TripleFiveSynth: 555 synth, one osc per note

[alexanderbrevig]

Hello everyone!

Today I had one of those ideas that just makes me drop everything and have a go!

I want to build an oscillator that uses a single 555 per note in the standard range (C0 -> B8). Yes, that's 114 555s, and 115 trim pots. Much wow!

The fun part of this idea was really how to figure out a practical way of doing it. The solution was a Rust program that (after some fiddling) uses R1=1k R2=9k+2k trim.

It tries to find a standard value capacitor, that will produce the correct Hz for the given note within 10k +/- 1k.

A 3296 trimmer has 25 turns. Which should give me 2000Ω/25/360° ≈ 0.22Ω/° and I feel confident that I should be able to tune all voices to usable notes.





For a tone of 440Hz, I should set C3 to 150nF and have my trimmer at 1431.9Ω or about 71.6%

https://github.com/AlexanderBrevig/TripleFiveSynth

I'm happy to take suggestions as to how to proceed with KiCAD (or anything really). I thought v6 has eeschema support for Python, but no luck.
I am considering writing my own generator for a kicad_sch file.

EDIT: Next steps, after getting all the 555s in, is to have it be voltage controllable. Futhermore, I think I will passively mix each 555 and then buffer the sum. Making sure to have an acceptable current draw with say 10 voices on. Probably my gate implementation will be a high side FET to entirely switch off silent 555s.

Why you ask? I don't really know.

[ledtester]

I'm happy to take suggestions as to how to proceed with KiCAD (or anything really). I thought v6 has eeschema support for Python, but no luck.
I am considering writing my own generator for a kicad_sch file.

I would just design a smaller board for only, say, 10 of the oscillators and then order multiple copies of that smaller PCB. All that matters is that each oscillator has a footprint for a cap, resistor and trim pot. You don't need to enter the exact values used into the schematic.

Next steps, after getting all the 555s in, is to have it be voltage controllable.

Can you elaborate on what you mean by voltage controllable?

Edit: If this is what you mean by voltage controllable:

Probably my gate implementation will be a high side FET to entirely switch off silent 555s.

I would look at putting a MOSFET across C2 or C3. If the MOSFET is on it would prevent the cap from accumulating charge which should stop the oscillation. Or maybe put it in series with one of those caps. Then you could use easier to use N-channel MOSFETS and ground referenced control signals.

[alexanderbrevig]

I would just design a smaller board for only, say, 10 of the oscillators [...]

Happy accident! This is exactly what I've been doing! Image attached.
1 octave per board.

Need to find some less proud pots though, these won't fit Eurorack.

Can you elaborate on what you mean by voltage controllable?

Following Eurorack CV standard 1v/octave. I won't bother with an analog solution. It will be some mCU doing the work. Probably STM, as I know them and still have a few on hand.

I would look at putting a MOSFET across C2 or C3. If the MOSFET is on it would prevent the cap from accumulating charge which should stop the oscillation. Or maybe put it in series with one of those caps. Then you could use easier to use N-channel MOSFETS and ground referenced control signals.

Attached an image of my current attempt. I went for a high side switch, but I am thinking I'll need a weak pulldown to GND from 555PWR. This way I will also start the oscillators in the same state, as reset will start low?

NB: This is my first 555 circuit :p

[alexanderbrevig]

Just wanted to show the new Schematic Properties thing. They take parameters for their C and R values.

How does one get these to work?

[ledtester]

Need to find some less proud pots though, these won't fit Eurorack.

You can mount those Bournes trim pots lying flat with the leads making a 90-degree bend.  In fact, they have little feet at the corners:

[alexanderbrevig]

Still, it will be a tight fit and probably finnicky to tune.
Maybe I should look at something like the 3224 series.

[fourfathom]

Any reason why you don't just do the top octave with 555s and then use dividers for the lower frequencies?

[ledtester]

Still, it will be a tight fit and probably finnicky to tune.
Maybe I should look at something like the 3224 series.

What about mounting your pots facing out one edge like this:



Then you can still access them even when the boards are stacked on top of each other.

[alexanderbrevig]

I really like the idea but have a hard time finding any good part. Also, I'm not envisioning having to tune after the fact. But...
Tonight I will make an effort. With 3224W they can at least stack inside a 8HP Eurorack panel, with a "mixer" board and a "brains"/motherboard.

[alexanderbrevig]

Any reason why you don't just do the top octave with 555s and then use dividers for the lower frequencies?

A really neat idea  , but I kind of want it be a "true" one per voice. Just for the gimmick of it! 
Also, I want polyphony. It would've required me to have some kind of output gate. I hope to achieve reset/zero crossing. So they all start the same place in the waveform.
Spice seems promising with high side switch.

[Zero999]

Rather than switching the power to the 555, switch pin 4 (reset) instead, or leave the 555 running and switch its output to the mixer.

Just one question: why not use an IC with more than one oscillator block on it? The 556 is an obvious possibility.

The 4093 has four Schmitt input NAND gates.
https://www.electroschematics.com/wp-content/uploads/2011/04/CD4093BM-BC-datasheet.pdf

If 6V operation and SMD are acceptable, there's the 74HC7540, which has eight gates.
https://assets.nexperia.com/documents/data-sheet/74HC_HCT7540.pdf

Also consider using the waveform at the capacitor, rather than the output, which might sound nicer than a squarewave, but it will need to go to an amplifier, with a much higher input impedance, than the timing resistor used in the oscillator.

[alexanderbrevig]

Rather than switching the power to the 555, switch pin 4 (reset) instead, or leave the 555 running and switch its output to the mixer.

As it's going into a Eurorack I was kind of unsure of the power consumption. Switching reset would be much easier, so maybe I should just make sure that quiescent is OK.
Will simulate and check! Thanks

Just one question: why not use an IC with more than one oscillator block on it? The 556 is an obvious possibility.

This whole project came about because I am currently building a modular synth for my dad for his birthday. I listened to TheAmpHour and episode 555 started playing. I realized I had never played with it, so that is why. However stupid it may be, 555s are here to stay

Also consider using the waveform at the capacitor, rather than the output, which might sound nicer than a squarewave, but it will need to go to an amplifier, with a much higher input impedance, than the timing resistor used in the oscillator.

I actually want the square and the slight PWM deviations it should make. A personal timbre to the instrument?
The modular already has two VCO with various waveforms. This is just a fun little addon.
There's also filters, so having the rich/harch harmonic content is kind of just cool

Here's a render of another module:

[fourfathom]

If you take the analog voltage off the timing capacitor, and use the 555 configuration where the output pin charges / discharges the capacitor through a resistor (instead of using the "discharge" pin) then you get a waveform approximating a sawtooth.  This sounds *much* nicer than a squarewave.  A friend of mine designed one of the first "string synth" electronic instruments (PAiA) using a triangle wave generator.

[Siwastaja]

You could even mix the sawtooth and square wave outputs with a potentiometer, to create a freely adjustable tone.

For the timing capacitor, use stable types to avoid having to re-tune the instrument all the time. If you use ceramic caps, C0G/NP0 instead of X7R, and so on. Failing to have stability specifications, you could guesstimate that higher initial tolerance parts are more stable, too. Definitely no aluminium electrolytic caps. Tantalums might work in a pinch.

[CaptDon]

Many years back RMI made a keyboard with a separate oscillator for every note with a trimpot adjustment. The nice thing is that you could easily do even temperament tuning or just tuning or anything in between. I repaired an RMI for a friend. Made in Rocky Mount North Carolina.
 

[ledtester]

Many years back RMI made a keyboard with a separate oscillator for every note with a trimpot adjustment. The nice thing is that you could easily do even temperament tuning or just tuning or anything in between. I repaired an RMI for a friend. Made in Rocky Mount North Carolina.

Did some googling and found this:

https://www.soundonsound.com/reviews/rmi-keyboards-retrozone

RMI's first keyboard was a four‑octave affair called the Explorer, a garish‑looking instrument with a green speckled underside and a blue speckled upper body. The Explorer's sound was generated using a dedicated oscillator for each key, and this architecture was later to become the standard for most of RMI's instruments. However, the Explorer's real claim to fame lay in its 'Flying Hammers'. These were weighted steel springs that, when you played a key, vibrated against an electrical contact, turning the oscillator on and off to produce what have been described as "mandolin and banjo effects". To be honest, I've never seen, let alone played, one of these, so I would welcome the chance.

And that page has this pic... are the round things the tuning pots?

[fourfathom]

"Those round things" look like inductors.

[Ground_Loop]

If the output of each oscillator is presumably 0-Vcc square wave will you be able to passively mix them?  You may find the combined output locked to Vcc with multiple notes pressed resulting in no audible tone or at best an annoying crackle.

[CaptDon]

NOT SQUAREWAVE, that would be pointless due to all the odd harmonic content. They were nearly sinewave, every one!! AND, like a real piano you could press every key at the same time and get every note!!! That is the magic of one oscillator per note. Note shown in the pictures posted are the little trimpots with the blue 'wheel-like' adjuster with the screwdriver slot which was inaccessible because the trimmers were side by side in a row. You had to tweek the trimmers with your fingers. The fact that you could vary the temper of the tuning meant many pleasing tuning creations. Most notably the calliope type tuning as opposed to 'even temper tuning'. There were some switches that produced some overtone distortion in the audio amplifier section to create 'Brass' or 'Oboe' type sounds. As for the guy wanting to use multiple 555's, well yes they would be square wave and would require probably at least a 3 resistor / 3 capacitor RC filter on each 555 to get a somewhat sine with overtone waveshape.

[Ground_Loop]

I'm guessing the good captain is responding to my post. But my comment was to the OPs plan to use 555 oscillators without filters or shaping.

[Dabbot]

Rather than switching the power to the 555, switch pin 4 (reset) instead, or leave the 555 running and switch its output to the mixer.

Just be mindful of the power draw if you leave them running. With 100+ 555s all running, at any time a good portion will be in the discharge cycle, which means their R1 is connected to ground.

[ledtester]

The 4093 has four Schmitt input NAND gates.
https://www.electroschematics.com/wp-content/uploads/2011/04/CD4093BM-BC-datasheet.pdf

Power consumption-wise using the 4093 as a gated oscillator makes a lot of sense.

https://diyaudiocircuits.com/4093-nand-gate-oscillator/



For Vdd = 5V, the frequency formula is \$f \approx \frac{1.23}{R\cdot C}\$ using Vp = 2.9 and Vn = 1.9.

[Siwastaja]

NOT SQUAREWAVE, that would be pointless due to all the odd harmonic content. They were nearly sinewave, every one!!

The harmonics are not pointless, they are the point. This is about musical instrument, not a test tone generator. You don't want a sine wave synth with one frequency per key. It sounds extremely dull, it's not used very much in music. Sine wave based electric organs (or acoustic, pipe-based as in church) mix several sine waves i.e., build more complex waveforms with harmonics. They have multiple oscillators per key.

But you don't have to build the sound by mixing 3-4 sinewaves per every note; the simpler alternative is to use waveforms such as triangle or square. They have the harmonics built-in. The odd harmonic content of square wave resembles oboe, it has this "hollow" sound. Harmonic content of triangle wave is "warmer" and somewhat resembles strings. See chiptunes, Nintendo Entertainment System for example works solely on these two "instruments": square and triangle wave. No sine wave available.

So unlike a single-frequency sine wave, square wave is at least workable, with a little bit of low-pass filtering (not too much to make it near-sine) it becomes less harsh.

The 555 circuit is just fine because it gives you the triangle and square wave options, and after mixing, it's trivial to low-pass filter the whole output to further adjust the tone.

[jonovid]

this could get interesting

tip
use ALL the pins of the 555 has to offer.  this old 1970s chip is noisy with spikes on its pin 3 & a current hog.
tame the beast.
do same timbre hypothesis testing in a digital audio workstation before buying anything
before trying to make the 555 antikythera mechanism of music.

I spend 5 yrs trying to make a heavy metal band in a box with the hypothesis that heavy metal music was just modulated noise with popping sounds added to the mix.
am currently retracing my teenage steps with my own 555 project.

the holy grail of any 555 oscillator based musical instrument  is the wheels within wheels timbre philosophy

[LaserSteve]

Do what the neon lamp relaxation oscillator synths did. Use a tiny amount of the upper octave very loosely capacitive coupled to the next to sync. Makes tuning easier. In the neon case they wrapped little pieces of foil around the lamp to couple in the desired coupling.  So keep the top octave tuning where you can get at it, add LDRs for trim and sync "downwards" . Trust me, it does not take much stray to
injection lock anything.

Had a case at work where two 32Khz crystals were locking to each other a meter and a half away from a fraction of a milliwatt of radiated emission.

I have a multi-pound bag of matched neon lamps, for when your done with the 555s.

I'm sure you can build divide by twos with CMOS 555s.
Power consumption with NE555 would be huge.

This Google  search string will give you hints on the voicing and filters
"electronics 1958 neon lamp organ"

Steve