WASP Keyboard

I began troubleshooting my EDP WASP clone (Jasper) keyboard, which I built a few years ago. However, the keyboard was not tracking accurately which I thought to be a feature. After servicing EMS and Serge capacitive keyboards, I observed that capacitive keyboards can track both accurately and quickly. This led me to believe that I could also get the WASP functioning smoothly. Indeed, after I was done, my keyboard tracked perfectly, with no issues or false triggering.

I discovered that the reason my Jasper wasn’t tracking correctly was that the contemporary CD4069UB behaves slightly differently than the NOS CD4069 from RCA. With the new chip, tracking is glitchy, and turning the sense potentiometer all the way up doesn’t cause autotrigger as it should.

Replacing IC35 with a NOS chip allowed me to calibrate the keyboard correctly. It may be possible to modify the circuit to make it compatible with the new chip as well, but I didn’t spend much time on this since I had few NOS chips in my possession. While WASP has many CD4069 chips, I currently see no reason to replace any chips other than IC35 with a NOS version.

For anyone servicing a WASP, here is my description of how the tracking circuit functions. Jasper is using the same component numbering as the original WASP, so simply obtain either the original or redrawn schematic from any source and follow along.

The circuit surrounding IC35 (CD4069) gates A and B functions as an oscillator operating at approximately 100 kHz (10 µs). IC46 (CD4016) is used to disable the oscillator when the pad multiplexer is advancing. Oscillator is disabled by bypassing the feedback resistor R136.

In WASP CD4069 is utilized in amplifier mode across multiple locations. In this configuration, the chip functions similarly as an operational amplifier in an inverting setup, with the non-inverting input connected to VCC/2. This arrangement allows the chip to operate between +5V and ground without the necessity of generating a separate reference voltage point.

The keyboard pads are connected through multiplexing ICs to the junction between R135 and R136, one at a time. Your body will act as a capacitor, which, for some reason, causes the amplitude of the waveform to decrease.

In the oscillator plot below, the digital +5V signal at the top is Q10 on IC32 (pin 14). The blue signal at the bottom is probed from pin 4 of IC46 and is biased to approximately 2.5V. You can observe individual oscillation bursts that correspond to each keyboard pad. The first burst, which aligns with Q10 being HIGH, represents the lowest C note. This is followed by the highest C note, after which the sequence descends chromatically. There is a gap between middle C and C# due to the note encoding circuit. The highest C pad is being pressed, as indicated by the amplitude of the second burst.

Port B on IC35 is utilized to expand the waveform. Due to the feedback loop, the amplification is not perfectly linear, which is advantageous for this application. There is an optimal resistance value among R140, P5 (the sense potentiometer), and Jesper’s additional trimmer, all of which are connected in series between pins 2 and 3 of IC35.

The yellow signal at the top of all these plots represents the Q10 output of IC32. The signal at the bottom signal is the output of the tracking oscillator, pin 4 of IC35. In the plot above, the keyboard sensing is calibrated correctly, resulting in the burst from the active pad being significantly louder than those from the other pads. In contrast, the plot below illustrates that when the sensing is calibrated too high, the bursts from both active and inactive pads become identical and sufficiently loud to trigger the gate.

At port B output (IC35|4) the waveform remains biased at 2.5V. Port C is then utilized to slightly amplify the signal while also shifting it closer to 0V by mixing the oscillator with +5V through R139. This adjustment ensures that only the active pad oscillation reaches the CMOS threshold, which is approximately 1V. Without this offset adjustment, the signal at the IC44 input would consistently remain HIGH. Replacing R139 with a 100k trimmer would allow for fine-tuning; however, a 47k resistor is close to the optimal value.