A simple converter that turns sawtooth into triangle

Nov 08, 2026 Last reply: 20 hours ago 4 Replies
Jump to replies (4)

When it is up and running, this little circuit takes a sawtooth at its input and hands you a clean triangle at the other end. It works on its own as a standalone waveshaping device, and it can also be wired up to a simple VCO, where it gives a SHAPES VCO an extra waveform to pick from. On the final scope shot the triangle looks proper, and the maker demoed it with musical tones playing over an audio track, which is exactly the kind of proof I want from a waveshaper.



The parts count is refreshingly small: one TL074 quad op amp, a pair of 1N4148 diodes, two 1uF foil or ceramic capacitors, two 200k resistors, one 20k and two 10k resistors. The maker worked out the design in Falstad first, then took it to the bench with an analog oscilloscope and a digital multimeter. There is even a hand-drawn schematic on dot-grid paper, which I find charming and honestly easier to follow than some CAD exports.



A few choices stand out to me. The input signal goes through an inverting buffer, and two diodes run in parallel into a single output node. Matched 10k resistors are used to correct gain mismatches between the two halves, which seems like a smart bit of care for such a minimal circuit. Pulling the output junction to a -12V rail before the output AC capacitor, rather than to ground, lets that node swing below 0V, and that reads as a neat fix for an otherwise awkward limitation.



It is not all solved, mind. Input DC offset shifts the waveform crossing point and distorts the triangle shape, so a high-pass filter at the input was added to move that crossing point back where it belongs. And ordinary 1% resistor tolerances still cause gain mismatches in the inverting stage, so how close the two slopes match depends on how lucky you get with your parts. That feels like a fair trade-off for a circuit this compact.



Project by Moritz Klein. Watch it on their YouTube channel:

formatting link


I am curious about the crossing point business. Did the maker explain how the offset gets dealt with in the final circuit, or is the high-pass filter the whole story there? I would love to know whether that filter value needs tweaking per VCO or if it just works once set.

The -12V rail trick is the bit I keep coming back to. Pulling the output junction down to the negative rail instead of ground lets the node go below 0V, which an ordinary AC coupling setup cannot manage here since the diodes block reverse discharge current. It costs one connection and solves a real problem, and that seems like exactly the kind of thinking this hobby needs more of.

Since standard 1% tolerances cause gain mismatches in the inverting stage, a concrete improvement would be to measure a batch of 10k resistors and hand-pick a matched pair before building. It costs nothing but a multimeter session, and the two slopes of the triangle should line up better as a result. Pairing that with the matched resistors the maker already uses would tighten the shape up further.

Yes, that part is covered. The input DC offset shifts the waveform crossing point and distorts the triangle shape, and to deal with it the maker added an input high-pass filter that shifts the crossing point back. So the filter is the fix for the offset, and the matched 10k resistors handle the separate gain mismatch issue in the inverting stage.

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required