# Design a triangular wave generator using a sinusoidal input

I've been going through online assignments regarding circuit designs. I've gotten a bit stuck on this one:

Using a sinusoidal input voltage of amplitude 5 V and frequency 100 Hz, I need a triangular wave output with peak-to-peak voltage of 20 V. The external biasing voltages are assumed to be +15 V and -15 V, saturation voltage magnitude of 13.5 V and maximum output current of 25 mA.

I thought of using a Schmitt trigger to first convert the sinusoidal wave to a square wave, and then use an integrator to refine it into triangular wave using suitable capacitors and resistors.

I've not been able to do so.

This is the design I've come up with:

This isn't working. There might be minor errors, but I'm not able to find them.

Can someone help me with the design, and the possible values of the resistors and capacitors I should use in the integrator and Schmitt trigger circuits?

• hint: your operating system has a functionality for screenshots. You don't need to use your phone for photos of the screen :) Jun 11, 2021 at 16:04
• Yeah, sure. Thanks. Jun 11, 2021 at 16:08
• with that out the way, what's the waveform you're getting out of this? Jun 11, 2021 at 16:10
• "This isn't working" won't win any prizes for Troubleshooter of the Month. What exactly is it doing. Jun 11, 2021 at 16:17
• Hysteresis feedback is probably 13.5/2=6.5V while sine wave is only 5V. Will it trigger?
– AJN
Jun 11, 2021 at 16:21

You’re close..

1. AC couple into a zero crossing comparator that limits at given saturation of +/-13.5V =27Vpp

2. no need for a Schmitt as it gets integrated anyways in next stage and Hysteresis will cause amplitude controlled phase jitter if noise was added to input signal. If it has DC offset, AC couple into some R to 0V.

3. choose the cap current equation wrt. Vpp swing which controlled by bipolar cap current step

4. Ic=CdV/dt the cap sees the output voltage 20Vpp or +/-10Vp with respect to the virtual ground which also happens to be Vin+=0V

5. dt= 5ms = 1/2f, for f=100,. C=Ic * dt/dV = 1mA * 5ms / 20V

6. Since square output current charges the cap relative to 0V , the resistor only sees the square wave voltage = +/-13.5Vp. Using the absolute constant current ramp from the bipolar square wave |Ir|= Vp/R= 13.5V/R

7. Ir=-Ic into Vin- ( now solve for RC)

8. Ic= C * 20V/5 ms = Ir = 13.5/R

9. RC = 13.5/20 * 5ms = 3.375ms

10. So consider 33k * 100nF = 3.3ms and tweak the difference

11. Since the integrator has >10^5 DC gain, add a Rfb about 30x Rin (then it’s hard to see nonlinearity, but actually easy to measure with cap current sag.

12. Simulate or build it. Verify all assumptions in your “spec” .That is usually much more detailed but this step is called DVT meaning ?(guess)

13. Too late to ask for new specs like constant triangle wave for variable frequency. remember ,always firm up all specs before starting design

All done. cheers.

## Post. Mortem

On 2nd thought the straight comparator with glitches from noise will round off the peaks and change the Vpp amplitude, so the comparator with %hysteresis greater than % (Noise/signal) in Vpp ratio is best even though it causes a phase shift.

• I see, I realised a little too late and deleted my comment just before you replied.
– user173271
Jun 11, 2021 at 18:23
• Thanks a lot! Actually, I wasn't acquainted with the zero detector circuit until recently, so I was trying to use Schmitt in the first place. Anyways, thanks again! Jun 12, 2021 at 5:08