TLC2252: Noise Issues on the Output

Hey Javier,

I think I figured out the root of the problem here. The reason you see the "noise" you mentioned when the filter's fc is increased from 215Hz to 1kHz is the interaction of the RC network of R2 and C1 with the closed-loop output impedance (Zout) of the TLC2252.

According to Figure 30 in the datasheet the closed-loop output impedance (Zout) is rising at 20dB/dec and thus Zout is acting like an inductor. Note the datasheet says “Zo”, but this plot is the closed-loop output impedance which I am referring to as Zout.

This problem can be determined by perfectly isolating the Zout of the op amp from the feedback capacitor C1 using a voltage-controlled voltage source (VCVS) at a gain of 1 in TINA. The asymmetry is removed when you run the simulation.

It is difficult to say where and what values of Zout are causing the overall asymmetry in your output because the sudden rising and falling of the square wave is causing the capacitors to momentarily short, but there are some simple solutions.

According to Figure 30 TLC2252 has Zout = 20 Ohms at 25 kHz and if you simulate the closed-loop bandwidth of your circuit you see the gain comes back up around 25 kHz. Meaning those higher frequencies of the square wave during the rise and fall won’t be as attenuated as much. See the closed-loop bandwidth gain below.

One solution is to add an RC filter externally to the output of the amp with R = 20 Ohms and fc = 25 kHz, thus C = 397.9 nF. This will attenuate down frequencies > 25 kHz that the amp is not as effective at attenuating. Additionally you’re matching the resistive Zout with the 20 Ohms. Of course you have to make sure this does not destabilize the overall filter. See solution and simulation below.

Another option is to use an amplifier with a larger bandwidth. This will prevent the Zout from rising too early and thus begin loading down C1. You want to make sure that the amplifier’s open-loop output impedance Zo is purely resistive as this will help for stability. I chose the OPA320 (20MHz GBW) and simulated this below.

Some other things to note are that the TLC2252 macromodel does not use the most up to date simulation modeling methods and some aspects could cause problems in your simulation. Either way you want to verify this is the real world. OPA320 will more accurately reflect the Zout and Zo characteristics in simulation. Although there are methods to verify Aol, Zout, and Zo in simulation compared to datasheet plots which I could help explain more if desired.

Also note that you are stretching the amplifier inputs pretty significantly so you will want to make sure the input stage of the amplifier does not have back-to-back input diodes. You could run the signal through a simple RC low-pass filter before the amplifier to reduce the intensity. Or just replace the entire filter with multiple RC filters as seen in the PDF I attached previously in this thread.

Best,

Peter Iliya

Precision Amplifiers Applications