Okay, this is my first post here but I need some help. In class we are having to make integrators from an LM324. Mine is made exactly like the one found here: http://www.allaboutcircuits.com/vol_3/chpt_8/11.html

My capacitor is a .001 uF and my resistor is a 1M. My input is a 350 Hz square wave that is 14 Vpp. I'm supposed to get a triangle out of this obviously. But for some reason the top of my triangle (when the square wave rises) cuts off and goes flat. The bottom of my triangle (when the square wave falls) looks fine though. Would this be because of the size of my capacitor and resistor? It has nothing to do with my rails on the Op Amp because I raised them both all the way up to |20 V|.

My original thought was to try to get RC = 1. But that didn't work at all. My output looked like a DC offset. The triangle wave I have right now is the cleanest one I've had so far. But it's still not a triangle.

 

1M is a little high for the LM324. Try going to 100k with a 0.01uf cap or 10k with a 0.1uF cap.

Also that circuit is an idealized integrator with no dc feedback, thus the op amp will drift to the rails due to normal offsets. Add a 1M resistor across the capacitor.

Why would you want RC = 1 (second)? That's much too long a time constant for integrating 350Hz.

 

What is the output voltage of the opamp when it goes flat? What is it at the bottom of the output? Are the slopes correct? What kind of capacitor are you using? In that range, I would imagine it is nonpolarized, but be sure.

It wouldn't take much DC bias to send your output to a rail and from there your circuit would tend to come down in voltage to a point and then go back up to the rail. You might want to put a bleed resistor across the cap, but you normally want it to be a couple orders of magnitude larger than the input resistor. Can you change your input resistor to, say, 10kohm and your cap to 0.1uF and then but the 1Mohm resistor across the cap.

Can you post a sketch of the waveform you are seeing (with time and voltage indications)?

 

It looks like crutschow and I posted at the same time and he beat me to the punch on the size of the 1M being a bit too large. I went and looked at the data sheet to see what the input bias current is and it can be as high as 0.1uA, which is a nonnegligible fraction of the poster's signal current of +/-7uA.

Also, the LM324 has an absolute maximum voltage supply rating of only 32V, well below the 40V the OP is applying. That may or may not be related to your problem, but you have overstressed the part and I wouldn't mess with it. Throw it away, get another one, and run it from no more than +/-15V rails.

 

Another point that may be worth making is that whenever the input bias currents of an op-amp may be of significant size compared to the input, it can help to equalise the resistance seen by both inputs of the amplifier.

In this way, due to cancellation effects between the input current voltage drops at the two inputs, the error is reduced to being approximately that due to the difference between the two input bias currents, which (for many bipolar devices if in their active region) will normally be a lot less than that caused by one bias current alone.