Double Integrator?, HOWWWWWW????

I'm trying to design a double integrator circuit, but I'm having trouble. Chaining two opamp integrators together doesn't work, even with an opamp buffer between them. How can I make one?

Okay, came up with this



Have two problems with it though:

1) It's unstable, even with a signal that has no DC component.
2) How would you implement the current "drain" on the second OPAMP using actual parts?

Thanks for any help...

This is the code if you want to plug it into the falstad circuit sim:

If the op-amp models have ideal outputs and no voltage limts, two chained integrators will indeed produce the correct result. You may not be able to see this if the result runs away.

Without global feedback, integrators have a nasty habit of running away due to very small offsets. For this reason, a low frequency pole is always present, e.g., a very large resistor across the integration capacitors, or a long time constant from output to input (or a "bias servo" which does the same thing).

Tim

keep in mind that sim, with no initial values, can decide that the initial voltage on C1 is 4BV, and C2 is -700MV.


anyways, if this is in relation to the DAC thread, I'll restate my post -- you're trying to force what you want into something that works well. For a sim, it can work, but its not a good practical design. As such, the more you force this to work with real parts, the more time (and maybe money) you'll waste with expensive parts and such.

Others have thought for many years on good ways filter things using opamps. They've also stepped past the math part to the part where the topology is analyzed for the practical limitations of the parts. nuhertz's website has some of the most straightforward descriptions of why you would choose one 3-opamp topology over another 3-opamp topology over a 1-opamp one.