[Fixed] Op amp integrator clipping

[nullpainter]

I'm building a triangle to quasi-sine wave circuit which is operating from 30Hz to 240Hz but cannot stop the output from clipping. For op amp integrator circuits, how do you control the output amplitude?

I'm using a single source 9V battery power supply and a LM386. I have simulated the circuit with the offset and peak voltage of the triangle wave as measured by my oscilloscope (output is 2.05v - 6.88v) and am seeing the same symptoms.

The output amplitude is clearly too large, but I'm not sure how to reduce it. I tried reducing the amplitude of the input triangle wave, but this resulted in saturated output. I also tried tweaking the input voltages and all other passive components but nothing is leaping out at me.

I read through Olin Lathrop's reply on elecronics.stackexchange to a regular op amp amplifier circuit to get the initial values. I increased C1 from his 10nF suggestion because the output was still fairly triangular, but I can't seem to find a happy medium between something which resembles a sine wave and an output which doesn't clip. Even C1 of ~50nF clips.

I am guessing my problems are related to C1, the biasing of my triangle wave, and the power supply circuit, but not sure how to resolve.

I am hoping for the output to be around the ~4.8V peak to peak mark to be consistent with the triangle wave, however this isn't super critical. Right now, I'm just trying to solve the clipping!

---

Screen Shot 20.png
Description:	Circuit from simulator, demonstrating clipped output and input signal
Filesize:	79.67 KB
Viewed:	213 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

[JovianPyx]

You are using a virtual ground (network on the left) to provide a reference point yet your input is referenced to the negative side of the PSU (shown as a ground symbol). Your input signal should be referenced to the junction of the 100K resistors on the left (the virtual ground). It would help to see this if you draw it so that the virtual ground has the ground symbol instead of the negative side of the PSU (9V battery I'm guessing). This also should include relabeling of +V and -V. In reality, with this virtual ground, you have a dual voltage supply of +4.5v and -4.5v. I believe doing this will help you understand why the circuit is doing what it is.

[nullpainter]

Thanks for that. That doesn't quite make conceptual sense to me yet, so I think I need to revisit op amps and virtual grounds.

Also, I think that's just an artefact of the simulated circuit confusing things - in reality, the triangle wave is the output of another op amp inverter. I tried putting the whole circuit into that little simulator, but I think I overwhelmed it!

Will try again in Spice and report back. Thanks again for your feedback.

[nullpainter]

I've simulated the full circuit with LTSpice and attached an image of the circuit, and the the square, triangle and clipped sinusoidal output.

The triangle wave integrator - basically, everything to the left of R9 - is from someone else's circuit. As you can see, it works beautifully.

I just naively doubled the integrator component (R9, C3 and U3) in an attempt to transform the triangle wave into a sinusoidal wave, and that's where I'm striking problems.

A bog standard RC circuit instead of the second integrator naturally works fine.

Any pointers as to what I'm doing wrong would be greatly appreciated!

---

opamp circuit.png
Description:
Filesize:	20.25 KB
Viewed:	250 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

---

opamp output.png
Description:
Filesize:	27.02 KB
Viewed:	269 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

[nullpainter]

Actually, I just bumped up R9 and everything looks fine, in Spice at least. It's not looking fine with my circuit proper, so will double-check I haven't done something daft!

[Grumble]

Where does the arrow on the -input of U1 go to?

[nullpainter]

[nullpainter]

With R9 at around 450k, I get a nice theoretical output in Spice - but it's almost rail to rail 0-9V and I'm using LM386's so I doubt it'll work out in practice. Also, the waveform clips badly when I adjust R2 and R3 to reduce the pitch (on the real board, this is a single potentiometer for the pitch but I couldn't find a pot in LTSpice)

How would I go about reducing the amplitude and biasing it so it's sitting in a similar range to the triangle wave?

---

sine.png
Description:
Filesize:	25.13 KB
Viewed:	234 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

[Grumble]

[nullpainter]

[Grumble]

np

[JovianPyx]

If I may...

I've seen perhaps better ways of labeling rail connections. Very standard is to label the positive rail with V+, Vdd or Vcc and the negative rail with V-, Vss or Vee. I personally prefer V+ and V-. Sometimes V+ and V- are given their voltage especially when there are more than 3 rails (such as +15, -15, +3.3 and ground).

When you add a virtual ground to a unipolar PSU (such as a 9v battery), you are creating a virtual bipolar supply and I think it's important to label the rails in a way that helps the reader understand what the circuit is doing. It is not important that one end of the battery is 9v compared to the other end - at least not to the circuit and in fact, if the virtual bipolar supply was replaced with a real bipolar supply, the circuit's function would remain the same. This is the logic I use to decide to relabel the rails in a virtual bipolar supply. In this case, it would be +4.5, -4.5, and Vgnd. I also feel free to label the virtual ground as Vgnd to alert the builder that it is not a real ground since virtual ground can have problems if the circuit using it grows and attempts to sink or source too much current from it. It's configuration is easy enough to recognize, but IMO, it doesn't hurt to label it.

Ground symbols are quite standard and representing how the rails function makes the circuit easier to understand. This is why I suggest labeling the virtual ground with a at least a standard ground symbol - because ground is the function and "virtual ground" is just an implementation. Doing this makes it obvious to the builder where the input and output reference ought to be.

I too have seen arrow up and arrow down symbols used for rails which also has some popularity, but using different styles of arrows instead of actual labels can be confusing to the reader, especially when the symbols are small. In my view, using an up arrow as a ground for drawing convenience is confusing. I've seen real ground symbols inverted like that which is not confusing because the symbol itself is used for nothing but ground.

As for unexpected amplitude levels, the integrator used to make sine from triangle wave is in reality a low pass filter. As such the amplitude and phase will vary depending on the input frequency. To fix over-amplitude clipping, I'd use a simple voltage divider at the filter's input as a starting technique, but there may be other ways to do it.

Good that it's working better now, just keep having fun

EDIT ADD:
Have a look at this: https://en.wikipedia.org/wiki/Op_amp_integrator going down the page a bit there is schematic labeled "practical circuit" which shows how to control an integrator's gain.

[nullpainter]

[nullpainter]

Finally fixed! My fundamental problem was that I was inadvertently constructing an ideal op amp integrator, which works beautifully in Spice but fails miserably in real life.

Adding a relatively large resistor in parallel with the feedback capacitor did the trick. Thanks to everyone for their help.

---

opamp circuit.png
Description:
Filesize:	23.16 KB
Viewed:	216 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---