experimental scope - poking the x0xb0x

[antto]

i want to know what happens in the different sections of the x0xb0x circuit
but i don't have an oscilloscope (as you probably know)
soundcard scopes (like Zelscopes) depend on the soundcard which is usually AC-coupled (mine too)
this means there's usually a simple RC high pass filter on the inputs with some low cutoff frequency
i thought "what if it's possible to undo the HP filter afterwards on the recorded material?"
asked the DSP fellows at KVR and someone made an example of such a filter
then i tested it, it worked..

some time later i have this:
http://www.youtube.com/watch?v=tL-Gb4q0uXw
this is the x0xb0x square wave

the scope is still in developement
it's going to be a VST effect eventually

i'll post more pictures/videos from interesting parts of the x0xb0x circuit later on

[Altitude]

looks right to me. I love how loosely the term "square" wave is thrown around.. You think the 303/x0x is bad, scope out what Korg calls a square/saw/tri on the Monotribe

[guest]

thats really cool
and i think a lot of people will find it useful
the triggering is really great

[antto]

re, i'm trying to figure out how the envelopes and accent works exactly
need assistance

as far as i understand:
the VCF envelope is born in C62 (let's call it MEG)
IC12B is controlled by the ACCENT CV from the cpu, and switches to shortest decay time on accented notes, otherwise the decay time is whatever the Decay pot points to
IC12A is also controlled by the ACCENT CV

in my "model" i have approximated the MEG pretty well.. but i'm still struggling with the accent


so i suppose what comes from IC12A is a copy of the MEG signal only on accented notes
and this goes to one side of the accent pot, the other side is connected to ground, the middle pin goes to (D24>R46>ResonancePot)
so, if the accent pot wiper is closer to ground - almost nothing goes to D24..
but otherwise what goes there is a copy of the MEG envelope on accented notes only
in my model, the ACCENT CV is 1.0 during an accented note, and 0.0 on normal notes, thus i approximate the signal before D24 as (Env*ACC)
i probed this with the scope - it looks and behaves the same

then, D24 makes a "sharp edge" out of this and passes it to R46 and the Resonance pot
there, C13 is a LP-filter, i approximated it to having a constant frequency of about 7.2Hz
the "sharp edge" from D24 now becomes smoother - this is then added to the VCF frequency and creates the "WOW" modulation sound you all know pretty well



good, but till now nothing explains the effect of "accent accumulation" - this is when multiple accents are played quickly, they each have their "WOW" but each next note gets a higher start on the VCF cutoff frequency
this accumulation i think can't be an effect of the C13 LPF, since it's frequency is too high to accumulate enough (i tested like 50 times with different setups and parameters, and couldn't approximate it)
so my conclusion is - there is something else going on..

in the VCA section, there's something interesting
C36 creates a RC-decay which only happens on accented notes
but it's different from the MEG envelope - it has a very slow decay time
this new envelope must be responsible for the "louder" accented notes.. it is probably added to the VCA envelope to make accents louder
this signal also goes thru R119 and from there, it connects to the wiper of the accent pot and D24
hm


i approximated the signal at C36 with a RC-decay with equivalent frequency of 2.15Hz and also multiplied it with my ACC CV, tested it - behaves the same as in the x0x
but from there, more questions arise

the envelope at C36 alone doesn't explain the accent accumulation effect, since on multiple accents, even at high tempo - this envelope always retriggers from the same amplitude
so there must be something acting as a LPF after it

but what?

NOTE: i'll put pictures inbetween this post to explain what i mean

[antto]

the VCA envelope i guess doesn't act on the VCF frequency
i tryied shorting C36 to ground, then raised the accent knob, accents were still as usual, except they were with normal volume.. so C36 is responsible only for the louder volume on accents, but doesn't explain the accumulation effect

i also tested shorting C62 to ground - as expected, no envelope comes out then, and the VCF cutoff is left at the ground level which is determined by Cutoff and Envmod pots
but also, accents are not possible then, neither does the "WOW" come from the filter, nor the VCA "bump" (C36)
i guess this is because C36 is retriggered to start at the amplitude where the main envelope is

[guest]

its definitely c13 that gives the increased high end
with repeated notes

it has a time constant of 1uf x 100k = 100ms
so it takes around 400ms to fully decay
and if you hit notes at a higher rate
it just builds its voltage up
i just double checked this on my x0x

[antto]

hm.. but this:

(signal at C13 with my soundcard probe)
is at insanely fast tempo (262BPM)
and it behaves like a LPF with a much higher frequency, also, it doesn't look like accumulating at all, while at the same time, probing the audio output of the x0x under these conditions - the accents were "flying" sky high at that tempo..
what's going on exactly?

[guest]

your soundcard scope probably has a lower limit
at which it no longer restores dc
its theoretically impossible to get all the way down to dc
so there must be a lower cutoff frequency

[antto]

excuse me, it was my dumb probe, i had a 100K pot on it, and i was probing at ~0ohms which seems to change the behaviour of C13
i set the pot to 100K and now i see the propper behaviour
i really hate myself now

i think i will try not to use my dumb scope, instead record audio from the x0x and use sonogram..

[guest]

the scope isnt dumb
although you bring up a good point
that high input resistance is critical to a clear signal

for this application
you would actually want 1Mohm or more
as the resistance of the circuit is around 100k

i made this a while back
http://www.openmusiclabs.com/projects/audio-sniffer/
for just this purpose
the schematics and such are linked from there
if you want to build your own

[antto]

cool, but it has a HP-filter on it's input..
and some of the parts are very hard to get here, like the audio jacks
i have a bunch of parts here.. opamps and such..

so i need a signal amplifier with huge input impedance and DC coupled

anyway.. this is gonna be difficult again
another aspect of the problem with the accent is how it's "amplified" into the VCF frequency
i mean, as cutoff is low - the accent sweep is with a low amplitude, but when cutoff is high - the amplitude of the acc.sweep is bigger (and this is not seen on C13)
even more weird things happen when the envmod knob gets involved

so i guess this happens in around the R71 R72 R63 and the cutoff/envmod pots..

btw, are you sure that C13 has a fixed cutoff frequency? because i suspect it varies depending on the signal
i mean.. i tested 100k+1uF=1.591Hz in my model and the rising part of mine looks too slow

[guest]

i didnt know the accent varied with cutoff frequency
could you describe this effect a bit more

c13 is like an envelope generator
it has different attack and decay
the attack is set by r46 and the resonance pot
and the decay is set by r72
except its not quite that simple
as they both vary with the resonance pot

decay is 100k + resonance position
so 100k to 150k
plus 2.2k or so because of r71 and q10

attack is ((47k + (50k - res. pos.))//100k) + (res. pos.)
where // means parallel resistor math
so 80k to 50k

[antto]

hm..
so i used the extreme values (which give slowest and fastest decay time for C13)
50K for attack, 100K for release, which is equivalent to 3.18Hz and 1.59Hz respectively
i feed my "C13" LP filter with a waveshaped version of the main envelope (the waveshaper is silly linear approximation of the diode D24)
and here's the closest i can get this way:

cutoff=1.0, envmod=0.0, decay=0.0, reso=1.0, accent=1.0
x0x is in blue, my synth is in red, when they overlap it becomes magenta

so, the first part of the accent sweep (the Wow part) is pretty close
but the accumulation on mine is not enough whatever i do with it

this was at tempo 172
now the same setup at 241BPM looks worse..

[guest]

where on the x0x did you probe
to get those waveforms

[antto]

both of these are sonograms from audio signal, in other words, from the audio output
this time it's real
since the resonance is at max - what you see on the sonogram is the filter cutoff frequency