2025 Unexpected Neutron VCF Tweak
You know how the Neutron filter sounds with full resonance… some like that, I don’t. When I modded my old Neutron way back, I somehow never got to the filter, but when putting on a cool Oddvolts panel recently, It was finally time look at that ole screecher.
The Neutron VFC, developed by Keith Moffat (hence, “Moffat Filter”) is pretty much an 2164 data sheet example type state variable filter you can find in many designs such as Rob Hordiijk’s Benjolin, DIY CS80 filter recreations, designs by David Dixon, and, of course, many of Èmilie Gilette’s filters, such as the Shruthi or the Blades, to name but a few. In the SSI AN701 Application Notes, Figure 13 you get pretty much the core of this, and the frequency and resonance control look close to what we can see in the MI Blades schematics. What Moffat did differently, I cannot tell since I didn’t trace the circuit in completion but instead focussed on making it sound a bit sweeter.
Anyhow, here’s what I did. Point 4 would be essential, and the rest as you like:
- Change the Resonance Potentiometer range: I changed the travel of the Resonance potentiometer (VR11) because I want the filter to start of with a bit more body when Resonance knob is at zero but not go all the way to “shred” when fully up. For doing so, disconnect VR11 Pin 1 from the OPCB (it is connected to ground) and put a 1,5K resistor between ground and VR11, Pin1.
- Limit Maximum Resonance Potentiometer Range: Simply replace R559 (680 ohm, on the back of the PCB) with a 3.3k resistor (SMD size 0603). This limits the max reso setting to a sweet spot where you get full self-oscillation across the entire range.
- More DC coupling at filter input: I replaced C111 (1uf polarized electrolytic) with a 220nf film box cap. This seemed to help reducing distortion a bit in the HP setting.
- Clamp that Resonance with zener diodes: You can find this trick in a couple of SVFs. By putting two zeners in the feedback path, you clamp the maximum resonance and round that reso off at the same time. Use two 5,1V Zeners and connect their anodes. (NB the cathodes face “outwards” towards the IC connection points, not inwards like on some MI designs). Then connect the cathode of one to IC 29 (TL074), PIN1 (output pin of part A) and the other zener cathode to the bottom terminal of R145 (30k) or, alternatively, to IC 29, PIN 13 (-IN of part D).

Here’s a picture of my test-run diode clamping (you want to do this nice and proper with two through hole diodes, not like I did there 🙂
Some sounds:
And the old modding material…
Just updated some info here over summer ’21. What you’ll find are some easy fixes for well-known shortcomings of the Neutron (mostly square wave level, clicky ADSRs, S&H losing charge and the slew circuit not being unity gain).
Further down the road, there are some more laborious changes, such as implementing a different type of VCO hard-synchronization, Linear FM and Soft-Sync for both VCOs, changing ADSR Speed, and altering the normalized signal path of the synthesizer (if you hate re-patching all the time). Finally, there’s a sweet mod for an insert in the delay feedback path, allowing for filtered delays or feedback control per VCA.
VCO Section
Square Level Adjustment

Because the waveform outputs of the pulse wave on the 3340 is considerably higher than that of the other waves (+12V in contrast to +10V for the ramp, for instance), it is usually attenuated more than the others. Not so in the Neutron, where the ramp (i.e. saw) and square outs are both followed by 18K resistors. While this seems to be a good value for saw, your filter input stage can be a bit overwhelmed by the full oomph of the ramp, especially when both are engaged and/or PWM is active.
With 26K for your new resistors, the balance between saw and square is ace, but 25K or 24k are also fine (with the latter, the square a tad louder than saw). For VCO 1 replace R61, and for VCO2 replace R63. (If you have not lifted SMD components yet, best do a bit of training first; there’s also a little guide in my ressources section).
Linear FM for both VCOs
The 3340 chip allows for a super easy implementation of linear FM, which might be desirable over exponential FM in cases where you’d like to retain the pitch of the carrier oscillator. Here’s how it goes: simply wire a two jack input sockets via an 1m resistor each to the two respective points labelled Lin FM on the back of the PCB. Use a switched socket the NC lug of which goes to ground.

On my unit I didtched the two summing mixers on the patchbay and rewired the sockets as follows: lin fm1, linfm2, ssync1, ssync2, vco1 out, vco2 out (the latter two just mirror the main osc outputs, so I don’t need a multiple when using vco inter-modulation.
VCO hard-synchronization
The Neutron offers oscillator synchronization, yet users soon noticed that VCO sync sounds different here to what one might be used from other CEM3340/V3340-based VCOs (for instance, the delicious electrosmith VCO submodules). Some generous and tech-savvy person opened their Neutron, provided a good analysis, and offered a video tutorial of how to change this. My version of this is fully functional but maybe not as elegant as I remember the other one. However, I am sure you can easily improve it.
What happens in the Neutron sync-wise as is, as far as I understand it, is shown in the CEM/V3340 datasheet in figure 4. This is indeed hard oscillator synchronization, yet only on the rising/falling edge of the waveform (which sounds a bit like soft-sync). What most of us are used to, from synthesizers such as the Prophet 5, is what Curtis termed “conventional hard sync”, as decsribed in figure 5 of the same datasheet (see also Electricdruid’s great page on the 3340). If you want this as a second option, as I did, you need to make a little circuit as shown in figure 5 first and then hook this up to your Neutron.
Let’s do this in reverse, since the connections to the Neutron are the fiddly bit.

These instructions are for syncing VCO2 to VCO1. As you see in fig. 5 pasted into the PCB shot above, you need connections from PIN9 of your second V3340 (that’s the soft-sync input, yes SOFT-sync input) and PIN10 of the same 3340 (that’s the triangle wave output). For PIN9, there is no alternative solder point, yet its corner position makes soldering a tad easier. For PIN10 there must be a good alternative (the terminal of the output resistor just after TRI OUT), yet I have not found it yet. NB soldering in these areas requires a thin tip and thin wires with already some solder applied (suction is your friend here!). Make sure you don’t have a solder bridge between the pins you just wired up.
You then need the ramp (saw) output signal from VCO1, which I took from the top terminal of R62 (the green blob between the two through-hole resistors from the square level fix). If you want to use an external VCO for your sync in as an alternative, you could wire your ramp output to the normal pin of a switching jack socket, so that’d give you even more options.
Finally, you need some ground, and a negative power supply (I took it from the bottom of C157, which buffers the negative supply voltage for the 2164 quad VCA (IC39). (Tested and confirmed by Aart – thanks!).
Then assemble the little circuit as shown in the datasheet, figure 5 and wire up.


What you also need is an on/off switch between the negative power supply and the transistor. Important: this on/off switch is not only necessary for deactivating hard sync but also needs to be off when powering up the Neutron since having hard sync on interferes with the VCO tuning routine at startup. This is not harmful, it’s just that the Neutron cannot tune VCO correctly and gets stuck in the routine (if that’s the case, switching sync off and powering up again does the trick).
And that’s how it sounds in comparison:
Advanced SYNC Implementation with latching switch
This is in from Kees, who worked on an ace way of using a latching switch that is automatically off when the Neutron boots. This way, you bypass the tuning issue/boot hangup described above. This is proper stuff with an indicator LED and Kees generously agreed to have the schematic shared (thanks!).


Soft Sync (lazy and proper)
As mentioned, the 3340 allows for several types of oscillator synchronization, so, for the sake of variety, I implemented a “lazy” version of soft sync, losely based on the excellent Digisound 80-3 VCO. In the Digisound implementation, you’d wire your sync in signal via an 1nf capacitor to Pin9 of the respective 3340 that is to be synced, and wired a 100nf capacitor between the NC lug of your input socket and ground to prevent glitchy stuff when no sync input is plugged. Now, on The Neutron, there are already two caps between each Pin9 and ground (C291 for VCO1 and C95 for VCO2) which I didn’t want to remove. So the lazy version I went for is to wire up two input sockets via 100nf caps, which I then conected to C291 and C95 respectively (C95 is only accessible on the front of the PCB).

This lazy version works well with different waveforms as modulators and you might want to test putting some 47k-68k resistor in between the jack socket and the 100nf cap too to “soften” the sync effect more. [Like on LIN FM, I don’t but simply use external attenuators if need be.]
Proper version (not tested) would be to remove C291 and C95 and wire up as per Digisound schematic (see above).
Envelope & Utilities
Declick the VCA envelope (connection between ENV1 and VCA control)
You can nicely reduce the click audible on the Neutron’s VCA on low attack and release values for ENV 1, which is pre-routed to VCA amplitude modulation. Sice this pre-routing is done by means of a switched socket on the Neutron’s patchbay (when a jack is plugged into the VCA CV socket, the signal from ENV1 is disconnected), and since the Neutron’s patchbay are accessible from the back of the PCB, you don’t need to take off the red faceplate. simply take your Neutron out of the case, flip it over and look for C318, which you enlarge by means of a 1uf capacitor.

How this works is that the additional capacitor “rounds” (filters) off the ENV spikes a bit while keeping everything snappy about the ENV itself snappy. The capacitor placed there also does not round off any other CV sent into the 1. VCA CV plug or 2. the ENV output at the respective socket on the OUT section of the patchbay. For 1. you would place your cap over C340 and for 2. over C319. As for the value of your additional capacitor, I find 1uf works well. If you use a polarizing cap (the blue ones with the white side), the cathode (white-side leg) goes where the lower leg on the picture is (away from the normal pin of the jack socket).
Envelope Speed Switches
The two envelopes on the Neutron are not unlike the classic MFOS ADSRs. What this means for modding is that changing overall envelope speed is easy but something like auto-looping not. For an ace approach to the latter see this source here.
Besides declicking, I went for two simple changes on my unit. I swapped the decay pots (1m lin) for better versions by Alpha, and I implemented two speed switches by means of which an additional 10uf timing capacitor is engaged, allowing for properly long envelope times.

The respective timing capacitors on the Neutron are C169 (ENV 1) and C233 (ENV 2), yet putting in an extra cap is as easy as wiring it to PIN1 of the Attack pot. Anode of your extra cap goes to Pin1, and cathode to an on/off switch, which then goes to ground. I used a 10 uf, 60V electrolytic cap, but if you wanted to go fully prim and proper, I’d suggest removing C169 and C233 and toggling between a 2uf and a 10uf tantal in their place.
Enlarging the Sample and Hold Capacitor
In the audio demo linked above, you can hear the S&H on VCO pitch. In the beginning of the file, no mod is engaged, and you can hear the pitches of the VCO falling flat after a little while. This is on account of the rather limited storage capacity of C42, which holds the sampled voltage until the next sample is taken. If the capacitance of C42 is increased (by just wiring another capacitor in parallel), the sampled value remains stable. This, amongst other things, allows for extremely precise “computer room” sounds 🙂

I’ve not tested the value of C42 nor experimented with different sizes for additional caps but got good results with 1uf immediately, so that’s what I’m going to solder in. but NB I like sloow S&H action – the larger your cap the lower your maximum uptake rate, i.e. if you need fast S&H, you want lower values.
SLEW CIRCUIT FIX
The Neutron Slew Limiter changes your CV, which results in bad pitch tracking, when used for pitch CV signals. You can incrae its accuracy by simply bridging R590 and C588 with some wire.

Delay Feedback Insert
Some of the really really cool Eurorack delay modules have an insert on the feedback path. With such a function you can plug a filter or eq or whathaveyou in the echo loop so that subsequent repeats have a different sound characteristic. With the onboard BBD on the Neutron not being the clearest delay (I personally like that lofi techno vibe!), you might not expect this to be the next PSP Audioware 85, as it were, but this easy mod is still useful! Insert a VCA in the chain and you can control feedback per CV; put distortion or a resonant bandpass in between and get wonky!
Here’s how you do it: remove R386 (22k) and wire an output socket to the upper PCB terminal for R386 and a IN socket via a 22k resistor to the lower PCB terminal for R386. using a switched jack socket saves you plugging a cable for keeping the feedback loop (–> “retun” jack tip lug to 22k resistor, NC lug of same socket to top PCB terminal where R386 was; wire NC lug of that “return” socket to tip lug of “send” socket).
