[sdiy] sync conditioning/simple triangle oscillator without linear voltage control

Tom Wiltshire tom at electricdruid.net
Sun Feb 24 00:30:57 CET 2019


This looks pretty good, Brian:

https://www.hobby-hour.com/electronics/s/oc2-octave.php

The advantage of the peak-based detection, rather than zero-crossing, is not so much in the accuracy (since that’s probably much the same) but rather the *phase* of the result. The reason this is important is as follows:
One way to generate an octave-down sound is by flipping the original input waveform at half its frequency. This generates the required sub-harmonic but maintains more of the original information than simply replacing it with a square wave from a flip-flop. But for this method *when* you do the flip is clearly important, and the peak-based method is generally considered to produce a better result than the zero-crossing method. Your tastes may differ, obviously…


HTH,
Tom

==================
       Electric Druid
Synth & Stompbox DIY
==================

> On 23 Feb 2019, at 20:43, rsdio at audiobanshee.com wrote:
> 
> I did not realize that the Boss OC-2 uses peaks rather than zero crossings. Still, the OC-2 suffers from the same common problem where its output frequency will jump octaves when the input harmonics change in relative strength. In other words, sometimes the 2nd harmonic is stronger than the fundamental, and then the output will jump up an octave. Due to the constantly changing timbre of plucked strings, the 2nd harmonic will eventually fall below the strength of the fundamental, and the output will jump back down. I’ve experienced conditions where these jumps will occur multiple times per note, meaning that the amplitude ratio between fundamental and 2nd harmonic is changing drastically.
> 
> Is the schematic for the Boss OC-2 available? I’d personally like to take a look at how they do peak detection on a cycle-by-cycle basis in the analog domain.
> 
> On that note, is there any evidence that peak detection is better than filtered zero crossing detection? It’s a fact that harmonically rich waveforms cross zero more than once per cycle. Just add a sine wave with its 2nd harmonic - if the 2nd harmonic is within 6 dB of the fundamental (and especially if they have the same amplitude), you’ll see the waveform dip below zero twice during the cycle. That said, harmonically-rich waveforms can have two pseudo peaks per cycle. It’s sometimes hard for even a human to determine which peak is “the” peak, so it seems likely that an analog circuit would have difficulties distinguishing peaks that are very close together (unless, of course, the circuit is highly tuned to expect a narrow range of frequencies).
> 
> A good paper to read is “Pitch Extraction and Fundamental Frequency: History and Current Techniques” by David Gerhard, 2003 (not that I’ve finished reading it!)
> 
> Brian
> 
> 
> On Feb 22, 2019, at 10:08 AM, David Moylan <dave at expeditionelectronics.com> wrote:
>> This issue has been tread a lot in the realm of guitar "octavers", mainly octave down effects, and guitar synthesizers.  The most popular circuit for guitar octavers seems to be the peak detector circuits rather than zero crossing detection.   Google the schematic for the Boss OC-2 for that type.  Some guitar synthesizers like Roland's GR-300 used zero crossing detection, but had separate processing per string which makes the job easier since you only need good tracking for ~2 octaves.  They also used filters that would switch center frequency based on the detected pitch.
>> 
>> The OC-2 method is certainly more economical circuitry-wise.  I've been curious lately about the history of that variable threshold detector.  Earliest use I could find seems to be the ElectroHarmonix Octave Multiplexer from the seventies, but couldn't find any patents related to it.  Anyone know of earlier instances of the same or similar circuits?
>> 
>> Dave
>> 
>> On 2/22/19 8:56 AM, ulfur hansson wrote:
>>> hello richie,
>>> 
>>> I already have constructed an instrument that works quite well, using high powered class D amplifiers and custom wound electromagnets to induce feedback in an acoustic set of strings. here is a brief description of my harp, along with demo recordings; https://ulfurhansson.com/SEGULHARPA-ELECTROMAGNETIC-HARP (kind of an acoustic synthesizer/organ?)
>>> 
>>> the internal circuitry has patchpoints for experimenting with additional processing/signal conditioning. right now I am using waveshapers to change timbre/harmonics as you press harder/lighter on the capacitive touch sensor keyboard on the front panel.
>>> 
>>> it works great, but i'd still like to explore other avenues for varying timbre control. I feel my idea of a synced oscillator tethered to each string is exciting as the sync artifacts will provide a much richer harmonic response, that is if i can figure out how to implement it with limited space and solve the challenge of signal conditioning before SYNC input.
>>> 
>>> it doesn't have to be perfect, and sync errors as the signal grows stronger could actually be an interesting addition to the instrument timbre. I would probably add a VCA to the oscillator output to control the level of output mixed into the feedback signal depending on how hard you press against the keyboard (aftertouch).
>>> 
>>> thank you for your thoughts on this!
>>> all the best,
>>> -úlfur
>>> 
>>> fös., 22. feb. 2019 kl. 11:43 skrifaði Richie Burnett <rburnett at richieburnett.co.uk>:
>>> Synchronising an oscillator to a vibrating guitar string waveform might be quite challenging. Different frequencies travel along tensioned strings at different speeds leading to a dynamic waveform with likely many zero-crossings per cycle. This is particularly so just after plucking where string tension is highest and high-frequency partials are loudest.
>>> 
>>> Attempts to extract pitch periods from harmonically complex waveforms based on zero-crossings are usually of limited success unless you can either heavily filter the input waveform to remove harmonics, or use something like a PLL that will tend to inherently reject harmonic energy in its reference signal if you slug the loop filter enough.
>>> 
>>> Are you trying to make something like the e-bow?
>>> 
>>> -Richie,
>>> 
>>> ---- ulfur hansson wrote ----
>>> 
>>> hello list,
>>> 
>>> I need to find a triangle oscillator design that meets two requirements; 
>>> 
>>> 1 - it needs to have a sync input that can sync to a signal of varying amplitude ( it is actually a single guitar string)
>>> 
>>> 2 - it has to be tiny! I plan to design it using 0805 caps/resistors and SOT transistors etc...
>>> 
>>> they idea is to use this oscillator to reinforce a feedbacking signal induced in the string using electromagnetic pickups and actuators. I have already constructed an instrument that works great, but I'd like to use this addtional circuitry to accentuate different harmonics of the resulting tone by mixing the synced oscillator in with the feedbacking string signal.
>>> 
>>> since the sync input signal is not a steady pure waveform,i will probably need to do some signal conditioning - any ideas regarding comparator designs for converting varying amplitude signal to steady square wave would be greatly appreciated! perhaps a simple diode compressor would be helpful too...
>>> 
>>> I hope this topic doesn't come across as too nebulous, but really any ideas or helpful thoughts comments would be greatly appreciated!!
>>> 
>>> all the best,
>>> -úlfur
>> 
> 
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> http://synth-diy.org/mailman/listinfo/synth-diy




More information about the Synth-diy mailing list