Deep Bass Nine – 12 mods

Deep Bass Nine Overview

Deep Bass Nine – unverbastelt

This analogue rack synthesizer is from the 90ies when several manufacturers jumped the TB303-Techno-train. Feature list:

  • the typical seven 303 parameters
  • Midi-Interface
  • Audio input
  • CV-/Gate in- and outputs
  • Cutoff frequency can be set by modulation wheel
  • CEM3340 Oscillator. (In contrary to what i’ve read it tracks perfectly over the whole range. Maybe it’s because my PCB revision is “issue C”)
  • VCF is NOT a 303 clone and does not sound that nice, but it’s ok. Interesting: the reso pot is stereo … one half of it raises the volume level when reso is increased to compensate volume loss :-)
  • good hardware: tough steelcase, integrated power supply, high quality pots screwed to the case.

First disppointments

After playing around with it for while i was a bit confused – many things did not work the way as i experienced from using more than hundret different synthesizers throughout the last 20 years.:

  • the Cutoff pot’s range is not wide enough, the filter does not close (without Env Amount). If you adjust PR3 accordingly it does not open enough.
  • turning up “Env Mod Amount” lowers the cutoff frequency at the same time (accomplished with a Stereo-Pot)
  • the impact of “Env Mod Amount” is to strong, difficult to adjust subtle changes
  • the ModWheel impact on Cutoff is to weak, the filter won’t open completely
  • Accent only has impact on cutoff, not VCA
  • Accent starts a veeery long decay for the Accent envelope. Eh? LONG?! It should be shortened!
  • Accent does not shorten the Decay of the envelope to minimal time (as the 303 does)
  • the worst: Glide is NOT engaged on legato notes but with CC # 65. Even that is implemented quite strangely: Glide will be activated if CC #65 will be decreased, starting with 127. It deactivates when increasing CC #65 starting with 0. No, not the extreme positions 0 and 127 are used, the moving-away-from-them changes the glide state. Strange.

In the following all these issues will be solved and some more useful mods will come on top. The changes will not convert the DB9 into a TB303 but it will be much more usable and flexible afterwards.

IMPORTANT SAFETY NOTE: CAUTION! The inside of the instrument has some potential deadly points and cables providing high voltage and current. Some of the insulation tubes have moved throughout the years. The Powerswitch on the frontpanel is also connected to dangerous voltages. The following mods should only be made by a professional! Keep the power cord unplugged when working inside of the DB9.

Internal trimmers

DB9 – inside

A view inside reveils 2 PCBs, one for MIDI/CV/GATE and another for the audio part. The latter has some trimmers with the following functions:

  • PR1 Tune for Oszillator CEM3340
  • PR2 Oscillator frequency span (Scale), adjust to 1 Volt/Octave
  • PR3 Filter Cutoff initial point
  • PR4 max. Resonance, can be cranked up to reach self oscillation (set waveform to X)
  • PR5 VCA DC Offset (adjust for minimal thumps while volume set to max and waveform to X)


PDF with mods

I did not find any schematics for the DB9 and had to spend several hours for re-engeneering. All i can offer is a highres photo of the PCB with some marked, important points and a rough PDF with my moddings. Nearly all mods (except PWM) can be soldered directly on the topside of the PCB. To have space for the modding pots the midichannel turning switch was taken out. I fixed the midi-pcb-contacts to channel 9.

Stereoping DB9_Mods_V1_0

Mod 1 – Cutoff-Pot range

Very easy: R51 leads the wiper of the Cutoff-Pot to the cutoff mixing point. I replaced the original 330k against 15k. The pot now ranges from completely closed to completely open even without any env-mod. If might be necessary to re-adjust PR3.

Mod 2 – Cutoff-lowering by Env Mod Amount pot

R43 removed, R44 increased with another 47k.

Turning up the Env Mod Potis does not only increase the envelope impact on cutoff. It also loweres the cutoff at the same time. I did not like that and completely removed R43 (150k) positioned to the right of the Env Mod Amount pot.

Mod 3 – Modulation depth by ModWheel

To increase the impact of the mod wheel on the cutoff frequency you can lower R200. I changed it from the original 100k to 47k.

R200 – the left pin is the positive injection point for cutoff keytrack

Mod 4 – Decay Error Fix

As described above, ACCENT starts a very long (about 3 sec.) envelope changing cutoff. The reason is switch C of the 4066 pin 9 which leads the (one and only) Mod Envelope to an OpAmp input (IC 1 pin 5) while Accent is high. On the next non-accented note the OpAmp’s input is cut off and floating free. It takes some seconds for the OpAmps input to consume the charge and the output to come down to 0V again. This can be solved by pulling the OpAmp input/4066 output to GND over a 10k resistor. The Accent envelope then gets exactly the length of the ‘normal’ envelope. As this ‘error’ can sometimes be quite interesting i added a switch to decide if i want to keep the “Decay error” or fix it (by pulling the input to GND over the 10k).

Mod 5 – Decreasing Env Mod Amount

For me the Env Mod Pot has a to big impact for adjusting subtle changes. To the right side of the pot you find R76 (47k). I lifted one of it’s legs and added another 47k in series. You can as well replace it with 100k of course :-)

Mod 6 – Accent Envelope Shortening

One of the typical 303 feature is shortening the modulation envelope to it’s minimal time on accented notes. We can add that feature to the DB9 quite easy. There is one unused analogue 4066 switch : switch ‘D’ on pins 10-12. They are unfortunaley all connected to GND, it’s easiest to cut their legs and bend them up. Switch B is already used for leading the Envelope through OpAmp1 to the ACCENT for accented notes. So we connect CTR B (Pin6) directly to CTR D (Pin12). The switch-pins 10 & 11 will be connected with the pins of the envelope decay pot to shorten them out with the 4066 for accented notes. That’s exactly what the 303 does. As i also like to have the option, so i added another switch to decide between long, default and short,303-like accent envelope.

Mod 7 – Glidetime

Also very easy: R1 (100k) will be replaced with a 1M log Pot. Variable glidetime.

Mod 8 – Pulsewidth (see PDF with Mods)

The DB9’s oscillator is a CEM3340 which got an input for voltage controlled pulsewidth. The PDF contains the schematic for driving the input (you can not simply lead 0-5 volt into the CEM). As it’s got a CV-mixer i also added an additional jack for driving the pulsewidth with a LFO e.g. Interesting: by default the PWM pin of the DB9’s CEM-chip is floating free. As a consequence the square waveform is NOT 50:50 but a tiny bit smaller. That’s why the squarewave misses the typical hollow square sound. The mod fixes that, by proper adjustment of P2 you will get the complete range from perfect 50:50 to 95:5!

Mod 9 – Octave +1 and -1 (see PDF with Mods)

That’s also relatively easy. A voltage divider is buffered to get +1 volt. With an inverter of gain 1 we generate -1 volt. These voltages are soldered to the outer pins of a ON/OFF/ON switch. The middle pin will be connected over 100k to the right pin of R12. Why an ON/OFF/ON switch? Well, we want to have a switch position (in the middle) where the pitch is unaffected, right?

Mod 10 – Positive and negative Cutoff-Keytrack
(see PDF with Mods)

DB9 – modded inside

I am a big fan of VCF keytrack. The cutoff frequency is altered by the pitch of the played note. With the option to switch it also to ‘negative’, VCF-tracking allows the ’emphasizing’ of lower notes as the VCF closes more, the higher the played notes are. The mod is easy: The frequency control voltage from IC1 pin1 will be connected to a 10k lin pot (the opposite pin to GND naturally). The wiper is soldered over 47k to a flip switch’s centerpin. Both outer pins are connected to R200 and R47.

Mod 11 – Accent to VCA and Squelch (see PDF with Mods)

Now we add Accent to the VCA. The Accent envelope (after the Accent pot) goes into the same schematics as in the original 303 to form a VCA envelope. The 3080 OTA of the DB9 behaves somehow different than the 303’s BA…. OTA, it did not work to just connect the envelope over a diode with the OTA’s bias input. I therefore added a current converter, it’s output goes through a 470k log pot (for Accent VCA amount) and 22k directly into the 3080 bias input. The pot has to be soldered the wrong way to feel right. You might also try to find an inverse log pot.

I wanted a separate “Accent to VCF” pot. Instead of connecting it’s wiper to the cutoff CV mixpoint it is sent through a 50k lin pot to a 1µ electrolytic capacitor. The wiper of this 50k pot goes to the cutoff CV summing point. Now depending on the wiper position of the 50k pot the accent envelope for VCF gets a tiny attack and generates the typical 303 ‘barking’. In the 303 this 50k pot is one half of the resonance pot.

Mod 12 – Glide engaging by legato notes

DB9 new Midi Interface and quad OpAmp TL074 for mods octave +/-, PWM and Accent to VCA

This is the most complicated part. By default the glide is engaged by these strange CC#65 changes described above. As there is no chance to change the midi interface’s behaviour we need a second midi interface generating the glide from legato notes. So i added a second optocoupler and a Tiny2313 microcontroller. It analyzed notes on channel 9 for legato notes and velocity above 90 to generate signals for glide and accent. It also got some pins left over to drive two LEDs for indicating glides and accents.

The accent signal would not have been necessary as it works well with the DB9’s own interface. But i wanted to add another feature: as long as the notes C3, D3, C4 or D3 on channel 8 are held down, glide or accent will be ignored or forces. Means it affects all notes for momentary effects as long as the notes are held.


Deep Bass Nine – modded, frontpanel

No, it is still not a 303. It misses the feature following accented notes to raise the cutoff frequency (.. the 5.333V mystery…) and it’s VCF is surely not a 303-vcf clone. Both would have needed much more effort and time.