[sdiy] TB303 Slide

[Richie Burnett]

The confusion about the TB-303 gate pulse timing can be explained as 
follows:

In 'normal mode' a gate pulse for a 16th note coming out of the micro goes 
high for 3 cycles of the tempo clock and then goes low for the remaining 3 
cycles of the tempo clock.  This gives a 50% duty ratio for the gate signal 
where it originates from micro, but this is not what is seen by the outside 
world either at the gate output or audio output jacks.  Transistor Q36 turns 
on quickly but delays the falling edge of the gate signal by roughly12ms due 
to the action of C72.  The unequal propagation times for rising and falling 
edges noticeably stretch the note lengths from the internal sequencer.  12ms 
might not sound like much but it is easily noticeable on a pattern of 16th 
notes playing at a fast tempo.  12ms is about half of a tempo clock period 
at 120bpm, so it appears as if the gate output goes low shortly after 
three-AND-A-HALF tempo clock cycles.  This is the impression that a casual 
observer would get looking at only the DINSYNC clock along with gate out or 
audio output, instead of looking at the internal IO lines from the micro as 
Colin Fraser has done.  As Colin says the state changes are approximately 
synchronised to the active-going edges of the tempo clock only, it just 
happens that the subsequent circuitry delays the gate line's falling edges 
somewhat and lengthens the notes.

It's also worth noting that the gate signals from the micro are only 50% 
duty ratio in normal mode.  In the case of 'triplet mode' the gate line from 
the internal micro goes high for 5 tempo clock cycles and returns low for 
the remaining 3 tempo clock cycles.  5 out of 8 gives a duty ratio of 62.5% 
for 16th note triplets.  Then these gate pulses again are stretched by the 
unequal propagation times.

This behaviour is consistent for both of the 303's i've checked here. 
AudioRealism and Phoscyon fail to get these things right in their plugins 
though.

-Richie,


>> Note in there he suggests the gate on/off times are not
>> 50:50, but 3.5:2.5 (out of 6!) - however this 'scope shot':
>>
>> http://ultra303.ultrafex.de/twonotesslide.jpg
>>
>> seems to contradict that, showing it to be 50:50.
>
> It's 50:50 on my scope too.
> Robin's info is not entirely accurate.