Hi,

I am designing an analog circuit that resembles a Gaussian shaping amplifier (takes voltage spike looking signals at the input and integrates it to a Gaussian shape voltage signal preserving the amplitude with a particular ratio). I chose the cascaded 2 stage Sallen-Key filter topology for that. Unfortunately, I get quite nasty ringing of give or take 13 MHz in my signal that I can not eliminate on my own (my input signal is around 1 us, expected output signal is around 8-10 us, I use LT1365 opamps).

In general, the circuit works and does what is required from it. But I can not get rid of this 13 MHz oscillation. (It is acquired after first stage, the second stage only amplifies the oscillation a little bit more.) What's interesting that this tone is not applied to all the signal. I usually get just a few high amplitude oscillations on the rising edge, but after that 13 Mhz tone is not present in the signal any longer. This might be a pole of the opamp, but I am not sure, since the original signal frequency is not that high... I tried to put a series 5.1K resistor on the output of the second stage and measure the signal through it. Signal gets a little smother, but that 13 Mhz rising edge stuff is still strong. I also tried to put an additional opamp buffer at the output to load the second stage properly, but that did not make any difference. There's a circuit diagram in the attachments. The 47K series resistor in the second stage feedback path (feedback to the positive terminal of the opamp.) is added to eliminate another unwanted effect: before I have added that resistor, my opamp was introducing significant undershoot right before the rising edge of the signal. So I added the series 47k resistor in the feedback path experimentally, and solve that problem. Other than that, I guess it's just better for you to take a look at the circuit and ask for an additional clarification if needed.

I have already made the PCB and yet made quite a few modifications to it (picture in the attachments). I did not preserve the factory look of the PCB, had to cut a few traces and solder some additional wires. In the end, it does not seem like I have introduced significant amount of parasitics to it, however I can guarantee, that I have strong decoupling for my opamps. I really need to make this board working one way or another, so changing the topology or redesigning a PCB is not useful suggestion in this case... I really appreciate any observations and help how should I solve this problem. Thanks.

 

I eliminated the overshoot and ringing by reducing the feedback capacitor from 220pF to 120pF (below):
The circuit response was some underdamped, so this reduced the Q of the circuit and eliminated the peak in the AC Bode plot.

(I see no reason for the 10kΩ load resistors so I left them out).

 

Mostly, ringing and steady oscillation are caused by excessive positive feedback. That might possibly be due to capacitor values in the filters, but more likely it is due to unintended coupling thru either the common circuit or the power to the op-amps. Leadarrangement and power supply bypassing are the most common causes and cures in a correctly designed circuit.

 

I eliminated the overshoot and ringing by reducing the feedback capacitor from 220pF to 120pF (below):
The circuit response was some underdamped, so this reduced the Q of the circuit and eliminated the peak in the AC Bode plot.

(I see no reason for the 10kΩ load resistors so I left them out).

View attachment 331786

I tried to replace those feedback capacitors with 120 pF and even with 47 pF. It did not change anything. I appreciate the simulation, but it seems it would not help much in this situation...

 

Are these problems in a simulation? Or in an actual physical circuit? There is a lot of difference!!

 

Are these problems in a simulation? Or in an actual physical circuit? There is a lot of difference!!

Problems are in the real physical circuit

 

OK, in an actual circuit system, unintended coupling can be a very irksome problem. So my comments about bypass capacitors and common connections are relevant. But that does not mean they are simple solutions. Or even the problem.

 

And given the blobs of solder I see on the PCB, a cold solder joint could also be the problem.

 

In fact, I do not see even one power supply bypass capacitor in the assembly. Cheap 0.1 MFD disc capacitors are fine, also the tiny ones used in digital applications, provided that they are rated for at least the supply voltage in use.