NOTE: THIS ARTICLE IS UNDER SUPERVISION, THERE IS AN ERROR I CANNOT DETECT IN THE CIRCUIT
Welcome to acidotunismo.com!
This is the first article I post, I hope many will follow. Basically it’s a site about DIY (do it yourself). It’s a sort of open log of some of my projects for anyone that could be interested. It’s in english and spanish.
What does it do
This circuit is a mod/hack to Korg’s Kaossilator (v1) to extract clean rythm pulses (BPM) from it and be able to control other home-made devices and synchronize them. Specially practical for the “tapping” BPM control and sync’ing with the automatic drums (P.90-P.99).
The circuit is electrically autonomous, feeding from the Kaossilator itself, generating 0 to 6V pulses approx.
It’s not exclusive to the Kaossilator, it can also be used to extract pulses from any LED integrated to a device.
Mainly it uses two BJT transistors to detect a negative and positive from a LED inside the digits display, then it goes two a monostable circuit using the famous LM555 to generate a reliable positive pulse and “absorb” small (fast) changes produced by PWM and digit iteration (see Why it works ahead).
Download LED pulse extractor Fritzing file (.fzz) (schematic & breadboard)
I’ve seen in other tutorials that they take ground (GND) from the battery spring, but that’s not optimal as pluging an external power source, the socket disconnects ground from the spring making it effectively disconnected from the rest of the circuit. The pins I found best to take as power source work for battery power and from an external source/transformer.
Int the next image it’s shown where you must solder cables inside the Kaossilator to use the circuit. Here’s a tutorial (third party) on how to open it: http://www.youtube.com/watch?v=wiZhB_BpINY
I’ve detectet that, no mater which condition, the Kaossilator turned on and off a LED in the numerical display with each rythm pulse and I imagined it could be extracted electronically.
As I don’t have an oscilloscope I’ve tried with a computer speaker: The sleeve of the TRS plug connected to circuit ground and the tip of the plug through a resistor to different pins on the numeric display to hear which one sounded more when the BPM LED was turned on. After detecting one, leaving the connection there, I started trying other pins with the plug’s sleeve to see which pin improved the signal. A common circuit bending technique.
Afterwards with a 9V battery and a generous resistor of 10kΩ (just in case) I tested and effectively with those pins, in the orientation thought to work, turned that led on.
Why it works
This is a more complete and technical explanation of the inner workings of the circuit.
7 segment displays + dot inside are usually 8 leds that share a common anode (+) or cathode (-).
Most of the times, to economize pin counts in controllers, a method better than one pin per led is chosen. All anodes (+) of LEDs of the same segment on each digit is shared, but each digit has it’s own cathode (-). They change at high speed, faster than what the eye can perceive, iterating each cathode.
The Kaossilator’s display has 3 digits monolithically sharing pins. With the classic method it would need (8 anodes) x (3 digits) = 24 pins, but optimized we have (8 anodes) + (3 cathodes) = 11 pins, cathodes are shared by each digit and anodes are shared by each segment on the digits.
The NPN and PNP transistors work as gates. When the NPN base conected to the LED’s anode detects a positive, it opens. Same happens with the base of the PNP detects a negative voltage.
Logically it worked, but I was left with a doubt, because to turn on/open BJT transistors, the voltage in the base has to be higher than the one in the emitter for NPNs or lower than the emitter in case of PNPs. In this circuit the emitters are connected to each other, apparently being an “what came first: egg or chicken?” situation. I’ve ended up testing the circuit in a LTSpice IV simulation to try to understand what was going on.
Download the simulation file for LTSpice IV: LED sensor SPICE
I’ve concluded that leakage currents were filtering in both transistors, generating a snowball effect until both transistors open completly.
This afterwards gos to a monostable LM555 circuit (explained in thousands of sites) to generate a stable pulse.
The fastest rate the Kaossilator handles is 300BPM = 5 beats per second, or 1/5 seconds per beat.
Half of that period is good for us: 1/10 seconds.
In between, the circuit absorbs pulses from the iteration of digit cathodes and PWM used to lower the intensity of the digits (and save power).
According to calculations for the 555 monostable, we have that the period T = 1.1 x R x C. I used 100kΩ y 1μF, ergo T = 0.11s almost the 1/10 seconds I was aiming for. Good enough.
My other DIY devices generally use 9 volts. To be able to connect them to this device that delivers 6V, a PNP transistor could be used as the right side of the schematic shows.
I hope this was easy to understand and worked for you, thanks for reading!