this file lists all the mods i performed, along with a discussion of how they work.

synth:

1. vca envelope decay time.  the decay time is set by C147 and P26.  dropping C147 to 4.7uF gave a more reasonable decay time.  the 22k attached to the P26 bottom leg sets the min decay value.

2. vca envelope gating.  stock has gating on the vca, with decay engaging after key-up.  attaching Q47 through R265 to the top of C147 makes the vca behave similar to a 303, where there is no sustain period, and decay begins immediately, and the note shuts off when the gate is released.  this allows for slid notes to decay out.  the shut-off time after key-up is set by R265 and C147.  the way i did this mod was to make 2 cuts, add a resistor and 2 jumper wires.  it would be easier to just cut between R265 and P26, add a jumper from R265 to the other side of P26, and add the 22k to the cut side of P26.

3. vcf envelope.  this was changed to just function properly.  R209 is removed to make accented note NOT have a longer attack.  the R288, C133, R273, R274 changes all work to have the LM393 latch circuit actually latch off when the envelope reaches its peak, as opposed to saturating U18B.  there might be better ways of doing this.  there is still the issue of accented notes not having a higher envelope (like they do in the VCA).  this could 'probably' be accomplished with a capacitor at the cathode of D20.  this would hold a sample of the accent value during the trigger phase, and use it as the comparison for the voltage level at which the envelope started decaying.  care would need to be taken to ensure it charged up in <5ms, and fully discharged at some point.  if it didn't fully discharge right away, that might be fine, and would be similar to the 303 increase in accent after a string of accented notes.  but, this mod would also make accented notes have a longer decay time, which is the opposite of what you'd want.

4. vca cv control.  R73 and R76 were removed to make the vca "normal".  with these resistors in, there is a sharp drop off in the vca response at the last bit of the envelope.  it makes the notes sound gated.

5. vca input level.  R58 is decreaed to 22k to give a bit of boost in the volume as a result of the vcf changes (discussed below).  it adds a bit of distortion at high signal levels, but the vcf already does this, so its not a real loss.  its acts as a bit of compression as well.

6. vco mute.  remove R37 to get rid of the vco mute, which causes notes to unnaturally stop before the envelope fully decays, but still has all the vcf noise getting amplified.

7. vcf cv control.  this is converted to exponential by adding 2 transistors in a simple configuration.  the 4.7M sets the min current, and the gain of U20A sets the max current.  the control cv needs to be inverted, so U18A is bypassed and the added 22k and R233 being 330ohms sets the gain.  the cutoff pot also needs to be inverted, so the limit resistors are removed, and new ones are added to the appropriate rails.  the 100ohm is a bit small, and the lowest frequency is down in the subaudible range, which is excessive, but i liked how it sounded.  the max frequency might be a little low, but this is limited by R169 and R123 into the filter.  but, the easier fix would be to use smaller C91 and C68, as you have to swap these anyways.  there is also the issue of the 4.7M being rather large and causing the control current to be less than exponential at high cv's.  the fix here is to use a lower resistor value (500k or 100k), and then add another resistor to offset U20A to +50mV or so, to reset the control current range.

8. vcf topology.  the stock vcf is setup such that increasing resonance decreases passband amplitude.  this keeps the filter from distorting at high resonance, but also reduces passband.  this mod switches where the feedback comes from to the more standard approach of having resonance just increase amplitude.  in order to compensate, the vcf level is reduced a bit so there is more headroom.  but, it still distorts at high resonance, which i personally like a lot.  the stock input gains were also asymmetric and way too low, so R173,155,144,124 were all adjusted to make them more symmetric between stages and give a bit more level, which makes SNR better.  C91 and C68 had to be adjusted to get the frequency range correct.  smaller values might also be ok.

snare:

1. decay time.  this is set by P22 and C106.  i dropped C106 to 2.2uF and change R47 to 12k to give a smaller range of decay values.  R47 sets the min decay, and C106 sets the max.

2. noise signal.  the stock topology bled noise like crazy, so it was set back to be similar to the 808.  R157 is 100ohms to set the noise level.  larger resistors here give lower noise level with less distortion of the noise.  the noise is pretty distorted and loud, but i liked it.  it sounded more metallic than the standard puff of white noise that the 808 has.  but, it can sound more like standard white noise at lower noise levels.  if its too loud, but you like the metallic sound, increase R80.  or, if you increase R157 for less distortion, but want more noise volume, decrease R80.  C82 is removed and a 33nF is used to reroute the noise to Q24 base.  the value of the capacitor will determine how much low frequency noise gets through.  C96 and R182 are removed as they bleed signal when the vca is off.  

3. drum head tones.  i pitched these up, but that might not be what you want.  the pitch is controlled by R92 and R115, with R84 and R116 setting the decay times.  decay times increase with decreasing pitches, with no change in these resistors.  i went for very long decay times.  the mix of the 2 tones is set by R102, and i made it a full mix, so either tone can solo.

percussion:

1. accent effect.  remove D50,51 and swap C174,213 for resistors to give louder drum sounds with higher accent.

2. noise removal.  i found the noise be redundant, so i repurposed P24 to mix the 2 drum sounds.  this involved cutting some traces and running some jumpers.  C120 was removed to reduce loading on the opamps.  R8 was removed and the signal ot Q37 cut, just to help eliminate noise bleed in the circuit and protect Q37 from possible over-voltage.

3. vca levels.  the input to the vca was way too hot and was distorting, so R296 is reduced to 100ohm.  C131 is removed to make sure the signal at both inputs are identical (but opposite).

4. compressor muting.  R253,R6 were changed to get the vca to fully mute at the very tail of the envelope.  this keeps noise bleed out of the signal chain (although not super important since the main noise source is removed via (2) above).  C2 is removed to give a faster response time.  but, since R6 is so small, and this is part of the decay circuitry, C149 needs to double.  this can be accomplished by removing R279 and replacing it with a 1uF capacitor.  not sure how relevant these mods are with the noise out of the circuit, but it also been a long time since i did these mods, so there probably was a reason.  you just dont want the VCA closing too soon, and it would be nice if it fully closed at some point after the drum sounds were finished.

5. accent on attack.  U22B feeds a sharp pulse proportional to the accent value right at the beginning of the note.  the amount of this addition is set by R225 or R153, and the duration is set by C216 and R153.  C216 is quite small, and im wondering if this actually does anything.  if youre not seeing any difference, try a much larger C216.  the 10k is required to give a discharge path for C216.

6. faster compression response.  remove C151 for faster compression response, pretty much required for the accent on attack mod.  C39 also slows things down, so adjust that if its still not fast enough.

mixer:

1. fix howl to synth only.  this will make the howl function only distort the synth.  the synth volume will need to be turned down as the howl is turned up.  remove R106,122 and reroute the signals for the synth.  i felt that the distortion only sounded good on the synth or kick, and in retrospect, i should have put it on the kick, as the synth already has an  independent out if you want to add effects.  on top of this, the filter mods above already give some distortion.  it might be worth it to just bypass the howl altogher, as it always adds some noise, even when turned full down.

kick drum.

1. pitch decay.  there are some limitations to how much pitch decay can be added, as P7 is very small, and the signal there is consequentially also small.  Q31 acts to pull P7 down, pitching the drum up, with the pitch drifting back down as Q31 closes.  there is a balance here of how long the decay time is (mostly set by C98 and R188), and how much effect it has (mostly set by R181), since the decay time of the whole sound is completely variable as well via P21.  R205 sets the attack time of the pitch shift.

hi-hats:

1. set decay times.  i left the max decay times stock, even though they were pretty long, as i felt the open hat could also be used a cymbal this way.  the minimum decay times essentially muted the voices, though, so i increased these va R208,202.

2. change how tune works.  tune was not actually tune, but a mix of white and digital noise, and the white noise didn't sound good, so i hardwared the digital noise.  this is done with a bypass resistor from U14A to C127, and by removing R118.  P10 is repurposed by cutting its wiper trace, removing R113, and adding new limit resistors.  the wiper is then routed to a new transitor to replace Q72, which sets the voltage on U35 (the digital noise generator).  a capacitor is added at the base of the new transistor to stabilize the voltage, and the collector is run to +12V.  sweeping the voltage on U35 pitches the noise up and down, but the volume of the hats also change, which higher pitches being louder, so the hi-hat volume knob needs to be adjusted when pitch is adjusted.

3. closed hat closes open hat.  for the open hat to get closed, U36 is repurposed as a comparator on the closed hat envelope, to shut off the open hat if its ever triggered.  a resistor is run from U36B output to C104 to drain the open hat envelope capacitor.  this resistor value sets the decay time of this process.  i removed R386 to set a fixed level at which this process occurs, as the hysteresis didnt give enough "off" time.

4. pitch difference between open and closed hats.  i didn't keep this mod, but would love to have it switchable or tuneable some way.  it was very interesting, but i just couldn't pick a setting.  its pretty simple, just a resistor from whichever U36 output you want to pitch down (R364 or R353 - open or closed).  if putting it on U36B, you might need a diode in series with the added resistor to ensure that there isnt a leakage path between the P10 wiper and the open hat decay capacitor.  this would cause the open hat to never close.
