Modification of the PAiA Synth Drum

I have successfully modified the Synth Drum of the PAiA 9302K Drum Tone Board to have front panel controls for Pitch, Decay (Q), and depth of frequency Sweep. At first, I didn't expect to have much luck at this because I've found that in ringing oscillators there's too much interaction between the frequency-determining components and the Q/resonance/decay components. But it turned out that the interaction can be minimized or completely eliminated by limiting both the frequency range and the maximum amount of Q that are possible. The parts values I've settled on allow tuning from what might be a high Bass, through low- and high Tom and up to Bongo or maybe even Wood Block. The maximum Decay available is more than adequate for percussion.

MY modifications were done with 9Volts powering the board. A higher supply voltage shouldn't require any changes. For +5V the board requires, I installed a voltage divider consisting of two 47K resistors and a 22uF electrolytic. This is good enough while the board is somewhat empty, holding only the Synth Drum, Snare and output op amp IC3 components. As I add the other voices I'll need to add a 78L05.

Note also that the parts identifiers in the following description are from the full Drum Tone schematic, and not from the PAiA page that explains the Synth Drum.

Further Note that the method used to control Decay has been greatly simplified in modifying the Conga voice and that modulating by LFO has been improved.


Warp diodes replaced with l.e.d.s 03/29/11

I determined that only a small portion of Q trimpot R28's rotation was usuable without the circuit breaking into self-oscillation. The usuable resistance was about 10k. I found the pcb conductor that connects one end of the trimpot to op amp IC1:C output (pin 8,) and cut it at the trimpot pin. Then I jumpered that pin to the trimpot wiper (the middle pin). I desoldered the lead of R27 that had gone to the trimpot wiper and bent it up vertically. I wired the then freed end of R27 to the center lug of a 10k potentionmeter. I soldered one "outside" lug of the pot to the hole I had made by lifting R27. To the remaining outside lug I soldered a 47k resistor and wired the free end of that resistor to the lead of R25 that connects to pin 8 of IC1:C.

Now with the 10k pot set for maximum Decay -- which should be fully clockwise when wired correctly -- R28 can be adjusted so the circuit doesn't self-oscillate. It needed re-adjustment as work progressed.

The next step was to get some control over the frequency sweep. I dislike extreme glissandos. They sound too disco. I removed diode D2, soldered two wires into the holes, and connected them to a 1Meg pot with a 10k resistor in series back to the hole connecting with D1's cathode.

(Update: The LFO option doesn't work as shown. This has been fixed in the modifcation of the Conga voice. Also, the 10k resistor labeled "temp?" was necessary when I was triggering the oscillator by touching +V to C13. I don't think it's needed when triggered by an op amp or other active device.)

To bring about control of the oscillator frequency, I desoldered R26, soldered a wire in the hole nearest to C17. The other end I soldered to resistor at the outside lug of a pot. The center lug of the pot was wired to the other hole of R26. (Ground.) I could have also gotten ground at the center lug of the Sweep pot. IT'S IMPORTANT that the pot be at the ground end of the series, otherwise the long wire going back to C15/C16 would pick up hum and noise, possibly causing false triggering. The values I chose for the fixed resistor and pot were 100k and 250k, respectively. Trimpot R28 was readjusted for no self-oscillation with maximum Decay at the lowest or highest Pitch settings.

Another enhancement was to change the character of the sound by adding crossover distortion (which for now I am calling Warp because "Distort" seems like something undesirable,) in the form of two diodes in the signal path to op amp IC3. I tried to do this with a spst switch but the result was too subtle, and I had to use a spdt. It's still too subtle, as far as I'm concerned, but I didn't want to cram any more components into this one voice. (Germanium diodes might've sounded better, but I don't have any.) Update: Tried Germaniums. Heard only a slight difference. However, Small, garden-variety red l.e.d.s vastly improved the "warped" sound. The first 22k resistor and the two diodes l.e.d.s were soldered to the switch's lugs. I desoldered R53 and put an upright 22k in the hole nearest IC3. IT'S IMPORTANT that resistor be there rather than on the switch to prevent hum and noise pick up by the op amp's inverting input.

The last feature added was optional Impact clicks, by capacitor coupling the trigger into the signal path.


Sound Samples

In the sound samples, the circuit was triggered by striking a piezo drumpad connected to an LM324 op amp in a configuration similar to that in the Drum Sensor board.

Low pitch with varied Decay
I think the low end corresponds with low- or mid-Tom?

Low pitch with varied Sweep

Different freqs with varied Decay, Sweep and dynamics. Also, Impact clicks have been added.
Note that when Sweep is in effect, lighter taps create lower pitches because a lower Sweep voltage is generated. This is heard best in the last group at 00:25:00 -- the controls were not changed, just the force with which the pad was struck. Bug or Feature?

Various pitches with and without Warp or Sweep Kinda resembles Conga or Bongo at times.

Fix for LFO Modulation

I found two problems when trying to modulate the pitch with an LFO, and both problems are caused by the very nature of twin-T "ringing" oscillators. First, just about any point in the circuit can be used to trigger it, so if the modulating signal's amplitude is too large, the circuit will continuously re-trigger at the modulation rate. Second, the twin-T configuration can also act as a bandpass filter, so if the modulating signal frequency is at an audio rate it will feed through to the output even when the oscillator isn't ringing.

When I modified the Conga voice, instead of diodes, I used an NPN transistor as the voltage variable element. This did not completely eliminate, but greatly reduced both the re-triggering AND the feed-thru. Here's how it would look if done to the SynthDrum:

The connection from R29 to the junction of C15 and C16 must be broken, either by cutting a pcb trace or clipping that end of R29. The values of the Sweep pot and R29 affect the amount and duration of Glide. Increase C17 to get a longer glide time.