Just got a Korg/ARP Odyssey Module, which I needed to open for cleaning anyways, so I thought why not trace stuff and test some mods. And then things escalated somewhat…
Let’s kick off with some additional filter responses. You hear 1 pass each: the normal 4023 and 4035 (types I and II), then 1 pole 4057, 2pol 4075, 4pole 4075, and bandpass…
Just found a link to the service manual: https://archive.org/details/korg-arp-odyssey-service-manual. And here is the old ARP service manual: https://www.synthxl.com/arp-odissey/
If you ever need to calibrate your KARP, here‘s some good info: https://modwiggler.com/forum/viewtopic.php?t=282489
Max contemplating the possibilities
If Max had had a say at Korg, this thing would be called the Max-Korg, of course. This might have caused confusion with the model 800DV though, the so-called Maxi-Korg…
Bottom lid off and all sub-PCBs dismounted. Some modifications can be drone from the backside of the analog board, so there is no need to disassemble the smaller PCBs.
Modularized Oddy with a tad reverb…
Test Points
Big thanks to Dimitree from MW for pointing me towards schematics, as this was super helpful when mapping test points and all circuits.
| T1: CV | T2: Square 1 | T3: Saw 1 | T4: Square 2 |
| T5: Saw 2 | T6: Ring Mod | T7: VCO mix | T8: ADSR out |
| T9: AR out | T10: white noise | T11: pink noise | T12: SH Mixer out |
| T13: SH out (pre lag) | T14: lfo square bipolar | T15: LFO sine out | T16: 4023 VCF out |
| T17: 4075 VCF out | T18: VCA out | T20, T21, T22: Pitchbend CV | T19, T23: Vibrato CV |
| T24: 4035 VCF out |
Oscillators
Both VCOs and the ring modulator are as per old ARP schematics, using equivalent SMD parts. To be honest, I don’t plan on modding anything substantial here, so not much to say.
The glue blobs around the expo converters establish an air of mystery – black epoxy might have looked more vintage though… well, here’s the veil of mystery lifted…R51 is a tempco, Q1 (dual PNP) a 2SC4207, Q2 (npn-pnp) a HN1B04FU. The FETs are 2SK209GR. Wish me luck workshopping this, because I actually managed to muck one VCO core up…
Individual Waveform Outputs: for waveform outputs, simply tap the following points. If you want to go extra safe, use op-amp buffers, but to be honest, I’m going in 2600-ARP-style and simply use 1k resistors.
External FM modulation inputs: the easiest way to have an insert FM input for VCO1 and 2 is to insert a switched socket between the SH/ADSR switches and their respective amount sliders. You can do this by lifting the cathodes of D59 (for VCO1) and D40 (for VCO2) and inserting the sockets like so:NC lug to cathode of diode, Tip lug to vacant PCB terminal of respective cathode. If you just want extra FM inputs without attenuators, simply wire external sockets via 100k resistors to: junction R46-R48-R214 for VCO1, and junction R52-R170-R487 for VCO2.
External PW modulation inputs: adding simple CV in is easy: input socket via 56k resistor to the junction of R78-R202 for VCO1 and the same separately to the junction R80-R87 for VCO2. Inserting a socket between the LFO/ADSR switch and the amount slider requires you to cut traces. check the picture below. You need to cut the trace leading from the middle pin of the LFO/ADSR select switch to the via marked in a yellow box. Insert your switched socket liek so: NC lug to switch, tip lug to via. Be careful when cutting the trace!
Noise generator
The noise generator makes noise or something. There’s a massive cap too! Awesome…
White and Pink Noise Outputs: for white noise out, wire a 1k resistor between your output socket and the node of R328 (2.2k) and R228 (1.8K). For Pink noise out, wire a 1k resistor between your output socket and T11.
AUDIO MIXER
One clear difference to the old ARPs is that that the VCO, noise, and external input signals are not simply AC-coupled into the filter input but also buffered by an inverting amplifier (gain of x1). Moreover, where the old Oddys AC-coupled the VCO signal via 220nf caps into the filter, we get 10uf on the Korg , which is just a wee tiny bit more. Not sure how this and the inverted signal polarity really affect things – sounds good anyways, and gives you an extra audio mix output mod point.
Audio mix output socket: simply connect an output socket via an 1k resistor to T7.
FILTER TYPES
Korg managed to implement three different types, all of which had been implemented in various reiterations of the classic Odyssey. Such variety is really cool! As for individual schematics, check Don Tillman’s excellent page. The modules you’re after are 4023 (a 12db LP), 4035 (the -24db transistor ladder) and 4075 (the -24db norton amplifier lowpass).
VCFI – 4023: two OTA cells followed by integrators are connected to make a 12db lowpass. This setup is not unfamiliar to those of Oberheim SEM and the 1047 by ARP, though there are a couple of twists. More later, but first some simple points of interest here:
-6DB output: as for modding this properly, more later, but here’s a sound clip comparing the -12db low pass with the output from the first filter stage (-6db low pass)…
VCFII – 4035 type: as is known from late night television and frog-induced dreams, a company called Moog liked the game Donkey Kong so much, they started implementing ladders into their filters. Other companies like ARP found this appealing too but weren’t allowed to play with ladders – the rest is Nintendo lore… oh well…
Hotter Filter Input 4035: switching between Type I and II/III, the first one is quite louder. If you want some more dirt and oomph in type II and III, however, that’s easily accomplished. For driving filter II harder, simply wire some 10k-20k resistor across R347 (18k). Here’s what a 10k across sounds (in the clip i play first normal, then modded, then the same with VCA DRIVE on):
VCFIII – 4075 type: this is similar to the 4072, of post-law suit legend where transistor ladder setups were to be avoided on account of Moog. The 4072 used in later 2600s is very similar, but this design is yet simplified. Sleek and slick filter we have here…
Hotter Filter Input 4075: For driving filter III harder, simply wire some 100k-200k resistor across R262 (220k). Here’s what a 100k across sounds (in the clip i play first normal, then modded, then the same with VCA DRIVE on):
Different filter responses: It’s super easy to get different filter responses (including simple pole mixing responses) from this filter. Listen, for instance, how low pass sounds at -6db, -12db, -18db sound compared to -24db:
An here is oh eof my favourites – asymmetric bandpass created by mixing the first and last filter pole at equal measure (using 10k resistors in this example):
What needs to be considered though, is that, in order to keep the resonance path intact, we cannot simply rewire C83 to another pole output (as for why, see the following section). One option would be to giove each filter pole a dedicated output (using a voltage repeater as an output buffer in order to avoid possible damage) and have a VCA input insert point.
FILTER SWITCHING
The way filters are switched here is simple (and genius). Actually, the idea seems to stem from the Odyssey service manual that describes exchanging filter submodules, and folks like CMS offer such modifications as well. Let’s have a look at the framework, i.e. how each filter circuit is connected to the rest of the synth. Let’s use the 4075 as an example:
The actual filter is only indicated by the square box, and I marked the main connections, such as audio signal input, filter cutoff control and filter output in red. Now, the cool thing is that for changing the filter type while using all single controls for each type, you only need to switch between two points, namely filter resonance (post reso fader) and filter output. Since the “unused” filters remain silent, connections like cutoff and audio in can stay permanently on.
Although the switching could be done by the actual switch itself (Behringer go this way), Korg use a Switch IC for the actual routing duties, while the switch itself simply pulls various control pins (A and B) to ground. Now, theoretically, we could sequence filter types…
The switch IC is a CD4052, powered by +7/-7V. With a filter amplitude of -1/+1V (if I’m not mistaken) that’s no problem. The 4052 is connected as follows:
The resonance slider (B100k) is then simply connected to PINs 13 and 3 of this IC . If you look at the schematics of the individual filter submodules, you find that the old 4023 has a 100k resistor wired across the resonance fader – this was meant to spread out the resonance increase more evenly across the fader, but that’s not the case in the new Oddy. Hence, you get not much resonance in Type I mode for a long while and then suddenly a lot at a small distance. A tradeoff for having three types, I guess.
SHARED CUTOFF and FCV INPUTS: While the resonance path and output to VCA are switches as just described, all filters share common cutoff (per cutoff slider) FCV (per modulation mixer) and audio signal inputs. Audio ins are R141 (82k) for VCFI, R347 (18k) for VCFII, and R262 (220K) for VCFIII. For cutoff this is always: slider via 140K resistor to respective expo converter: VCFI – R88, VCFII – R344, VCFIII – R442. For FCV this is always: FCV node via 10k trimmer followed by 90.9K resistor to expo converter – trimmers: VCFI – PT26, VCFII – PT2, VCFIII – PT10.Implementing independent cutoff sliders would mean disconnecting the respective 140K resistors and rewire for new potentiometers/sliders (100kb connected between GND and -15V, with the wiper going to the 140k resistor). As for FCV, you’d need to take the trimmers out: since PIN1 and PIN2 of each of these trimmers are connected, so you’d need to lift the entire trimmer out and just put PIN3 back in.
Modularising the Filter Types/ Independent VCFs
Here is some information if you want to use one or more of the filter types as independent VCFs. This requires disconnecting the filter from cutoff in, FCV in, audio in, and most of all, the filter type switch. Some resistors best need taken out and then put in separately on a sub-pcb that also hosts your I/O and controls.
Here’s a start with the 4075: R442, R30 and R262 need replacement. The connections to filter switch (IC 19, Pin15, Pin2) need to be broken. +15V and -15V need to be supplied (check the OUTER cables of Connector Board CN1A), as wll as ground (middle cable CN1A).
After testing independent resonance control for the 4075 I found that inserting a 1uf-10uf capacitor between filter out and resonance pot in does help a lot with stable operation.
High Pass Filter between LP and VCA
Yup, the simple HP between main filter and VCA is nice for thinning things out, if need be, but maybe I’ll recycle that control for other duties in the future. The circuit itself is simple and can simply be bridged. Simply remove C106 and bridge its PCB terminals, and remove R303.
VCA
In contrast to the old Oddys, Korg opted for a discrete, albeit bare bones VCA design. The ARPs had designs around a CA3080 OTA, which could have been well substituted with a LM13700, but maybe the discrete design cost less…
BTW the resistors limiting signals from the VCA gain slider and the ADSR/AR fader have higher values on the actual board (22k and 33k) than indicated in the service manual. I halved the respective resistance on my unit and like that so far.
VCA AUDIO IN: Adding a a separate Audio input that does not break the internal connection to the filter out is easy: simply wire an input socket to a 10uf cap, which is then connected to a 100k resistor. That resistor goes the following pin of the VCA overdrive switch (as seen from component side) – middle pin, right hand row (facing the VCA GAIN slider).
VCA AUDIO IN (INSERT): Adding a an audio input that breaks the connection between internal VCF and VCA, you need to do this: Lift the anode of C105 (be careful, SMD electrolytic caps are hard work…) and insert a switched socket (NL lug to PCB terminal of anode C105 and Tip lug to actual anode of C105).
Extra VCA Control Voltage: wire a fast switching diode (anode facing the socket) between an extra input and a 6.8K resistor. The resistor, in turn, connects to the junction of R304-R280-R279 (see “CV IN” label on pic above)
Repurposing the Vibrato VCA: Cool thing is, there is a vibrato VCA built exactly like the main VCA at its core, and it is indeed tempting to reuse that VCA for non-vibrato purposes…
Decoupling this VCA from vibrato (sine LFO to VCO pitch), you would need to remove R411 (27k) which is the sine LFO input… Instead, use a 100k resistor and external input socket. Decoupling the VCA outoput can either be done crudely by cutting the trace above T23 or by removing R322 and R323 (each 100k), which send the vibrato CV signal to the VCOs. Your new output could be T23 via 1k resistor to external output socket. If your extra VCA needs to be AC-coupled, simply use a 10uf cap between T23 and that 1k resistor. If oyu want to decouple the CV input too, just lift the ANODE of D33 and wire an CV input socket to that lifted anode. In case you want more CV swing, decrease R405 (10k) to 6.8k or 4.7k.
ADSR AND AR ENVELOPE GENERATORS
ADSR output is at test point T8. Modding an ADSR out socket is easy, wire a 1k resistor between T8 and your out socket.
Notably, C37 (old C8) is a 3.3uf tantal SMD capacitor, not an electrolytic. (The advantages of tantal specifically in this envelope have been discussed in the context of 2600 modding.) Using a switch for faster ADSR speed would toggle between a 1uf tantal capacitor and the existing 3.3uf one. Do this as follows: connect the anodes of both capacitors to the PCB (base of Q87) and put an A/B switch between ground and the respective cap cathodes.
Never mind the AR out label. AR output is at T9, in case you want an output socket for this (use a 1k resistor in between). Not much of note here, except that the Oddy AR is improved over the 2600 AR in that a transistor (instead of a diode) charges the circuit.
Dedicated AR GATE: In case you want to modularise your Oddy and want to use AR and ADSR independently of each other, you want a dedicated AR gate. For this, however, you need to cut a trace – at the base of Q91 (see teh yellow Q5 marker on the pic above) there’s a through via – this connection between via and base of Q91 needs cut – then rewire the via through a switched socket (NC lug) to the base of Q9 (tip lug) – best use a fast switching diode between your tip lug and base of Q91 for protection. ((the base of Q91 is connected to SW13 “AR_KYBD_GATE”, on case working on the switch is easier than cutting a trace)
LFO
The LFO is a simple relaxation oscillator with TROI wave, square (clipped to unipolar amplitude by means of a diode), and (diode shaped) sine wave output.
Deactivate LFO reset: One of the few modification documented for the new Oddy… People who do not like that the LFO resets when a key is struck (or midi note comes in), might want to do this: first remove D5 (CR2 in the old service notes). Then take a new diode (1N4841 or similar is good) and wire the anode to the upper PCB terminal where D5 was removed. The cathode of your new diode goes to a 1m resistor, which, in turn, goes to -15V (use a nearby OPA supply). You can make this circuit switched by using a toggle switch that A/BS between your new diode and the lower PCB terminal where D5 used to be. Using a switched socket is also cool!
LFO SPEED MOD: LFO speed can be changed either by replacing the blue through-hole 100nf capacitor or by changing the value of R41 (R33 of old Korg, 390K). I find the latter quite convenient, because, if you want to make the LFO faster, you can easily do this form the back of the PCB. Simply wire some 100k-200k resistor between the terminals of R41.
Fitrst normal, then 240k across R41:
100k, then 240k across R41, then normal:
LFO Waveform Outputs: LFO SINE you get at T15, unipolar SQUARE you get at the junction D29-R245, and TRIANGLE (usually not utilised in the Odyssey) you get at the junction of C79 and R243 (ie.e. the terminal of C79 that faces the LFO frequency fader).
Sample and Hold (and Lag)
Good people, this is what a SH should look like. Mixing different sample inputs, having options for LFO trigger or keystrike trigger, and a lag (glide) circuit right after. As for mods, hmmm. at lest two things might come in handy – adding a separate sample input and having a trigger insert.
Extra Sample Input: Simply wire a 100k resistor (connected, in turn, to an input socket) to the node of R201, R222, R91 (see pic below). This way, you can feed in more sources for sampling.
SH Mixer Output: Since the SH input mixer can mix VCO1/2/Noise, you can easily misuse it as a small sub-mixer. Simply connect a 1k resistor to T12, and connect an output socket to that 1k resistor. A voltage buffer (op amp, for instance) wouldn’t do any harm here.
SH pre-lag output: could be fun to have a stepped-only SH output (tap T13 and connect via 1k resistor to output socket) that can be used in parallel with lagged sample and hold.
Lag (glide) insert socket: the lag circuit is a unity gain glide thing. You could use this for portamento purposes or other stuff too. To play this safely. we’d need the following: remove R237 (10k) and put a 10k resistor to T13 and on th eother end of that resistor teh normal lug of a switched socket. The tip pin of that socket then connects to an op-amp buffer (simple voltage repeater). The output of the buffer amp then connects to the junction of former R237 and VR45 (Lag Slider), i.e. the point I marked “lag In” below.
Track and Hold button: if you put a momentary on pushbutton or a switch between ground and the junction where C69 meets D6 and R203 (100k), you got yourself a track and hold thingy. For the time the junction C69-D6 is grounded, sample input passes through, and as soon as you let go, hold is active until the next sample step is engaged.
No, how to prevent eitehr note gate or LFO to trigger the next SH step? simply put the trigger select switch exactly in the middle position…
Modular Extension Panel
Not being a synth “player” in the sense of gnawing on my lip while pressing aftertouch on a keyboard, I’m inclined to replace the octave select switch and pressure point vibrato thingies on the left side of the panel with an I/O panel. Will need to measure a bit the depth available for parts, but would love to instal something like this with jack sockets and pushbuttons:
More when I have more…