Tuesday, September 17, 2019


Got a friendly message from none other than Mike Walters of Mystery Circuits. He had been tasked with repairing an odd-looking synth that resembled something I would make. Sure enough, it was the heavily modified Yamaha PSS-30 I had built on commission some 8 years ago. I let him know that the keyboard was probably a total mess inside, being that it was one of my earlier mods, and that it would be easier for me to have a go at it since I built it. The owner was OK with it, so he sent it my way. I always love seeing my old creations years later. I sometimes wish i could go back in time and help my former self with advice or some basic knowledge I've learned through experience. Needless to say, the PSS-30 was a total mess that not even I could decipher... the keyboard was still pretty much functional, with the exception of a few broken wires inside and some sticky switches. However, the audio signal path was just so dark and thin. I was not great at mixing signals at this time in my modding career. I really wanted to rebuild some of the circuitry inside, and get a good strong signal, but the pile of circuits all nested together were too annoying to try and separate and reverse engineer, so I decided to just rip everything out, and rebuild it all properly. The theory of the mods were all pretty basic, so I was confident I could rebuild it in no time. Of course it ended up taking a lot longer than I thought it would. I have tried to make time for electronics when I can, but being a dad takes all the best hours of the day. I am able to squeeze in a few hours after my kid goes to bed, but those late night hours are never as productive as a fresh day to myself. Oh well, I'll take what I can get. 

I was able to re-design and rebuild everything over the course of a few weeks. I ended up making a lot of changes to it to make it more stable and usable. For one thing, I replaced the 74LS624 pitch oscillator with an LTC1799. This gives the sequencer a lot more range, and the LTC1799 uses a lot less power. I replaced the crappy AR generator with 555 timer type. There was a knob labeled "release" that really just controlled the trigger length of the original AR generator. Since it had no use in the 555 circuit, I created a voltage divider and sent it to pin 5 on the 555. It acts as sort of a range control. At lower attack and release times, the range control will hold the envelope at its peak for a good while before releasing. I replaced the drum voice circuit, too, with an arduino with four PCM drum sounds. The original drum sounds were analog, but needed a gate-to-trigger circuit for each one, and the "accent" knobs were actually more of depth/decay controls. each drum sound would have to be fine tuned to be rebuilt, and I didn't think it was really worth it for those sounds. Instead, I programmed an arduino with some code from DSP SYNTH with some custom drum sounds. The "accent" knobs now control the pitch(bitrate) of the individual samples. Other than that, everything else is pretty much the same. At least functionally, that is. It was a lot of fun rebuilding this thing, and it was great to get reacquainted with my workshop. I am feeling confident that I will be able to spend more time building electronics and tinkering with new ideas, so keep watching for more to come. And thank you to everyone out there that is still reading my blog and writing to me. It is very encouraging to get your emails.








(incomplete schematics)
(just for reference)


Saturday, April 21, 2018


I was recently sent back this keyboard that I circuit bent some nine years ago. I remember spending a considerable(relative) amount of time making it, and in the end I was not very satisfied with how it turned out, but it worked, and I ended up selling it. More details on the original build can be found HERE. The keyboard had stopped working, so the owner contacted me to ask if I would repair it. I agreed, and before long, it was back home on the bench.
When I first powered on the keyboard, the first thing I noticed, other than no power was that the supply was being shorted, so I opened it up to see if it could be remedied easily. I was greeted by a massive pile of rainbow ribbon cable, perfboard circuits and lots of hot glue. It really took me back seeing that mess, and how I used to have to build things. However nostalgic I was feeling, I was in no mood to try and dissect the circuit and all of those wires. Instead, I felt it would be easier to just completely gut the keyboard and rebuild it, and it was easier, even if it took a good week or two to finish. 

I stripped the keyboard down to the single COB PCB, and plugged it into a breadboard. I then whipped up a VCF, two LFO's, an amplifier, an auto-power-off disable circuit, some summing and boosting amplifiers, a CMOS distortion circuit with PWM, and a bunch of CV modulation buffers for the pitch bend, cutoff, PWM, LFO's, and distortion. Everything was pretty simple an straight forward, so the circuit came together pretty quick. 

The original faceplate was looking pretty rough, and most of the hardware was nasty, so I decided to cut a new faceplate, and use all new hardware. It was pretty straight forward, but space was limited, and when I was trying to lift up the keys, one key broke off.. The broken key set me back about a day, but I was able to find a replacement set that was only slightly smaller, but nice and new. The new keys also gave me a lot more room to work with since the original keys were and old set of used casio keys on top of clicky typing keyboard switches with lots of hot glue. They worked well enough, but took up a lot of space. I decided to add dust cloth to the back off the keyboard, since the PCB would be located pretty much under the keys, and debris could fall through the keys and compromise the circuit. Dust cloth is pretty useful. I also used some for the speaker cover after noticing the speaker magnet was sucking up all kinds of cut-off component leads, and making all kinds of buzzing.


Once the interface layout was decided, and the faceplate was all drawn up, I engraved and cut out the faceplate on my lasercutter. I got it right in two tries. 

The PCB was straight forward enough, but it was a little difficult going back to single sided CNC engraved circuit boards after being spoiled with PCB software and board-house PCB's. I only needed one though, so the CNC would have to do. The main PCB ended up taking a couple of days to design, cut, and populate. In the end, there were very few errors, and they were easy to remedy since the traces are so big. That is one advantage to engraving PCB's on a CNC.


The keyboard's new functions are essentially the same as they were before. Maybe a little bit more reliable though, and more effective. The keyboard chip on its own has several functions. Some are more redundant, like volume up/down, and demo songs, so I left them out. In fact, the auto-power-off disable circuit is just an oscillator the repeatedly triggers the volume up switch. When the keyboard reaches its maximum volume, the switch makes no sound, and resets the A-P-O circuit continuously.
the functions I did include are the eight drum beats, sixteen timbres, tempo up/down, and record/playback. All of the functions are selectable with the eight push-button switches. The modes of the push-buttons are selected with the respective right or left shift switch. The four buttons on the left can select rhythms 1-4, or playback, record, tempo up/down. The four buttons on the right can select rhythms 5-8, or the 16 timbres with a combination of of two button presses. The lower octave of keys can also be shifted to one octave above the higher octave. This was a feature on the original mod that I though should be included since there are only two octaves of keys.
The keyboard's main voice output is sent to an amplifier circuit to be scaled for the VCF input. the output of the amp is also sent to another amplifier with very high gain. That amplifier sends it's signal to a couple of NAND gates that have a sort of PWM filter. The effect is a heavily distorted signal that has a sort of high-pass filter, or "d-color" as I've called. The distortion signal is mixed back with the clean signal via "clean" and "distortion" volume pots before being sent to the VCF. The VCF is my go-to LM13700 12db resonant LPF. The VCF signal is then sent to an LM386 power amplifier to drive the output. There is also a line input that is summed to the main keyboard signal, so external signals can be sent through the distortion and filter circuits in the same way the keyboard is. The line in on the original mod did not work very well, from what I remember, but this is a huge improvement. There are two LFO's for modulation. Each has triangle and square wave outputs, and the LFO rates are controlled with a joystick. The LFO's have depth control pots, depth/frequency indicator LED's, and four-way select switches to modulate CV parameters. LFO1 can modulate "d-color", "pitch", or "cutoff", and LFO2 can modulate LFO1 rate, "pitch", or "cutoff". The joystick can also be set to modulate any combination of cutoff, pitch or d-color as well.

This keyboard is pretty sweet now. It is a whole new beast, and I am glad that I was able to finally get this keyboard to work in a way that I am actually satisfied with. Over the years I have taken on some pretty huge and long winded projects that this original mod doesn't even come close to, but at the time, this project was pretty defeating, and I think that stayed with me in a small way. It is nice to see this new thing connected to that time, and to see how far I have come since then. Now I must say goodbye again, and return it to its owner. See you again, maybe in another nine years :) 

Sunday, March 18, 2018


UPDATE: 4/9/18






 Yikes!!... Nearly two years since my last post.
Well, i haven't had too much time for electronics since i went back to work as a cook. Selling my custom instruments for three years didn't really pan out, and i was forced to rejoin the wage-earners. Not really a bad thing, though. It is nice to be around people again. Especially cooks. They(we) are the best(worst) kind of people you will ever meet. 
Oh yeah, i'm a dad now too. Needless to say my time for electronics has become extremely scarce, but i have still found time for little things like repairs for friends, developing circuits for future projects, and also a lot of circuit bending digital cameras. More on that later...

About a year ago, before i found out i was going to be a daddy, work in the kitchen was slowing down. My boss was having trouble finding hours for me, so i was only working about two days per week, which was fine with me at the time. I had more time for electronics. In fact, a lot of time. I had so much time i decided to come up with a new NT model, and really try to do it right so that i could make a lot of them. Coming up with the circuit was one thing, but i wanted these new noise-makers to be clean and organized. I also wanted to be able to fit as much as possible into them. So i sat down and learned how to design my own printed circuit boards(PCB's) with some really great, free software, KiCAD. I highly recommend this software if you aren't already using something else. It is very easy to learn, and very easy to draw your own parts, which can be very useful.
So as long as i am designing my own PCB's why not go surface mount? Hell yes! I made a small investment in surface mount parts from ebay. This is really the way to go. The parts are insanely cheap, and very easy to store and keep organized. No more piles of cut-off component leads all over the place. 

(SMD parts... thousands and thousands)

Once i was confident with my first PCB designs, i sent them off to be made in china by Seeedstudio. They are super cheap, and the quality is very nice for the scale i am working at. The only catch is the shipping cost, but at least shipping is fast, and they track, so it's worth it in the end. It can really add up though when your design has mistakes and you have to keep reordering several times like i did, but i guess that is just the price you pay for this kind of thing. In fact, it would have been a miracle if it was perfect the first time, and i knew that going in. 

So, the plan was to come up with a super slick project file for the NT04 so that i could build them easily, even if years had gone by. I generally try to take good notes on all of my projects, and update them as i go, but it is sometimes hard to know what is important information for the future when you are immersed in the moment. So i would keep an inventory list of all of the components needed and have all files and diagrams updated and organized... Yeah right, still working on that. 

(sanding, painting boxes)

Work picked back up in the kitchen, and i had to put the NT04 on hold for a whole year. I had nearly assembled six by the time i put the project on hold. They just needed main boards and wiring. Thankfully i had successfully built one working unit before the hiatus, so at least i knew the design worked. 
Fast forward to now, i am only working two days per week again, but i am a Mr. Mom now. There is still not much time for electronics, but there has been enough time to finish the six NT04's i started. 

Ok, so what's inside the NT04? 
 The NT04 is a programmable 8-step sequencer with 8 touch sensors, and one tuning knob. When you touch a sensor, the sequencer will jump to that step, and the step will be held as long as you are touching it. When you let go of the sensor, the sequencer will continue from that point. The sequencer can also be turned off with the run switch(run), and the touch sensors can be used like a keyboard. When the record button(rec) is on, and a touch sensor is pressed, that sensor's respective step will record the setting of the pitch knob(ptch). Each step has its own LED indicator. The sequencer can run in either direction depending on the up-down switch(u d). If the sequencer is in down mode, reverse mode(rvrs) can be enabled. Reverse mode is simply another clock signal that toggles the sequencer between up and down mode. The rate of the reverse clock can be set with the "rvrs" knob. Both the sequencer rate and pitch setting can be controlled with the Y axis of the joystick when each's respective "joy" switch is toggled. The sequencer clock rate can be modulated with the sequencer's recorded steps. The "sqnc" switch sends the CV sequence to the sequencer clock VCO for some weird swinging rhythms. 

The sequencer modulates a pretty basic square-wave VCO (VB). VB has four modes; normal(3), -1 octave(2), -2 octaves(1), and XOR. XOR is kind of like a ring modulator but for square waves. VB is the carrier, and VA is the modulator(or vice-versa). VA is PLL VCO that tracks the frequency of VB. The tracking filter of the PLL is intentionally designed to be inefficient for its wild phasing, chirps, buzzes and other strange but consistent anomalies. The VA track knob adjusts the tracking filter from jittery-fast to portamento-slow. The VA tracking filter also has two additional filter modes, "B" and "C", which are really just different sized filter capacitors for more crazy ranges for the PLL VCO. If that wasn't enough, the PLL VCO also has four octave ranges that can be set with the VA "range" switch.  VA-4 is matched with VB-3, and VA 3-1 are all octaves up. The VA tracking filter can be modulated by the X axis of the joystick when its "joy" switch is toggled. Both VA and VB are mixed together with simple on/off switches (VA and VB) before being sent to a 12db lowpass VCF. 

The VCF has a cutoff knob and a resonance knob. The cutoff can be modulated by the X axis of the joystick if its "joy" switch is toggled. The VCF can also be modulated by a simple envelope generator. The envelope generator has an on/off switch and a decay knob. The depth of the EG is set to mid range, so the initial cutoff setting sets the overall range. When the sequencer is disengaged, the EG can be triggered by the touch sensors. The decay of the envelope generator can be modulated by the X axis of the joystick if its "joy" switch is toggled. 

After the filter stage, the signal is sent through a volume control knob, and out to the speaker or 1/4 inch line out. The NT04 can be powered with a 9 volt battery or a standard 9 volt adapter. Additionally, there are two clock sync jacks. One is input, the other is output. The output will send either clock signals or trigger signals, depending on whether sequencer is in "run" mode or not. Any signal coming to the input will advance the sequencer whether the sequencer is in "run" mode or not, the only difference is that when in run mode, the input signal can trigger the EG. Any signal input will be sent to the unit's output, so syncing multiple units in series is possible.
Wow, what a mouth full!


(sanded box unpainted)

(gut-shots and failed prototype on the left)




Here is a quick video that covers some basic functionality and sync-ing options.








(Making 9 volt battery compartment work)
These plastic 9 volt battery compartments can be pretty useful. Unfortunately they do not come with terminals, and using wire-lead snap-connectors makes changing the battery a pain, so installing terminals is a must. First i created a drill hole guide for the terminals. Then using battery terminal springs, m2 screws, m2 lock washers, m2 nuts and solder tab eyelets(all found on ebay), bolted the terminals to the bottom of the battery compartment. Lining up the screw through the spring and into the hole inside the compartment was tricky, but i found the best solution was to use a little piece of tape to wrap the screw and screwdriver together, and then push the screw through. More effort, but saved time in the long run.