Here are two bends for the Yamaha PSS-20. Feedback overdrive has an effect similar to pulse width modulation. The clock interrupt produces weird glitches, speed-ups, and drop-outs. A 1/8" switching output is also fitted. The build has been documented from start to finish in hopes it would help show the entire process.
Here's the keyboard before any modifications. It's a simple square wave monophonic keyboard with blip-bop drums, as well as an arpeggio option. Different tones and rhythms are selected by holding down the large green or blue pads and pushing a key. Let's take a look inside for good modifications.
After removing the screws holding the case together, the circuit board was exposed. There are two chips on the board: the custom Yamaha YM7108 synthesizer IC and a LM386S amplifier chip. In the photo above, the two rows of 9 solder traces belong to the synth IC and the 8 solder traces in a row beside it belong to the LM386S amp chip. When trying to find bends, these are the obvious places to start.
Because the YM7108 runs on a crystal clock, the usual method of changing the pitch by swapping a resistor is not possible. However, it is possible to glitch this type of clock by degrading the clock signal. In order to do this, locate the crystal on the board and follow the traces until it meets the CPU. The red dot on the YM7108 traces indicates the pin receiving the clock signal. By routing some of this signal to ground via a 25K potentiometer, the signal can be loaded until it produces glitch effects.
There are usually not many interesting bends with keyboards that use such a simple synthesizer IC. To find more bends, look to the analog sections of the circuit. In this case, the LM386S amplifier chip provides many possibilities. It is also advantageous that a datasheet is available for this chip. One effect that works well on this chip is feedback, which will produce overdrive and ringing.
Usually this involves joining the input and output pins with a potentiometer. The feedback circuit I've devised for this keyboard uses a different approach. According to the data sheet, pins 1 and 8 control the gain of the amplifier. By placing a 10uf capacitor across these pins, the gain is doubled. Interesting things happen when the output on pin 6 is feed back into the amplifier through the gain pins. I used a 10uf capacitor in series with a 25K potentiometer connected to pins 8 and 6. The potentiometer sweeps from normal amplification to oscillating blips at the extremes, with pulse width modulation sounds in between.
After finding the bends I intended to use, I needed to figure out how to fit the two potentiometer controls inside the case. This is not a large keyboard and it does not have much front panel space. A suitable cavity was located next to the speaker. In order to mount the two potentiometers, two screw support tabs had to be removed. The two areas in red were removed with clippers and smoothed with a lighter. Care was taken to avoid cracking the keyboard case.
Holes were drilled for each potentiometer. A small drill bit was used to start the hole to help avoid cracking and keep the hole centered. A cardboard shim was cut to size to level out the plastic the potentiometers would sit upon; this keeps the controls perpendicular to the keyboard. Note that the potentiometers are not flush with the sides of the case, as the bottom part of the case occupies the inner rim of the top part.
The potentiometers are attached directly to the plastic case by nuts. These are tightened with needle nose pliers and stay strongly in place. Plastic knobs are placed over the knurled shafts of the potentiometers and held in place by friction. I was able to find matching colors for this keyboard!
After the controls have been fitted to the case, wires are run to the different points indicated earlier. The controls are fitted before the wires to keep wire length to a minimum. This makes better use of the space and results in better sound due to a shorter signal path. Very little solder was used, as the wires sit under the key bed. It is important that stranded wire is used, as solid wire tends to break due to the flexing that occurs when closing the case. Components like the capacitor can be directly attached to the potentiometers.
Each new solder joint was reinforced with a tiny amount of hot glue. This provides extra strength to hold the connection in place when closing the case. Though not pictured, a switching jack was also added to provide a line output. It was placed between the positive line going to the speaker and grounded on the battery pack.