Tuesday, October 28, 2014


This project was supposed to be a quick quick one, but it ended up taking nearly the entire month of October to complete. it was't terribly difficult to build, but this month has been so full of distraction. initially all i had intended on doing was modding up a toy keyboard with some simple mods that could be built quickly, and would yield some interesting sounds. i started with a Rhapsody programmable electronic organ. i forgot to take a picture of mine before i totally dismantled it, but i did find a pretty good picture of it on google images.
The keyboard comes equipped with a built-in sequence recorder and rhythm generator, so i thought it would make a great base for a novelty synthesizer. the rhythm generator was actually just two separate pulse generators that were filtered slightly before being sent to the amplifier. my first thought was to add a step sequencer that could be counted up by one pulse-generator, then counted down by the other. it seemed to work, but for whatever reason, i decided not to include the pulse generators or the internal sequence recorder in the end. i think the reason was that the pulse generators made a popping sound whether they were connected or not. i only wanted to use them as a clock source, and not hear them at all. they could also be heard when the recorded sequence was playing back, so i just got rid of both features and moved on. 

Keyboard case cutting before and after.

I came up with a rough idea of what i wanted the keyboard to do, and breadboarded the circuit. the keyboard pitch would be sent through a bi-directional 8-step sequencer. the sequencer would also control the pitch of a little drum voice chip i found in the basement. it was an HT82105 from an old drum toy from the 80s. the chip has four drum sounds that would be triggered by the sequencer clock. since the pitch of the drums would be sequence by the same hardware that pitches the keyboard, i thought it would be a good idea to have the two voices be pitched in opposite directions. so basically, pitching the keyboard voice up will effectively pitch the drum voice down. the drum chip is kind of unique in the fact that when the voice is triggered, the drum sound will repeat as long as the gate is high. at higher pitches, the voice is able to repeat more than at lower pitches. to get a little more versatility from this effect, i added a duty-cycle control to the sequencer clock to be able to adjust the gate time to the drum voice. 
The melody voice would be sent to two CD4046 PLL circuits. one would output 1/2 and 1/4 the frequency of the keyboard voice, and the other would output X2, X4, X8, and X16 the frequency of the keyboard voice. each PLL circuit would have individual filter controls, to modulate the tracking speed of the circuit. one of the great things you can do with PLL chips is drastically reduce the tracking speed and efficiency by replacing the filter resistors and capacitors with ones that are not nominal. the result is squirly, phasey, rough and beating tones reminiscent of circuit bent sounds, but much more stable and not overdriven. the voices have a mind of their own, and that is what i want!
Anyway, then the two PLL voices would be mixed together with the original keyboard voice via switches, then sent to a 12db lowpass filter before being sent to the power amplifier. the filter would be slightly resonant, and the cut-off frequency could be controlled either manually, or with an envelope generated by the sequencer clock. there is not really any decent way to shape the modulating envelope to the filter without effecting the sequencer speed and the drum trigger gate time, but because there is a duty cycle control, the envelope can be changed. everything being connected in some way is kind of a recurring them in this synth.

Schematics and PCB layout.

Once all of my ideas were on a breadboard and working, i drew up some circuit boards and faceplates. surprisingly the circuit boards came out with no problems at all. i think i'm getting good at this. the one thing i forgot to include in the faceplate design though, was the Key Hold switches. unfortunately there was no way to hold notes on this keyboard without making a physical connection to the note switches. i wanted to be able to hold any of the 25 notes, so i would have to use multi-position switches. the keyboard chip has 8 data lines, and 4 common busses, but the 4th is only responsible for one note and all of the other unused functions, so i left that one out, so the high F can't be held. also, because the PLL circuits need a gate signal to disable their VCO's while a note is not playing, i used the gate signal coming from the 4th common buss. unfortunately though, the PLLs cannot be gated when the high F is played, only the original keyboard voice will sound. 
Anyway, to hold the 24 notes, i used a combination of a single pole four position switch with a single pole eight position switch. the 8-position switch will select the note address, and the 4-position switch will select the one of three common busses, or off. unfortunately the 8-position switch i used is a "make before break" type, which means that the note cannot be changed with the 8-positions switch alone, the 4-positions switch will need to be turned to activate a changed note. sadly, i didn't realize this until it was already built. good to know for next time though. 

Sequencer board on the right.

Fitting everything in the keyboard enclosure was kind of a challenge. there was not really a lot of room to work with, and it all got filled. the finished keyboard is actually pretty heavy for its size. i had thought about painting the keyboard enclosure a different color just because of how old and sun-stained the white plastic had become, but then i started to like the stains, and started to think that painting the whole thing might just make it look too "arts-and-crafty". it's not like it wouldn't be obvious that i painted it. also, it's been super wet here in Portland lately, and that has caused me more than a few paint-job heartaches in the past. best to let it be itself as much as possible. i am glad i did, because i don't think i could be happier with the way it turned out. 

 Tight fit.


Tuesday, September 30, 2014


This project started shortly after completing the NT02 series. i wanted to get back to experimenting with the CD4046 PLL chip, and i thought it would be kind of cool to use a toy enclosure as the body for the new synth. i shuffled around in my basement for a while until i found the perfect toy. it was an old radioshack animal sound keyboard. originally i had planned to use the internal circuitry of the toy, but soon realized that the sound quality would not be suitable. instead, i grabbed a little toy keyboard i had laying around, and began prototyping around that circuit. the keyboard was nice because the keys were all different colors, and they were small enough to mount in the toy organ body. i thought it was a perfect match since the keys kind of resembled a rainbow and the organ body kind of resembled a cloud. 

Once i had a rough idea of how i wanted the keyboard to look with the toy organ body, i began prototyping the circuit and coming up with a general idea of what the final product would do. i knew that i wanted to use a PLL chip to multiply the voice frequency, and then send that frequency to a pseudo-noise generator made from a 74ls273 shift register and CD4030 XOR chip. i was pleasantly surprised with some positive results, but unfortunately there were problems with the toy keyboard chip i was using. the toy keyboard was a very cheap generic toy keyboard. it put out a square wave voice for a clock-determined gate interval, so changing the pitch of the keyboard would effectively change the gate time of the voice. then an envelope capacitor would shape the output of the square-wave. the problem came with the gate time, since not only did the gate time dictate the length of the note, it also put the chip in standby mode each time the note was allowed to end. then when another key was pressed, the chip would have to power back on, and the voice would spike. the sound was especially noticeable since the signal was being sent to the PLL frequency multiplier, so i decided to scrap that keyboard chip, and go with one that was more compatible. i had an old purple DSI keyboard that had just the chip for the job. the chip had a square wave output that would stay on as long as a key was pressed in"organ" mode, or hold the last note continuously in "piano" mode, granted the envelope circuitry is removed from the circuit. the keyboard also puts out a gate signal for the length of the held note in both modes. the chip also had a record and playback mode, but i decided to leave those out since i knew there would be space limitations on the control panel. the only drawback to the chip was that there is a battery alarm sound that chimes after a minute or so of inactivity. unfortunately the battery alarm timer will only reset when a key is re-triggered, so re-triggering redundant functions like "play" mode and "piano" mode with an LFO was not an option. 
moving on, i came up with a general idea for the architecture of the signal path. the main voice coming from the keyboard would be split in two directions. one voice would be put through a frequency divider. the four sub-frequencies would then be mixed together with a joystick before being sent to a VCA. the other direction of the signal would be sent to the PLL frequency multiplier, then sent to the clock input of the pseudo-noise generator circuit. that signal would then be sent to another VCA that would be mixed to the other one. the two signals would then be sent to a PT2399 echo circuit. after the echo circuit, the signal would be sent to a resonant 12db LP filter, and finally the power amplifier stage. i then added two individual ramp LFOs to be able to modulate the VCAs and VCF. LFO "A" has "RISE" and "FALL" inputs to control the shape and speed of the LFO. LFO "B" has a rate knob and a three way toggle switch to set the shape to saw, ramp, or reverse-saw. the filter's cutoff can be modulated by LFO "A", but the LFO input is inverted from that going to the VCAs, so their peaks will be out of phase 180 degrees. this makes for a more versatile effect, i think. once everything was built and working on the breadboard, i designed, cut, and populated the circuit board for all of the new circuitry. i also started playing with ideas for the faceplate, cutting prototypes out of old cereal box cardboard. 

It was about this point in the project when i had received an inquiry into buying and modding my last NT02 and my long-for sale NOISE synth. see THIS UPDATE. that took my attention away from the keyboard for a week or so, and then i got really sick... i haven't been sick for a long long time, and this time i was laid out for a good ten days. my cold finally passed though, and by the time it had, i realized that all i really had left to do was wire up the hardware, cut a faceplate, debug, and assemble the keyboard once and for all. debugging took a few days, and the faceplate gave me a few problems, but eventually this little creature came to life. this little guy really turned out great!

 Re-wiring the key pads was tricky, but i've done this all before.

 Some guts-shots after all of the hardware was wired up.


 Blue LEDs under red lenses.

 Speaker in the back. remarkably loud, considering...

 Portamento knob on the side by the line out.

My schematic. WARNING: there may be errors.

Watch this

Wednesday, August 20, 2014


This is the latest set of NT(noystoise) series novelty synthesizers. following the overwhelming reception of the original NT01 series, it was pretty clear that if i build them, they will sell. the NT02 is simpler in many ways than the original NT01, but the sound and character are a bit more sophisticated than that of the NT01. the NT02 consists of a square wave VCO with chorus/delay, a white noise generator, a ramp LFO, and individual 12db resonant lowpass filters for the VCO and white noise generator. the pitch of the VCO is controlled by one axis of the VCO joystick, while the other axis controls the delay time of the chorus. the chorus circuit is basically just your typical PT2399 delay chip setup, except there is no feedback loop. the VCO is fed to the delay chip, and the delayed signal is recombined with the initial signal before the filter stage. the effect makes the VCO sound much bigger and warmer with the chorusing effect. almost like an old analog poly-synth with detuned VCOs. the white noise generator is your basic two transistor type found in many old synth designs. a common issue with the two transistor noise generators is that they require at least 9 to 10 volts to operate. that's just fine if you are using a plug in power supply, but when using a 9 volt battery, the noise generator usually wont work. in my case, the noise generator wouldn't sound with any less than 10 volts. however, the noise generator doesn't draw all that much current, so i simply used a 555 charge pump voltage multiplier to get the circuit where it needed to be. the problem i ran in to on the breadboard though, was there was some cross-talk from the 555 charge pump oscillator to the VCO, so i only used the 555 charge pump method for the first NT02(revA). for the others i simply buffered the VCO, and made a charge pump out of that. this way, if there was any cross-talk, it would be harmonized and unnoticeable. however, once the NT02 revA was built and working, the cross-talk was gone, so i guess it wasn't necessary. both the VCO and the white noise generator have their own respective filter with resonance control. the cutoff frequency of both filters are controlled by their respective axis on the VCF joystick. the center switch on the joystick toggles either the VCO or noise voice on or off depending on which joystick switch you press. each voice will remain on or off until the switch is pressed again. this feature was not included in the revA NT02. its voices are permanently on. the NT02's LFO is ramp shaped and can be set to modulate one of four parameters, or turned off. the LFO can modulate either the chorus delay time, VCO pitch, VCO filter cutoff, or the noise filter cutoff. on the revA NT02, the VCO pitch setting controls the maximum threshold of the pitch joystick, whereas the revB VCO pitch setting is independent of the pitch joystick. the LFO rate is controlled by the knob above the VCF joystick. the NT02 also has a volume knob, a 1/4 inch switching line out jack that bypasses the built in speaker when plugged in, and a center positive polarity protected DC jack that bypasses the internal battery when plugged in. the internal power supply is regulated, so any DC adapter between 9 and 18 volts will work just fine. unfortunately, unlike the NT01, the NT02 does not have CV inputs or outputs. the NT02 revA(black one) does have two 0-5 volt CV inputs for the VCO and LFO, but the revB models do not because their circuits' operating voltage is 8vdc instead of 5, which is kind of an unusual number... i would have liked to have scaled CV inputs that could turn 0-5 volt CV input to the necessary 1-8 volt CV input that the VCFs, VCO, and LFO require, but space was too limited. the inputs are in there though. i still kind of regret not including the ability to connect CV inputs to the NT02, but it is just safer this way. these were not intended as modular synths. they can be easily modified though, for any advanced users out there. hopefully these little synths do as well as the NT01 did so i can start on an NT03... the NT02s will be available at my for sale page until they are all gone. have a look, tell your friends about this BLOG!!!
Thanks for reading,


REV-B PCB (a ton of parts)