Simple but Fully Discrete Triangle Wave Generator

[PinePolarBear]

Hello. I'm working on a fully discrete class D amplifier. I'm looking for a triangle wave generator design. It would be preferably simple, but it needs to go above 250 kHz. Thanks.

[JPortici]

Component selection and how to get around all the gotchas to achieve an triangle that as pure and symmetric as possible is left as an excercise to the reader

[Benta]

You should add that the triangular output is from the first amp.

[T3sl4co1l]

I've used some variant of this, many times:

https://www.seventransistorlabs.com/Images/Flashlight2_Schematic.png

IC2 is a astable multivibrator and PWM comparator.  Q4A isn't necessary, but it improves linearity.  If not needed, it can be replaced with a pull-up resistor.

Note that you don't want to skip the rest of the circuit: IC3B compares the inductor current to the current setpoint, providing a first line of defense against destruction of transistors or other components.  The output current simply cannot be anything other than what it can be set to, within a controlled range.  IC3A then controls that current, in order to regulate the output value (in this case, a combination of output current and voltage for driving LEDs, but just voltage would be fine for a power supply or amplifier).

The chained error amplifier topology also avoids putting the double-pole output filter inside a single loop.  Instead, the inductor is in the first loop, and the capacitor and load in the second.  This is much easier to compensate, and gives much nicer waveforms.

Tim

[David Hess]

Oscilloscope sweep generators meet your requirements so you might want to study how they work.

I would also consider bipolar implementation of a PUT based relaxation oscillator with the resistance replaced with a constant current source.

https://www.instructables.com/id/Circuit-Collection-of-the-Programmable-Unijunction/