>>
>>>> On Feb 3, 2010, at 00:35, Dave Manley wrote:
>>>>
>>>>> When an encoder becomes unreliable is it because
>>>>>
>>>>> 1. the switches are worn out
>>>>> 2. the switches are contaminated and need cleaning
>>>>> 3. the mechanical cam that operates the switches is worn out
>>>>> 4. debris from wear on the cam/switches is blocking the switch  
>>>>> operation
>>>>> 5. ???
>>>>> 6. any and/or all of the above?
>>>>>
>>>>> Anybody take one of these apart and do an analysis?  Does spraying
>>>>> cleaner
>>>>> into an encoder really do anything?
>>>>>
>>>>> -Dave
>>>>>
>>>>> _______________________________________________
>>>>> Synth-diy mailing list
>>>>> Synth-diy at dropmix.xs4all.nl
>>>>> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
>>>>>
>>>>
>>>> _______________________________________________
>>>> Synth-diy mailing list
>>>> Synth-diy at dropmix.xs4all.nl
>>>> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
>>>>
>>>
>>
>>
>
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy