From: email@example.com (Tom Bruhns)
Subject: Re: Problem with Spice simulation of Linkwitz-Riley filter
Date: 18 Sep 2002 10:17:55 -0700
NNTP-Posting-Date: 18 Sep 2002 17:17:55 GMT
Seems very strange...unless either (1) the input resistance to your op
amp circuit is negative and decreases as you increase the gain, or (2)
the output impedance of your op amp buffer circuit increases as you
increase the gain. I would not expect either of these to be
significant with decent op amps. You might try replacing your op amps
(including the gain setting resistors) with simple voltage-controlled
voltage sources for testing. Also, if you can post the spice netlist,
we might see something strange in it. (Maybe don't post the whole
opamp subcircuits, if they are long.)
Additional things to try: Just simulate your first opamp stage, and
"measure" the input and output impedances and the gain. (Input
impedance = input voltage/input current, but for output you'll need to
run it with and without a load.) In the circuit you have, look at the
frequency response to each node (or from one node to the next)...is
the change you're seeing (as you change the first amplifier gain)
isolated to one area? Can you tell from that what's causing it?
firstname.lastname@example.org (Carlos) wrote in message news:...
> Hi all,
> I'm having a strange problem with a Spice simulation of a Lowpass 2nd
> order Linkwitz-Riley filter (that's two 1st order Butterworth in
> series). I have the evaluation version of Spice, which should work ok
> with two opamps and such a simple circuit.
> If I simulate the following (it's very simple so I'll just describe
> Vin -> R in series -> C to ground -> unity gain non-inverting opamp ->
> R in series -> C to ground -> unity gain non-inverting opamp -> Load,
> I do get the expected result and cutoff frequency (With R = 110 k and
> C = 18 nF, F6 = about 80 Hz).
> The problem is that if I make the gain of the first opamp > 1 with the
> usual couple of resistors in the feedback loop, the cutoff frequency
> goes down a lot (in some cases to 18Hz). I've done the maths and also
> double checked in Lancaster's Active Filter Cookbook, and apparently
> there should be no problem doing this. In fact, funnily enough, if I
> remove the second opamp (so it's just R/C -> opamp with gain>1 ->
> R/C), it works fine.
> Things I have tried:
> - Changing the order of magnitude of the feedback resistors in the
> first opamp.
> - Putting a resistor (several values) between the second R/C section
> and the second opamp.
> - Putting a resistor (several values) in the feedback loop of the
> second opamp.
> - Both at the same time.
> - Making the first opamp unity gain and the second one gain > 1.
> - Using different opamp models (LF411, LM324, uA741).
> In all cases, as long as one of the opamps has gain > 1, the cutoff
> frequency goes down noticeably.
> What's going on?
> Many thanks in advance,
> Carlos (I'm really at clorenzogNO@SPAMnavegalia.com, you know what to