Thursday, December 19, 2013



This project has been long overdue. the motivation behind it was to create an easy to use, easy to build, and easy to sell drum-synth. the drum synth would be based around my coveted collection of radioshack "elephant keyboards". over the past several years i have been snatching up these keyboards off ebay when ever i see the chance. i have been known to spend $60+ on them. it's kind of funny too because the people who sell them can't understand why i am offering so much money for a child's toy. the reason is that inside the elephant keyboard is a pretty decent monophonic toy-organ chip, as well as an amazing little two-voice rhythm chip. the toy organ chip is similar to the ek001in that it has an envelope I/O,
vibrato I/O, and the two-octave keyboards' switches all share a common bus(ground). it is monophonic though and there is only one timbre: 1/1 square wave. one unique aspect of the toy-organ chip is that the clock oscillator only seems to run when a note is pressed. i found this useful for creating a gate output since the oscillator always returns to the high state once the note has completed, whereas the audio output's initial state changes from high to low depending on the frequency of the last note, and the note before that, and so on... generating a trigger was pretty simple since there is an envelope I/O. not as simple as the ek001 but nothing too complicated. the rhythm chip is really what is cool about this toy keyboard. the chip is separate from the toy-organ chip and has its own clock and everything. despite the elephant keyboard only having 4 accessible rhythms, there are actually 4 additional that are not being used. i haven't compared them side by side but the 8 rhythms sound pretty close if not the same as the ek001's. the main difference being that there are only two drum sounds where the ek001 has 4. the rhythm chip has a start/stop, tempo up, and two tempo down inputs. why two?.. the rhythm chip has individual outputs for each drum sound which is pretty sweet! some older models ek001 variants have this but one thing that NO ek001 has is envelope I/O's for individual drum sounds. that's right, there are 2.2uf capacitors to shape the decay of each oscillator. one of the oscillators is a pretty straight forward square wave that generates the kick/tom sound. the other is a pseudo-noise oscillator. through the control of a 1meg resistor, the decay of each voice can be modulated individually. pretty cool.

well, after years of hoarding up all of the elephant keyboards i could find, it was finally time to use them. in the past i used a elephant-keyboard rhythm-chip to drive a little sequencer. it was really great because the sequencer was driven by the drum sounds themselves so changing the envelope of the two sounds would then change the sequence pattern. i definitely wanted to recreate that effect but i also thought it might be kind of cool to use the sequencer to modulate the envelope times of the drum sounds too to create endless pattern variations. not to mention causing the two drum voices to sound like 8 or more different voices. in the end i decided there would be two four-step sequencers running in parallel, driving themselves to one of the preset rhythms. an 8-step sequencer would also be driven by the pattern, and would control the pitch of the toy-organ chip. the toy organ oscillator is clean and holds notes but it doesn't play very well. at lower pitches it even exhibits a little latency/gating. for this reason i decided it didn't need really nice big player keys. instead i just used the ones that came with the keyboard. i figure they will only really be used to select the root note of the sequence. however the keyboard can be used without the sequencer.  
the main voice coming from the chip is a nice clean square-wave that is ready for shaping. i decided to use this project as an excuse to try out this idea i had been thinking about. in the past i have tried to make single oscillators sound bigger. using frequency dividers helps a little but its not what i'm looking for. no matter how i divided the frequency, the oscillations would always be in sync. i think that what makes analogue oscillators so big sounding is that while their frequency is the the same, the oscillators are out of phase with each other. i know, pretty basic synth stuff here. the point is i wanted to be able to take one oscillator and turn it into two. in my limited experience with synth-repair, i have learned that this is called "chorus". basically you send the oscillator through a delay chip(pt2399 in my case), and the out put is then recombined with the original sound. the phase relationship between the two oscillations will depend on the clock rate of the delay chip. to make it even better, add a subtle amount of low frequency modulation to the clock speed. i have always used pt2399 delay chips as echo processors, but this opens things up quite a bit. i have never really been too satisfied with the pt2399 echo either. it does the job but as a chorus it is more useful i think. 


i prototyped the circuit and was instantly in love. my next idea was to send each of the two voices though its own VCF(if you are still reading, you know what a VCF is). each VCF would have its own envelope generator that could be triggered(and re-triggered) by key-on-messages, or by divisions of the sequencer pattern. once i had tested and confirmed that all of the circuits would work, i began designing the enclosure for the keyboard. the design is not terribly original, but it is good and i know that i can build more of them easily. i used my sketchup demo to draw the box, then coreldraw to turn the drawing into parts that could be fastened together after being cut with the laser-cutter. once the box was glued, puttied, dry, and sanded, i put on red tolex. i only have two colors, red and powder-blue. once the tolex was on, i designed and cut the cheeks. for some reason i didn't think to look at the cheeks on the keyboard in sketchup before i cut out, stained, and varnished them. they look extremely good. definitely the best results i've gotten so far. unfortunately i didn't really like the way they made the rest of the keyboard look. they are just kind of.. big. not so big that they make the keyboard look bad, i kind of just fell like they make the synth look smaller or as small as it actually is. if that makes any sense. the next one will likely be different. or not, i don't know yet.

 next step was the circuit-boards. i have really enjoyed having a CNC-engraver. i really don't think i will ever go back to perf-board, even for the smallest thing. CNCs are just too much fun! i recently had to order a bunch of new engraver tips because the two that i had were totally busted. i decided to go for the finer pitched tips, thinking that if the tip breaks off, the diameter of the broken tip will still be usable. little did i know how much of a problem these new tips would be. after spending days drawing all of my circuits "vero-board" style, it was finally time to engrave my main-board. when it was finished it seemed fine. the etchings were very thin due to the new skinny tips, but the edges were very badly burred. this hadn't happened with the other tips. i used steel-wool, sandpaper, mini-files, and a utility knife to be sure that there were no shorts. everything looked good so i moved on to populating the board. once it was fully populated, i moved on to the circuit boards that would hold all of the hardware. i took extra steps to avoid getting the burred edge on the engraving, but it still need some work in the end. once all of the circuit boards were populated and wired, it was time to flip the switch. actually i think i was so nervous about turning it on that i subconsciously found ways to distract myself from it for a whole day. when i finally did turn it on, there were problems. apparently all that de-burring i was doing was lodging tiny threads of copper into the grooves and creating shorts all over the place. the circuit was turning on though, and it was making some strange sounds. so i was able to start debugging, which in this case was basically going thorough all off the circuit and clearing all of the shorts one by one. by the time i had solved all of the shorts and found/fixed certain mistakes in my circuit design, the pt2399 was not putting out audio and the keyboard would not re-trigger the envelope generators. also, one of the keys just stopped doing anything. no signal on the pin at all. both the pt2399 and the keyboard chip were fried... i don't mind losing a pt2399, but the keyboard chip was a real loss. thankfully the drum chip was fine, so i extracted it and will save it for another day. the keyboard chip is impossible to remove from the keyboard main-board. i am usually really good at extracting parts but this little guy was  impossible. instead i had to replace the entire keyboard main-board with a fresh one.

once all of the circuit was repaired and in working order, i began designing the face-plates. after a day or two of designing them, it was time to cut them out of acrylic. when i was ready to cut, i realized i  only had one sheet of clear acrylic left... ok, no mistakes! i was calm and calculated while setting up the job, but it didn't matter. one of the belts on my Y axis had gone limp and it caused the laser to mutilate my design... it was all for the better though. it's been too long since i've "got in touch" with my laser. regular tune-ups are critical. the next day i hit up TAP-plastics' scrap table. now i'm set up for at least a year. once the laser was fixed, i cut the face-plates out. they came out almost perfect. the only problem was that the protective sheet on one had peeled back, and the hot fumes created a scar on the face. i tried to buff it out, but that didn't work. then i tried to melt it out with my MAP torch. no luck there either. i only noticed the scar after i had painted the face-plates and removed the protective sheet. it was late at night so i didn't want to cut a replacement. i would have to wait til the next day. since it was already a loss, i wanted to see what would happen if i hit it with some 1500 grit sand paper. it got rid of the scar and made a really interesting texture on the acrylic. i kept going and buzzed the whole panel until the texture was consistent. it was a little too frosty so i hit it again with the MAP-torch. the result was a glowing/transparent/brushed-steel look. i loved it so much i woke up Heidi and made her look at it. she sleepily agreed it looked good so i did the same for the other one.

so now everything is ready to go. all i need is a good day or two to assemble it all. everything comes together quick and tight and i'm having fun. before i know it i only have one thing left to do; connect the power supply. for this project i am using one of my re-purposed switch mode power supplies that i find at the thrift store all the time. it has a 12volt rail, 5volt rail, and common ground. all i have left to do is connect the 12 to the 12, 5 to the 5, and ground to ground... i triple-check the terminals to be absolutely sure the voltages are going to the right place. then my dyslexia takes over, and for some reason i connect 12 to 5 and 5 to 12. i turn the keyboard on and hear some distorted garbles from the drum chip, then nothing. i turn it off and then back on... still nothing. i turn it off and check the voltages. backwards! i flip them back the way they're supposed to be and hope that nothing too devastating has happened. i turned it back on and found that the pt2399, keyboard chip, and the rhythm chip are all fried. i took the rest of the day off...

the next day i gutted everything to get at the main-board that held the pt2399. i disconnected the whole elephant keyboard and grabbed another fresh one. to my surprise the whole process was a lot easier than i thought it would be. the overall design of this keyboard is very accessible. once everything was back in place and working, all i had to do was wrap up. i quickly cut two back panels for the power jack, standby-led, and 1/4 inch output jack. i also had an extra buffer in my circuit so i made an extra pattern sync output. it is and active-high 5v logic signal directly from the 1/1 pattern. it's not midi but at least it's something to sync to. 

this keyboard will be for sale for a limited time at my FOR SALE PAGE. i plan to build more in the future but it might be a while before i have time again, so don't wait. check out the video for more info or email your questions directly to me at



  1. I love your designs...just wondering why you do not incorporate a resonance knob in your designs? I can clearly here resonance but at a fixed value.

  2. Sorry meant hear.

  3. i just didn't think it was too critical with this design. the resonance of these filters can be pretty unstable without proper filtering, which would mean additional circuitry, and space was a huge factor. this synth is smaller than it looks.
    thanks for your question:)

  4. Wow, I love your creations!

    As a synth collector AND an industrial designer I've always been fascinated by circuit bending however I'm not usually fond of the physical execution and poor design. You're instruments bring the art of bending to a whole new level. The level of design and attention to detail is very impressive. I would love to own one or more of them.

    keep up the good work.


    from British Columbia