« Back to DIY

DRAFT - the wiring below is translated net for net from the verified 11×14 stripboard layout, but this breadboard version has not been built on the bench yet. Build it once, fix anything that fought back, then delete this banner.

Dual-Wave LFO

Triangle and square from one chip, on a breadboard · By Onion Madder · Mess O' Pedals

An LFO is the effect behind the effect: too slow to hear, it exists to wobble other circuits. This one makes a triangle and a square from the two halves of a single NE5532, chasing each other from 2.5 to about 8 Hz, with a rate knob and a blend knob that morphs between the two shapes. It is the first build in this series that makes no sound at all on its own - and the first with an op-amp.

This is the seventh build in the breadboard series. I planned it as an 11×14 stripboard layout first - a board that needed seven track cuts, every one of them just to keep the chip's two pin rows apart. On a breadboard the center channel does that job by itself, so all seven cuts simply disappear and the layout collapses to eight nets and eleven jumpers. Power is 9V - though 12 to 15V is genuinely better here, see the supply note before power-up.

⚠ Before you start tap to expand / collapse

Two parts care which way round they go: the chip (pin 1, the corner with the dot, goes in F10 - anywhere else and nothing works) and the electrolytic (the stripe marks the minus leg).

Two resistors must not swap. The 68K and the 100K set the trigger window between them; exchanged, the window is wider than the chip can swing and the LFO never starts. Meter them rather than trusting the bands in bad light.

Build with the supply off. The two meter checks near the end are there so the first power-up is boring, and hole D17 stays empty on purpose - it is your probe point for the 4.5V bias that everything else is judged against.

What you'll need

The parts list lives on the build sheet too - this version adds how to spot each part: the printed marking, the package, or the polarity clue.

  • NE5532 dual op-amp, DIP-8 - spot it: 5532 in the part number on top, a notch at one end and a dot beside pin 1. An LM358 is the same pinout and is actually specified for a 9V supply - see the supply note
  • 47K resistor - sets the fast end of the rate sweep. Spot it: bands yellow-violet-orange
  • 68K resistor - trigger window, triangle side. Do not swap with the 100K. Spot it: bands blue-gray-orange
  • 10K resistors - the mid-rail divider; a matched pair matters more than the exact value. Spot them: bands brown-black-orange
  • 100K resistor - trigger window, feedback side. Spot it: bands brown-black-yellow
  • 1 µF film cap - the integrator cap; it sets the rate, so film, not ceramic. Spot it: a little box or oblong, either way round
  • 47 µF electrolytic, 16V or higher - bias decoupling. Spot it: a little can; the stripe marks the minus leg
  • 100nF ceramic - supply decoupling, not on the stripboard layout; added here because the NE5532 is a fast part and long breadboard rails are an invitation. Spot it: printed 104
  • 100K linear pot - rate. Spot it: marked B100K. Only the wiper and lug 3 are used
  • 1K linear pot - blend. Spot it: marked B1K or B102. All three lugs; its wiper is the LFO output and never touches the board
  • A breadboard - all of it this time: both banks and both rail pairs. The build spans columns 2 to 22, so a half-size board fits. Plus about 11 jumpers, at least one of them insulated wire rather than a bare leg
  • output lead - tip and sleeve, to whatever you are wobbling
  • 9–15V DC supply - a 9V battery works, but 12 to 15V is the better choice here and costs nothing in rate (see the supply note). Four salvaged vape cells in series land at 14.8V nominal, inside that window - untested on this build, so meter it first

Reading the breadboard

This guide names exact holes, like F10 or C22. Letters run across the board and numbers run along it, and both are printed on the board itself - the printed labels are the authority, whichever way up you hold the board. The five holes in a bank that share a number are one electrical point.

Unlike the rest of the series, this build uses both banks - the chip straddles the center channel, A to E on one side, F to J on the other - and both rail pairs. To keep the wiring unambiguous, the rails are named by their neighbors: the A-side rails are the pair beside row A, and the J-side rails are the pair beside row J. Red stripe is +, blue stripe is −, on both pairs. Before you wire anything, meter each rail end to end - plenty of boards break the rails at the midpoint, and if yours does, bridge the break with a jumper.

Quick reference - the whole build on one card

The chip map with the hole every pin lands in, then the four corners of the circuit at a glance. The steps below walk through it slowly; this is the bench card.

Notch toward column 10, pin 1 - the dot corner - in F10. Op-amp A (the F row) is the integrator that draws the triangle; op-amp B (the E row) is the trigger that snaps the square. Hole D17 stays empty - it is the bias probe point.

Power

  • Bridge red → red, blue → blue between the rail pairs
  • A-side red rail → A10; G13 → J-side blue rail
  • 100nF: B10 → A-side blue rail

Bias - 4.5V

  • 10K: A17 → + rail; 10K: B17 → − rail
  • Jumpers: E17 → F17; C17 → C12; G17 → G12
  • 47µF: + H17, stripe → − rail
  • D17 stays empty - probe it for 4.5V

Integrator + rate

  • Jumpers: G10 → G7; H11 → H5
  • 1µF: I5 → I7; 47K: J5 → J2
  • Rate pot: wiper → H2, lug 3 → C22

Trigger + blend

  • Jumpers: D13 → D7; B11 → B22
  • 68K: F7 → E7; 100K: C7 → C11
  • Blend pot: lug 1 → J7, lug 3 → D22, wiper = output

Step 1 The chip

The whole circuit hangs off eight pins, so this is the step to get exactly right. Every other part is measured from these four columns.

The chip
StepDo
1Seat the NE5532 across the center channel at columns 10 to 13, notch toward column 10, pin 1 - the dot corner - in F10. The F row then reads F10 = 1, F11 = 2, F12 = 3, F13 = 4, and the E row reads E10 = 8, E11 = 7, E12 = 6, E13 = 5

Steps 2–12 The jumpers

Eleven of them: two rail bridges, power to the chip, the bias distribution, and the four signal buses that carry the op-amp pins out to where the parts live.

Jumpers
StepDo
2Jumper: A-side red rail to J-side red rail
3Jumper: A-side blue rail to J-side blue rail
4Jumper: A-side red rail to A10 - power to pin 8
5Jumper: G13 to the J-side blue rail - pin 4 to ground
6Jumper: E17 to F17 - hops the channel so the bias reaches both halves of the chip
7Jumper: C17 to C12 - bias to pin 6
8Jumper: G17 to G12 - bias to pin 3
9Jumper: G10 to G7 - pin 1, the triangle, out to its bus
10Jumper: H11 to H5 - pin 2 out to the integrator bus
11Jumper: D13 to D7 - pin 5 out to the trigger bus. This one passes over the chip, so use insulated wire, not a trimmed component leg
12Jumper: B11 to B22 - pin 7, the square, out to its bus

Steps 13–17 The resistors

The matched pair first, then the three that set the speed and the trigger window.

Resistors
StepDo
13Resistor 10K (brown-black-orange): A17 to the A-side red rail
14Resistor 10K (brown-black-orange): B17 to the A-side blue rail
15Resistor 47K (yellow-violet-orange): J5 to J2
16Resistor 68K (blue-gray-orange): F7 to E7. It stands upright across the center channel at column 7 - one bank to the other, same column
17Resistor 100K (brown-black-yellow): C7 to C11

The 68K and the 100K set the trigger window between them: the 68K carries the triangle into pin 5, the 100K carries the square back into the same node. Swap them and the window comes out wider than the chip can swing, and the LFO never starts. If the bands are hard to read, meter them.

Steps 18–19 The small caps

Film and ceramic capacitors
StepDo
18Cap 100nF ceramic (printed 104): B10 to the A-side blue rail. Keep it short and right at the chip - it is the supply decoupling
19Cap 1 µF film: I5 to I7. Either way round

Step 20 The electrolytic

Bias capacitor
StepDo
20Cap 47 µF: + leg into H17, striped − leg to the J-side blue rail

Steps 21–23 Off the board: pots and output

Both pots fly off the board on leads. The blend pot's wiper is the LFO output itself - it never touches a breadboard hole.

Pots and output
StepDo
21Rate pot 100K: wiper into H2, lug 3 into C22. Lug 1 is genuinely unused - tying it to the wiper at the pot is harmless good practice
22Blend pot 1K: lug 1 into J7 (the triangle side), lug 3 into D22 (the square side)
23Output lead: tip to the blend pot's wiper, sleeve to a blue rail

Both pots here are wired unlike anything else in the series: the rate pot is a rheostat that uses the wiper and lug 3 only, and the blend pot is a divider whose two ends sit on two different signals - triangle on one lug, square on the other - so its wiper sweeps between the waveforms instead of between a signal and silence.

Steps 24–27 Checks, then power

Two meter checks with the supply off, then power, then the reading that proves the heart of it is alive.

Checks and power
StepDo
24Continuity, supply off: column 17 in the A-to-E bank to column 17 in the F-to-J bank (through the E17-F17 jumper), and both to C12 and G12. Then confirm C12 does NOT reach G12 by any route other than through column 17. Finally A10 to the red rail and G13 to the blue rail
25Resistance from the red rail to the blue rail: roughly 20K - that is the two 10Ks in series, a useful sanity figure. Anything near zero is a short: find it before you power up
26Supply off: positive to the red rail, negative to the blue rail. 9V minimum, 12–15V better - see the supply note below
27Power on with the rate pot at mid-travel and put the meter on D17, DC volts: it should read half the supply, steady - 4.5V on a 9V rail. Then probe F10 and E11: both should visibly wobble rather than settle

If D17 does not read half the rail, stop there - the bias is what every other voltage in the circuit is judged against, and nothing downstream can work until it is right. At these speeds a cheap DMM is scope enough: 2 to 8 Hz shows up as a reading that will not sit still.

If it doesn't work

Symptom → check
SymptomCheck
The bias at D17 isn't half the rail.The 10K divider, or a missing rail bridge. Fix this first - nothing downstream can work. Check both rail bridge jumpers.
Bias good, but pin 1 (F10) and pin 7 (E11) both sit dead still at bias.The loop never starts - usually the 68K and 100K swapped. The 68K goes from the triangle to pin 5, the 100K from pin 7 back to pin 5.
Pin 7 latched hard at one rail and staying there.The 68K is open or its cross-channel legs aren't seated. Without it the triangle can never pull pin 5 back - check both ends of the upright resistor at column 7.
Pin 1 drifts slowly to a rail and parks.The 1µF isn't making contact, so the integrator runs open-loop. Reseat it at I5 / I7 and confirm continuity from column 5 to pin 2.
Runs, but the rate pot does nothing.The rate pot's wiper and lug 3 are swapped, or the wiper lead is in the wrong column. Wiper goes to H2, lug 3 to C22.
Only one waveform across the whole blend sweep.A blend pot lug in the wrong place. Lug 1 to J7 (triangle), lug 3 to D22 (square).
Waveform has fuzz or steps riding on it.No supply decoupling - that is what the 100nF is for. The NE5532 is a fast part and long breadboard rails are an invitation.
Runs, but the swing looks squashed and clipped.The 9V problem - see the supply note. Raise the supply to 12–15V.

How it works, in one breath

Half the chip is an integrator: feed it a steady voltage and its output ramps at a steady rate - that ramp is the triangle. The other half is a trigger with a window set by the 68K and 100K: when the ramp crosses the window's edge, its output slams to the other rail - that slam is the square. And the square is what feeds the integrator, through the rate pot, so every slam reverses the ramp. The two halves chase each other forever, the 10K pair holds the halfway voltage the whole dance swings around, and the blend pot's wiper rides between the two outputs. Because the speed is set only by resistors and the film cap, raising the supply voltage changes the amplitude and nothing else.

The supply note, or the 9V problem

TI's NE5532 datasheet recommends a supply of ±5V to ±15V, and a 9V single supply is only ±4.5V - below the minimum, with about a volt of headroom left at the trigger's input. The stripboard original is verified and does run at 9V, but it is outside the part's spec. Since the rate formula has no supply term in it, running this at 12 to 15V costs you nothing and fixes it. If you are committed to 9V, use an LM358 - same pinout, and designed for exactly this.

What the rate pot gives you

The 47K plus the dialed-in part of the 100K pot set the ramp current. The film cap is the coarse control: 470nF roughly doubles every figure here, 2.2 µF roughly halves them.

Rate pot position → frequency
Rate potTotal resistanceFrequency
fully down47K7.8 Hz
quarter72K5.1 Hz
half97K3.8 Hz
fully up147K2.5 Hz

What the meter should say

Every reading is judged against the bias at D17. On a 9V supply, bias is 4.5V; on 12V it is 6V - always half the rail.

Meter readings on a working build
PointExpected
D17Half the rail, steady - the bias. Everything else is judged against this.
Pin 3 (F12), pin 6 (E12)Bias, steady.
Pin 2 (F11)Bias, steady. It is a virtual earth - if it wanders, the integrator isn't in control.
Pin 1 (F10)Wobbling around bias - the triangle, roughly ±2V.
Pin 7 (E11)Wobbling around bias - the square, swinging as close to the rails as the chip manages.

Things to try

  • Give it an eye. Hang a 4.7K resistor from column 22 (the square bus) to a spare column, and an LED from there to the blue rail - it pulses in time. The square is about 1.5× the triangle's size, so don't expect the blend to feel like a level-matched crossfade; that asymmetry is the design.
  • Change the span. Swap the 1µF film for 470nF and every rate roughly doubles; 2.2µF and everything halves. The rate range is the one thing the supply voltage cannot touch.
  • Wobble something. The output sits at half the rail - perfect as a control voltage into anything with a CV input. Into a pedal or an amp input, put a series cap after the wiper first, or the DC will thump.
  • Feed the family. Point it at the Hive's inputs or an Optical Theremin's photocell (LED taped to LDR = homemade vactrol) and the LFO starts earning its keep.

The circuit came out of the Experimentalists Anonymous DIY archive, where synth schematics are kept alive. This breadboard recipe, the stripboard layout it was translated from, and the words are my own. Build it, bend it, make it yours.