Wednesday, April 8, 2015

Trigger in Mod for TR-9090

I've been building Trevor Page's TR-9090 Drum Synth, a MIDI-controlled clone of the voice circuits of the Roland TR-909.  Since finishing the main build and testing the voices, I started adding whatever mods I could find or think up.  

One feature I wanted to add is a row of analog inputs so that I can trigger the voices with external signals from my modular and sequencers.  I tried unsuccessfully to find a circuit that would have a low-parts count and be relatively simple, not require ICs or rare parts, etc.  

Eventually, I asked a manufacturer of Euro-rack modules if he would be kind enough to share his method for developing an appropriate trigger pulse from just about any external gate signal.  While nervously planning an appeal to SDIY, I got a reply.  He sent me an original design that has been used in his own products and gave me permission to share with the world.  (Thanks again! ;D)

Here is that circuit, redrawn for the TR-9090:

This is merely one possible topology for the circuit, but it is basically what i used to test functionality.  I made it narrow so that each sub-unit can be inline with the socket.  The TR-9090 will need ten of these circuits to control all the voices, so some additional design will be required to get ground and power to each sub-unit and all units on one PCB.

Note: this mod by itself does not include control over Accent level.  The trigger inputs alone will not produce any audio output from the TR-9090.  To test this circuit, I tied the Velocity In for the BD to the Aux Power connector's +5V.  (see below for more on Velocity patch points)

Below are the points on the TR-9090 circuit boards where this circuit could be inserted.  PCB 1 uses jumpers from PCB 2 to get the triggers from the MIDI Interface to the BD, SD, LT, and MT voices.  These jumpers can be used rather than cutting any traces or lifting components for these voices.  

For the other voices, either traces need to be cut or components must be lifted. By "lifted", I mean that the component leg circled in yellow should be desoldered and raised from the PCB.  The now-empty solder pad provides a tapping point for the trigger signal from the MIDI Interface, while the free leg is the trigger return from the added circuit.  

If you decide to cut traces instead of desoldering, the straight lines show the origin of the internal MIDI triggers for the voices of PCB 2.  

Note: the trigger signal for the HH voice is connected to two components, shown circled in yellow and joined with an orange line.  Both points need to be isolated but must remain connected together.  

I'm not sure what effect lifting or cutting traces will have on the signals feeding the Tom Noise circuit.  See the noise schematic below.

Here are the Velocity patch points.  For Board 1, I recommend using the jumpers as the send and return points as all four voices have two destination points (circled in yellow and joined with an orange line).  For HT, HC, and RS on Board 2, I recommend either cutting the traces from U32 or lifting the pins of the IC as described above.  Note: C107 and C125 are missing from the schematics.  For Cymbal voices, lifting the circled leg and inserting a socket as described in the Trigger In mod is possible.  I plan to use a 3-way switch for Velocity so I can toggle between +5V, MIDI control, and External control.

Monday, November 10, 2014

More A-155 Tricks: A Way to Shuffle Your Patterns

   Continued from my previous A-155 posts, but applicable to any analog sequencer being driven by an external clock... 

   Another difference between the current range of Doepfer Sequencers is the ability to apply a shuffled groove to your pattern.  The MAQ16/3 and A-155 do not have this feature, but the DarkTime does.  However, I don't have a DarkTime ... so I wanted a way to shuffle patterns on my A-155 using other modules.  [I say "using other modules" because the easiest way may be to program a shuffled trigger pattern in another sequencer or drum machine and feed that to the A-155 through either a MIDI/CV interface or the drum machine's Trig Out socket.   This way you are able to have rests in your pattern without affecting when the shuffled note occurs (as you might if using a Switch module)].   Since I like to have more manual, hands-on control, I wanted to create the shuffle and rests within the modular system.  There are a few modules available today that can provide a shuffled clock, like the Shuffling Clock Multiplier from 4MS.  But I don't have one of those either....yet (update: still).

   I'm assuming you all know what Shuffle is (also called Swing), but just to be clear (although, in my own words), it is the shifting in time of certain notes/steps in a pattern.  A "Quantized to 16ths" pattern would have the same period of time between all notes, a shuffle ratio of 50%.  A "shuffled" pattern would typically shift even-numbered notes to be later in time.  If looking at the pattern on a grid, the even notes would shift to the right.  In this image, you can see that the Quantized pattern has a 50% ratio ... the Even steps (Red) are half-way (50%) between the Odd steps (Blue).  The Shuffled pattern shown has a ratio greater than 50% (more than half-way). 

   My preferred method....

   The best way I found to generate a shuffled A-155 pattern from a constant clock source requires only a trigger delay, like an A-162 TDEL, a Trigger Combiner, and some multiples.  Some other modules can be added to this patch to make it more versatile, like manual switches and clock dividers, but they are not necessary for a quick test.  It isn't obvious in the images because the first mult is not shown (instead represented by two arrows coming from the clock source), but you MUST USE A TRIGGER COMBINER FOR THE SECOND MULTIPLE otherwise the two multiples are connected causing feedback into the trigger delay. This could be an actual trigger combiner module (like a multiple but with a diode for each Input, all connected to a single Out), a Logic OR circuit, a linear active mixer, etc. Here's how it would feedback...  

  Take a clock source in 8th note time for the tempo in which you wish to work. This can be from an LFO (preferably one like Doepfer's A-146 which has variable pulse width), MIDI/CV Clock out, 8th note trigger pattern, etc.  Using an 8th note pattern gives tons of space to work with between pulses, but to tighten things up a bit more and have all notes the same length, you should shorten the pulse to resemble a 16th note.  This would be referred to as 25% PW.  Either adjust it with the LFO's pulse width control, or use half of the A-162 to create a timed pulse (no delay, length adjusted to fit tempo).  Alternatively, if your LFO provides both Square and Pulse Outs and you are using a Trigger Delay with features like the A-162, you can eliminate the first multiple (only needed to split the LFO into two signals) by sending the Square Out to the Trigger Delay and the Pulse Out to the Combiner.  The A-162 can be used to reshape the Square to a similar width as the Pulse.  If using a MIDI/CV Clock source, shorten the MIDI notes to 16ths spaced 1/8th apart.

   Split the clock source (shown above as two arrows from the LFO) and send one path to half of the A-162 and another path to your Trigger Combiner.  Send the A-162 output to the Trigger Combiner as well.  Patch Trigger Combiner Out to the Clock Input of the A-155.  The A-162 creates the shuffle by producing a delayed version of the 8th note clock.  Of course, timing it to 50% would produce a quantized 16th note pattern, but that defeats the purpose...

(To calculate this trigger delay period in milliseconds, use this formula: 
a 16th note's duration is 15/BPM.  For example, if your song tempo is 120BPM, 15/120=.125sec or 125ms)  

With this method, the shuffled notes can be placed > or < 50%, which is quite unique as most machines with shuffle only allow ratios >50%.  Perhaps ratios <50% are techincally known by another name?

  If a rest occurs in the original clock signal (the odd steps), no shuffled note will appear in the following even step as there is now no pulse from which to create the delayed pulse. Therefore, in my example, the shuffling clock remains consistent, and rests are programmed using the Gate/Trigger assignment switches on the A-155 sequencer itself (shown below).  This allows the user to change the pattern as it plays, inserting rests on any step, odd or even.

   Adding Shuffle ON/OFF ...

   If you happen to have a few more modules, you can add an ON/Off switch so you can go from a set shuffle to no shuffle without changing your trigger delay setting.  The modules you'll need are an A-160 Clock Divider (or similar) and a manual A/B switch.  As shown in the diagram above, you will get two versions of the clock signal: a /2 version (16ths) and a /4 version (8ths).   

[edit: I realize after reviewing this that increasing the clock rate and using two divisions is likely unnecessary, and I can't recall why I did it this way.  It's been over a year, so perhaps there was a reason that I've since forgetten.  It seems appropriate to use the master clock and a /2 division as the 16th and 8th note patterns...] 

The manual switch selects either the 16ths, or the shuffled 8ths as in the previous patch.  I built a simple module with a few SPDT toggle switches for this.  You could use an A-150 VC Switch and a CV source (A-176-type CV Source, an un-sprung Joystick, etc) to switch them also.

  A similar shuffle method ...

   If you are using a constant clock source (16th notes with no rests) and creating rests with the A-155 trigger assignment switches, you can use an A-151 Quad Seq Switch (or similar) to cycle between a normal trigger and a shuffled trigger.  This method does not require using a Clock Divider.

   A 16th note clock source is sent to a multiple and from there to an A-162 TDEL and to the Trig In of the A-151, as well as I/O1 and I/O3.  The TDEL output is sent to another multiple and from there to I/O2 and I/O4 of the A-151 Seq Switch.  Each master clock pulse generates the normal clock, the delayed variation of the normal clock (through the A-162 TDEL), and the trigger which forces the Quad Sequential Switch to alternate between them.

Shuffle using A-162 (or A-142) and A-165

  After publishing the above methods, I was contacted by musician/author Florian Anwander who realized a clever way to achieve a manually adjustable shuffled clock using Doepfer's A-165 Trigger Modifier module.  I don't have this module so had not considered its applications.  Many thanks to him for suggesting this patch!

  The A-165 has two functions: it can invert a trigger input signal (high input = 0V output, low input = high output), and it can produce a 50ms pulse at both the rising and falling edges off the input signal. Rather than using the A-162 TDel's DELAY setting to create the shuffle, now you can use the LENGTH setting instead. As the A-162 is a dual module, one sub-unit can be used to set the space between 50ms pulses (the shuffle ratio), and the other sub-unit can be used to set the pulse width of the final signal used for Sequencer Clock and/or ADSRs.  Just make sure that the Length set by the second sub-unit does not overlap the next master LFO/Clock pulse.  

  The example patch below shows a common Square LFO shape (50% pulse) as the clock source.  Alternativley, an LFO with a variable pulse width (like the A-146 in the above examples) could be used instead of the first A-162 sub-unit, and an A-142 VC Delay/Gate module could be used in place of the either A-162 sub-unit.  The second sub-unit's function is not entirely necessary. 

Saturday, October 26, 2013

Input to Comparator : a mod for the Doepfer A-119 and A-142

While working on what will hopefully be my next post, I was aware that I still do not have a dedicated comparator module in my system.  However, I do have several modules which contain internal comparators.  So, I decided to modify some of these to allow me access to the comparator function for use with other external signals.  

A comparator circuit is often used to derive a gate signal from a varying voltage source based on when the source exceeds a certain voltage (defined by the Threshold setting). The modules I have with internal comparators like these are the Doepfer A-119 Ext. Input/Env Follower, the A-142 VC Decay/Gate, and the A-143-1 Quad AD/LFO.  All these modules use a very similar circuit to Doepfer's A-167 Analog Comparator module.  The A-142 VC Decay/Gate (like the A-143-1), generates its Gate Out (Comparator) signal from its Env output while the A-119 generates its Gate Out from its Envelope Follower Output.

There are times when I want to use the VC Decay or Envelope Follower (or its pre-amp) but don't want to generate a Gate from their own outputs.  Instead, I'd like to use the internal comparator function with an external source.  By performing a simple modification to the A-119 and A-142, an input socket can be added to allow one to use the internal comparators with any signal (LFO, Sequencer, Min/Max, Random, etc) while still being able to use the VC Decay and Ext In/Env Follow functions separately.  Note: this mod is only described for the Doepfer A-119 and A-142, but it probably could or probably has been applied modules from other manufacturers. 

If you are already familiar with this type of socket, you can skip ahead to the videos or instructions below them.  

But, if this is your first time working with this type of socket, this might explain how it works (and other sockets with normalized connections:

You'll notice if you look at a socket on your module that the legs/pins on either side have a "normalized" connection, meaning the opposite pins are connected when NO plug is inserted into the socket.  This allows signals to be routed through the socket and interrupted with patch cables.  One leg/pin bends into an arm at 90 deg on the top of the socket and lies underneath the other side's longer arm.  I'll refer to the shorter arm as the normal arm and to the longer arm as the switched arm. Think of the socket as having a switchable input and an output for each terminal (Tip and RIng of a mono 3.5mm patch cable), and this is regardless of whether the job of the socket is a that of an Input or an Output.  Normally (when no plug is inserted into the socket), the signal at the NORMAL pin passes through to the OUTPUT, or switched pin.  But when a plug is inserted, the SWITCHED arm lifts off the normal arm, breaking the normalized connection, and instead the switched arm is only in contact with the plug's signals.

Watch these two short videos to see the Comparators reacting to external signals.  You will notice that the Env LEDs to not match the Threshold LEDs.  

A-119 "Input to Comparator" Mod from N K on Vimeo.

A-142 "Input to Comparator" Mod from N K on Vimeo.

Disclaimer:  Hang on.  Yes, this is a simple modification.  But be aware that any mods done other than by the factory will void your warranty.  Also, you should know and follow safety procedures and have well maintained equipment.  You should know how to solder.  Don't solder while in the bath.  Turn off power and remove from rack/enclosure before attempting to modify.  I cannot be held responsible for damage to your modules, burns on your fingers, spilled coffee, etc.  If you follow the instructions and look at the images, you should be fine.  I am neither an electrician nor an engineer, but I managed it.  If you're unsure, stop.  Ask a question in the comments.  If you do need to remove a PCB from the faceplate, be careful not to scratch it.  Do not touch your tools to the faceplate! It is very easy to scratch them. 

A-119 Mod:

On the A-119, the Envelope Follower signal passes through R24 (100K) to the comparator circuit.  We want to place the interrupt socket on the Env Follow side of R24, so the signal you patch in will pass through R24. 

Now rather than drill a hole in the faceplate or use a blank 4HP plate to mount my insert socket, I chose to re-purpose the 2nd Audio Out socket on the A-119 for use as my Comparator Input. You need to carefully de-solder and remove this socket from the board, or if you have extra replacement sockets, you can cut it from the board using a Dremel or wire cutters, then desolder what's left of the legs.

Next I bent the rear pins/legs of my socket up at a 90 degree angle and soldered two short pieces of wire to the pins. See Image.

Solder a 2" wire to the normal pin of the socket.  Solder another 2" wire to the switched pin of the socket.  Desolder and remove R24 from the circuit board.  Resolder the front two Ground legs of the socket to the PCB, or if you are building this mod on its own faceplate, you must connect the front two pins of the socket to Ground somewhere on the module, like the solder points for the pot brackets.  

Twist and loop together one end of R24 and the switched wire and solder together (1, 2).
I slid some shrink tube over R24 and its connection to the switched wire (3). Resolder the other end of R24 to the open through-hole on the PCB (4) and cover R24 with the shrink tube (5).  Solder the normal wire to the IC2/Pin1 junction point (6).  My wire was too heavy to get through the hole at R24, so I soldered it to some of the component legs at the same junction (6,7).  

Now the Envelope Follower signal will be sent through the socket to R24 and the comparator circuit.  Patching an external signal into the socket interrupts the internal Env Follow signal and replaces it with the external source.  

A-142 Mod:

Modifying the A-142 is a very similar process to the A-119 mod above, however the module doesn't have an "extra" socket which I can re-purpose.  Instead, I used a 4HP blank plate which I already had on hand and was already drilled to fit the Doepfer-style 3.5mm Cliff Sockets.  I threw some more sockets in the other holes and made some extra Ins and Outs.  Here's what it looks like now (left) and what it might look like if I rebuild it (right).

The resistor we want to desolder and remove is R21, located here near pin 1 of the TL084.  In this image, the signal path is travelling from right to left, so the right side of R21 is the junction of the VC Decay circuit and the comparator circuit.  We want our external signal to pass through R21. Since my 4HP expansion will be positioned on the left of the A-142 in my rack, I want to move R21 to the solder-side of the PCB, just like in the A-119 mod above.  

Basic mod:  prepare a spare socket by soldering a 2" long wire to the normal pin, another 2" long wire to the switched pin, and a 2.5" wire to the ground pins as shown here.  Twist together the end of the 
switched wire and one end of R21 and solder together.  Slide shrink tube over the switched wire.  Feed the other end of R21 through the hole in the PCB (shown left, "In to Comparator") and solder in place.  Move the shrink tube down to cover R21.  Solder the normal wire to the other solder pad (shown left, "Out from "Env Out signal") from R21's original position.  Solder the Ground wire to a GND pad on the module, pot bracket pads are good (shown left, "Ground"). Mount the socket on a blank plate.

Since I had extra holes in my plate and spare sockets, I added a few more Ins and Outs:
-  a socket is connected to the Trig In socket of the A-142 and simply acts as a multiple for the Trig In source.  
-  a socket is connected to the CV In socket and acts as a multiple for the CV In source.
-  a socket is the Comparator In socket from this mod
-  a socket is connected to the "switched" signal from the Comparator In socket.  If no patch cable is inserted into Comp In, this socket is a multiple of the A-142 Env Out.  If a patch cable is inserted, this socket is a multiple of the inserted signal. 

Thursday, August 22, 2013

An Example of Timing Multiplication with the Doepfer MAQ16/3

A few months ago, a fellow MAQ Sequencer user asked for some suggestions to spice up his sequences.  I recalled an audio/video recording I had made previously when I was experimenting with the MAQ's "Step Duration", where one sequence row affects the timing of another row (or two).  Basically, the controlling row (Row 2 in this example) is set to progress slower than the affected row (Row 1, which creates the note sequence), so that multiple Row 1 events might occur during each Row 2 event.  The value of each Row 2 step determines just how many Row 1 steps will occur.  Confused?  Follow me?

Either way, watching this video might help demonstrate.  In this video, the top row is creating a note sequence which you can hear right away and which continues through the song.  The middle row is controlling the timing of the top row.  Watching the lights, you will notice that the top row occasionally appears to speed up.  Some notes at the beginning and end of the song were muted in the recording, but you can hear the effect almost immediately. The sound being sequenced with the MAQ16/3 comes from the A-100 System, modified manually during the initial recording.  The other parts of the song were added afterwards. 

MAQ16-3 Step Duration (using t.4. on Row2) from N K on Vimeo.

Saturday, June 22, 2013

Using the Doepfer A-155 to Re-arrange Audio

   After thinking about the differences between the currently available Doepfer sequencers, I decided to try out a patch which would randomly switch between free-running beat loops using the inputs to Row 2 of the A-155, a feature unique to the A-155 in the Doepfer range.  
   Have a look at the video below and follow the LEDs of the top of the module.  You will notice that each step plays a different AUDIO beat loop. 

A-155 as Beat Slicer from N K on Vimeo.

about this patch....

  To make this patch, I took eight individual drum loops from old tracks of mine and loaded them into Ableton Live.  Each loop was routed to a separate mono output of my MOTU 24I/O soundcard, and from there patched directly into the sockets on the bottom of the A-155 module.  A few of the loops got some effects applied to them in Live, modulation effects like flange, so each time that beat is heard it is at a different point in the flange sweep.  What you are hearing is the CV output of the sequencer....which would normally be a set of DC voltages for controlling pitch of a VCO, etc.  But in this patch, the output is actually audio.  By patching into the lower row of sockets, you interrupt the voltage being sent to the Row 2 CV controls, which now act as volume controls for each input signal.  

  This beat-switching effect is not wave-sequencing, because each new step does not re-trigger the selected drum loop.  Instead, all of the drum loops are always playing in sync, and the sequencer is choosing which one will be heard at any given moment.  The result is similar to a DJ cross-fading between two beat-matched records, except there are eight records!  This patch is in no way restricted to beat loops, as any signals can be sequenced with this method:  different LFOs, eight more A-155s (yes!), pads, vocal tones, sound effects and drones, arpeggio loops.....

  The sequencer's audio output was processed with an A-109 Signal Processor with S+H to cutoff with some resonance with manual adjustment of the A-109 and A-199 Spring Reverb.  

Another example....

  Here's another example using the Doepfer A-155 Sequencer to manipulate audio signals.  This time I used some Jungle/DnB loops, and none of those loops were processed pre-A-155 unlike in the previous example.  I thought more "realistic" sounds might make it easier to hear the loops change.

A-155 Jungle Sequencer from N K on Vimeo.

  This patch was very similar to the first, using the same trigger pattern to control the sequencer clock and the S+H filter effect.  The A-155 has two CV outputs per row, Pre and Post, which in this case are the same signal.  Pre Out went straight to the mixer, and Post Out was split to the A-109 with S+H and an A-124 Wasp VCF with FM (fast modulation from an A-110 VCO and slower modulation from an A-145 LFO).  Both filtered versions were added to the Pre Out signal with an A-138 mixer, boosted with an A-119 Ext Input, and sent to an A-199 Spring Reverb.  Some manual adjustment of Wasp FM, A-109 Fc/Q, and Reverb Mix was made during recording.  

Saturday, April 6, 2013

Octave Cat SRM - Fixed Schematics and Layout

Original documents vs actual PCBs...

original schematic
original component  layout

The Octave Cat SRM documents you can find on the web are all the same, at least all the ones I have ever seen.  They probably all came from one source, likely Kevin Lightner's Synthfool siteThe problem though is that some of these documents (the schematics and the component layout) don't actually match the circuit boards in the SRM synthesizers that I have seen.  Nor does the schematics image match the layout image. The units I have looked at all have traces which have been cut, jumper wires intstalled, missing components, and/or extra components on the back side (solder side) of the PCB.  Not only do these models I've seen not match the documents, but they don't match one another either! 

Before I post any instructions or information on the Factory Mods that were offered by Octave Electronics, Inc., I thought it best to attempt to correct the original documents to match the actual circuit boards we have in our synthesizers.  Perhaps the corrections I show on these documents will match your own unit and save some confusion when performing the modifications, general troubleshooting/repair, or tuning.  I chose not to relabel every component on the schematic, but I got the ones that were hardest to read. 

Note:  All Cat SRMs I have looked at have some different resistor values (and in some cases, trimmer values) from the schematic and from one another.  Don't worry if your values are again different than those shown here.  

Below you will find two versions of the SRM "Board B", which is the VCF/VCA section of the synthesizer.   I haven't done this work for Board A yet.   Thank you to Tony from OakleySound for confirming that the two CV Scale circuits I drew are both viable circuit options (see pages 2 of SN:998 and SN:3866 Schematics, below).

My own SRM's Board B:

My SRM is Serial Number 998.  It has some cut traces around the Volume control and jumpers to route the control to other spots on the board.  Neither the way the board was made nor the routing change matched the documents.  Also, the CV Scale part of the VCF control circuit is different from the original documents.  The component side of SN:998 is fully populated, with the exception of there being no Temp Co resistor (R193) and that C77 (not shown on the original schematics, but does have a place on the PCB) is mounted on the solder side of the board.

The SRM SN:998 schematics, page 1:

click image to enlarge

The SRM SN:998 schematics, page 2:

click image to enlarge

The SRM SN:998 component layout:

click image to enlarge

Another Cat SRM's Board B, owned by Gregory Cox. 

Thanks to Gregory Cox for providing some excellent photos of his SRM SN:3866.  This unit has the same cut traces and Volume control jumpers as my board.  However, this model has no C77 installed, and is missing another unlabeled resistor.  The CV Scale circuit is again different from the schematics and different from my own unit, with resistors which are shown in the schematics mounted on the solder side of the board (as they have no place on the component side).  This SN:3866 Board B does have the Temp Co resistor included in the circuit.

The SRM SN:3866 schematics, page 1:

click image to enlarge

The SRM SN:3866 schematics, page 2:

click image to enlarge

The SRM SN:3866 component layout:

click image to enlarge

Do these match your SRM?

If you have an SRM that doesn't match either of the examples shown above, and can take some good photos, please contact me via the comment section below.  Using your photos, I can alter these files to match your Board B and add them to this page and/or email them to you.

Sunday, March 31, 2013

Octave Cat SRM - Factory Modifications

photo courtesy of G. Cox

This is the first in what will be a series of posts about modifications for the Octave Cat SRM synthesizer.  The series will cover some of the original factory mods as well as some that I've done myself.  I hope to include enough detailed information to enable anyone else to perform the mods on their Cat SRM.  

the factory mods

Here are links to the documents in which Octave Electronics described the mods that they offered:

-  Cover page
-  Mod list
-  Package price list
-  Page 1 ( Mods 1 - 6 )
-  Page 2 ( Mods 7 - 10 )
-  Page 3 ( Mods 11 - 12 )
-  Page 4 ( Mods 13 - 17 )
-  Page 5 ( Mods 18 - 19 )
-  Page 6 ( Mods 20 - 21 )
-  Page 7 (Mods 22 - 23 )

I bought my SRM from a seller on eBay sometime in 2006-2008.  It came to me having had several modifications performed at the factory over 20 years prior.  This is the only modified Cat SRM I have personally seen, and I have seen several SRMs over the years.  It came with these documents shown above, along with calibration instructions and the addendum to the user manual (five pages titled "Instruction Manual Modifiactions").  I will include these other documents in a later post.  Some of the factory mods to my SRM are NOT on the list and appear to have been specifically requested by the original owner.

my modified Cat SRM


Shown here are several factory mods:

Output Mute, Pitchbend Range and Normal switch, Keyboard/External Control select switch, Glide On/Off switch, LFO DC-Offset switch, Keyboard or LFO to Sample + Hold Trigger In switch.


Some more factory mods and some of my mods: 

12/24dB slope switch, HP/LP mode switch, Resonance Send/Return Loop, Noise Send/Return Loop and Routing switch


Another factory mod and more of my own:

VCO1 Audio to VCF Mod1, VCO1 and 2 Send/Return Loops and VCF/VCA routing switches


I added these Doepfer-style sockets at the edge of the case.  These sockets have "normal" connections, allowing a signal to run through the socket until a cable is plugged in.  The cable interrupts the normal signal and inserts the external signal. The sockets shown here are Audio and CV inputs to the VCF and VCA.  The VCF Out signal can be interrupted via the VCA In socket.  The External In sockets mix with the VCF Out instead.

 These sockets, switches and pots are part of the factory mods, however they don't all match those listed in the documents above.  I haven't yet determined what function the Gate socket provides.  The LFO, VCO2, Interface, and Pedal inputs are all CV inputs for use with a pedal or other CV source. 

The Pedal Input is connected to the VCF, VCO1, and Pitch switches and pots, allowing you set a mod depth for those three parameters with the pots and to have on/off control with the switches.