[sdiy] TB303 Slide

Tim Stinchcombe tim102 at timstinchcombe.co.uk
Sun Jun 10 16:28:08 CEST 2012


> > OK, point taken! I must be looking at it in some cock-eyed 
> way... :-0
> > 
> > Or maybe I *am* being fooled by the 'terminology' - confused now :-/
> > 
> > Tim
> 
> The two *are* the same. You can check the maths below, but I 
> think it's ok (plaintext is not ideal for maths). The 
> equation for a cap charging is:

<<snip>>

Yes, thanks Tom - with the benefit of being 'after the event', I now have
several pages of similar equations myself!

I'm still smarting from my foolishness: the inquisition into where I went
wrong led to the discovery of many new 'cognitive biases' that I was
previously unaware of (anchoring or focusing effect; confirmation bias; bias
blindspot etc., to name but a few - I saw only what I wanted to see...), but
mostly I think I was guilty of the opposite of 'paralysis by analysis'
(which unfortunately doesn't seem to have recognised name) - I simply
assumed the 'internet chatter' I read on this subject after some Googling
was gospel, and so neglected to do _any_ analysis of my own which would have
indicated otherwise! (And apologies to Mattias for my wrong-headedness.)

So, rather than saying the slide is complete when we reach a certain
_percentage_ of the step-change, if we take it to be when we get within a
certain 'delta' of the final voltage, it is easy enough to get this
expression for the time taken:

t = RC(log Vs - log delVs),

where Vs is the step-change in voltage, delVs the chosen delta from it.
Since log delVs is constant, this shows the time has a log relationship with
the step voltage Vs. In order to try and guard against myself continuing to
talk utter bollocks, I plotted this curve in both Mathematica *and* checked
it against some SPICE simulations - they show good agreement.

Using the original simple R-C circuit as before, I stepped the simulation
from 0.1V to 5V in 50 steps:

http://www.timstinchcombe.co.uk/synth/const_time_glide_multiple.gif

I then arbitrarily decided to take delVs to be 4.2mV, thus corresponding to
the pitch being 5 cents away (5/1200 = 4.2m) from the final pitch, as I am
under the impression that being such an  amount 'out of tune' is hard to
tell (and taking either a smaller or larger value will, by the above
expression, simply move the curve either up or down respectively, without
altering its basic shape). Plotting the time to reach this point for each
curve against the step-change voltage gives the expected log curve:

http://www.timstinchcombe.co.uk/synth/const_time_glide_time.gif

Not so very constant after all then! (Perhaps we should change the original
'constant time slide' moniker to 'constant time-constant slide', which at
least would be more accurate, if more of a tongue-twister?!) Another thing I
will add is that not having a TB-303, I don't know how such things as the
gate closing etc. may 'curtail' whatever slide may be in progress.

Tim
__________________________________________________________
Tim Stinchcombe 

Cheltenham, Glos, UK
email: tim102 at timstinchcombe.co.uk
www.timstinchcombe.co.uk






More information about the Synth-diy mailing list