I'm working through a tutorial on filters at https://www.electronics-tutorials.ws/filter/filter_5.html (yeah, why not ask them? Because I get such good answers here, and I cannot add attachments to questions there!). They have an active low-pass filter example exercise - an RC filter amplified by a non-inverting opAmp - which I thought I cracked. I used LTSpice to simulate the circuit, the Bode plot came out like I thought (and they said) it should. Then they "simplified" the circuit, moving the capacitor into the opAmp feedback loop. The idea is the Gain will become frequency dependent and provide the filtering attenuation. My Bode plot kind of works fine around the cut-off frequency, but at modest higher frequencies things go wrong (10KHz, so I'm assuming the choice of opAmp is not critical). I'm trying to understand what is going on.

I tried the circuit with about3 different opAmp models that I thought should work, all gave similar results. So the crux of my question is "should the second "simplified" circuit be behaving like the first one, and if not, what have I not understood, or what is causing the difference in these Bode plots?"

Thanks for any insights.
Peter

 

hi cspw,
This is what I see on LTSpice.
E

 

Once you reach the corner frequency the gain reduces until xC approaches zero. You’re left with the gain of 1 from non inverting opamp. Try the inverting version.

 

For the first filter the signal keeps reducing with frequency due to the input RC low-pass filter (as expected).

For the second, the signal is reduced until the impedance of C2 becomes very small as compared to R5.
So what is the gain of that amp when the impedance across R5 becomes zero (an AC short)?

 

This is what I see on LTSpice.

This is what I see.
I don't know where that peak you show is coming from.
My simulation below does not have that.

Note I changed the AC input to 1V which gives the circuit output gain in dB.

 

hi Carl,
I did not make any changes to the TS asc file.

I will run it on another PC.

E
Update:
Same result. this is on a 64 bit Win10 LTS XVII.
The earlier sim was a 32 Bit Win 10 LTS XVII/
E

 

hi Carl,
I read the other day that you did not run LTS updates on your system.
I do, I wonder if there is a bug in one of the updated models.??
E
BTW: tried AC= 1 and 3 , no effect


Update:
Swapped the AD549 to LTC. OK.
The AD549 test cct looks OK.

 

I read the other day that you did not run LTS updates on your system.
I do, I wonder if there is a bug in one of the updated models.??

Well I still run version IV since I don't like some of features of the newer version (such as not being able to alternate easily between the AC and Transient analysis without having to re-enter the parameters), so that could be the difference.

You might try to do a bode plot of the open loop op amp model and see if there's anything odd.

 

hi @crutschow
I would say that the AD549 I am using is at fault.!
E

 

This is what I see.
I don't know where that peak you show is coming from.
My simulation below does not have that.

Note I changed the AC input to 1V which gives the circuit output gain in dB.

View attachment 221859

hi cspw,
This is what I see on LTSpice.
E

Thanks. The peak might be because we're driving the op amp into its rails. My original Vcc and Vdd were only +-5V, and my Vin was 3V with a gain of 10! My bad!

The advice to change the AC to 1V so that we see the gain in dB was an "aha" moment for me, .

It is interesting that a few of us seem to be seeing quite different things. I took the bait and upgraded my LTSpice when it was offered. Do we know of any persistent "global" settings, like step sizes, accuracy of convergence, etc. that could make some unstable simulations appear quite different on different systems?

 

Thanks. The peak might be because we're driving the op amp into its rails. My original Vcc and Vdd were only +-5V, and my Vin was 3V with a gain of 10! My bad!

The advice to change the AC to 1V so that we see the gain in dB was an "aha" moment for me, .

It is interesting that a few of us seem to be seeing quite different things. I took the bait and upgraded my LTSpice when it was offered. Do we know of any persistent "global" settings, like step sizes, accuracy of convergence, etc. that could make some unstable simulations appear quite different on different systems?

I got another reply in another forum which makes a lot of sense, as did the one here. At high frequencies, in my second circuit, the opAmp gain is 1, it doesn't keep falling like the first circuit. Along with that, the phase shift "recedes. So in my original plot the phase shift turns around at about 500hz (value -55 degrees) and ends up at 10kHz with a phase shift of -9 degrees.