Problem with a simple low pass filter

[jg001]

I have a problem with a simple LPF circuit. The circuit, input voltage and output voltages are shown in the attached PDF file. My designed LPF has a cutoff frequency of 10kHz. I input a 5kHz signal in it and measured it with a pair of wires (between ground and the output) which are connected to an oscilloscope. My problem is that when I measure the output with these two wires, the LPF output seems to have a very high frequency signal (15.8MHz) that is being modulated by the 5kHz output. But, when I connect a 1k-ohm resistor in series with the output and then measure it using the two wires, the output is a nice 5kHz signal. Once again, if you look at the attached plots, you'll see what I'm talking about.

I don’t know why this is happening! Any help? Thanks.

 

[LvW]

As your simulation plot shows everythings seems to be OK - in principle.
If you observe some high frequency components during measurement - it surely is a measurement problem.
The wires constitute a kind of capacitive loading and this causes the amplifier to oscillate.
The observed frequency is a clear indication (15 MHz) - it is app. the unity gain frequency of the opamp.

To solve the problem you could try to reduce the tendency of the circuit to oscillate by increasing the gain of the opamps. There are other designs of the Sallen-Key architecture (v=2 or even larger) which are somewhat more stable. Of course, the parts values (R,C) have to be recalculated.
Or you can switch to other topologies (multi-feedback for example).

 

[FvM]

It may be an oscillation caused by excessive capacitive load, but the frequency seems to high for a LF356. Did you place a different amplifier type?

 

[jg001]

Clarification (sorry for not mentioning it in OP):
The plots that you see are not from simulation – they are measured waveforms from my circuit.

FvM – No, I didn't use different amps.

 

[FvM]

No, I didn't use different amps

O.K., capacitive load should be considered as cause anyway, possibly in combination with insufficient supply bypass. A 50 to 100 ohm series resistor should be enough to stop oscilllations.

 

[LvW]

Clarification (sorry for not mentioning it in OP):
The plots that you see are not from simulation – they are measured waveforms from my circuit.

Sorry - and it was my fault to recognize only the first two plots which were looking quite OK.
If it is not to much redesign for you I really recommend to use gain-of-two amplifiers (instead of unity gain).
At first, this improves somewhat the tendency to oscillation, and secondly, you can use two equal valued capacitors.

Simple formulas: dc gain Ao=2, Qp=sqrt(1/k) with k=R3/R2, pole frequency wp=1/(R2*C*sqrt(k)).
Formulas (indices) apply to the 1st stage.

 

[jg001]

Thank you both LvW and FvM. I will try the "gain of two" method.

Also probably should confess that in a haste to build the circuit (what I showed you is just a part of a slightly bigger circuit), I forgot to use supply bypass caps for all the ICs. In my next iteration, I'll definitely use them.

 

[LvW]

OK, good idea. Good luck. For the gain-of-two amplifier, of course, both feedback resistors should be as equal as possible.
The circuit is relatively gain-sensitive.

 

[jg001]

An update for any future reference:

My HF oscillation seems to be due to power supply bypassing (lack thereof). I added small bypass caps as close to the power pins of each LF356 chips. The improvement in output is remarkable! See for yourself in the attached PDF. I didn't even optimize the value of the bypass caps. With more care, I'm sure the output can be improved even further.

Just thought I should share.