Teensy 3.2 ADC opamp buffering

[Nominal Animal]

I'm considering using a TI TLV2372 rail-to-rail I/O (dual) opamp to buffer certain signals to Teensy 3.2 ADC.

As suggested here by Paul, I'd use a resistor between the opamp output and the analog input pin, and a 10 nF capacitor from the analog input pin to ground to act as a reservoir for the ADC, to avoid ringing.

Paul suggested using an 1 kΩ resistor, but I'm hoping to keep the bandwidth relatively high, say 50 kHz or better, so a much smaller resistance is needed. (I'm hoping to catch transients; with one half of the dual opamp a buffer after a voltage divider, and the other half in an instrumentation amplifier configuration over a current sense resistor, to monitor both voltage and current to an Odroid HC1 single-board computer. Or any other 5V device consuming up to 8A, I guess. Full schematics are here.)

Now, the TLV2372 datasheet claims that "a minimum value of 20 Ω should work well for most applications", "for capacitive loads of greater than 10 pF". I'm considering something like a 20 Ω to 47 Ω resistor.

Unfortunately, I dont have the test equipment to verify that has any hope of working at all in practice. In particular, I don't have an oscilloscope or a signal generator, which makes investigating this .. difficult.

Does anyone have practical experience using such buffering with a Teensy 3.2, targeting say 50 - 150 kSamples/sec?

 

[WMXZ]

Unfortunately, I dont have the test equipment to verify that has any hope of working at all in practice. In particular, I don't have an oscilloscope or a signal generator, which makes investigating this .. difficult.

Does anyone have practical experience using such buffering with a Teensy 3.2, targeting say 50 - 150 kSamples/sec?

This may not be explicit applicable, but it explains in the text how to choose both resistor and capacitor.

 

[Nominal Animal]

Thanks, WMXZ!

According to this e2e.ti.com answer, the TLV237x opamps have an output impedance of about 300 Ω (at frequencies from 2 kHz to 2MHz). Combined with the TI slau51a information, the minimum output isolation resistor value is about 33 Ω. Assuming the Teensy 3.2 ADC sampling capacitor capacitance is ≤ 500 pF (so the 10 nF capacitor suggestion fulfills the 20× recommendation also in TI slau51a), 33 Ω + 10 nF (and the 300 Ω output impedance from this particular opamp) gives an open-loop response with a pole at 48 kHz and zero at 530 kHz, and 39Ω + 10nF with 47 kHz and 408 kHz, respectively.

Therefore, according to that, with this particular opamp (with 300 Ω output impedance and 3 MHz bandwidth) a 33 Ω or 39 Ω resistor and 10 nF capacitor looks like it should work.

In comparison, the LMV358A that Paul mentioned he likes to try first in such situations, has an output impedance of about 1200 Ω, which would indicate it would require a 130 Ω or larger resistor when used with a 10 nF capacitor (using the formulae in slau51a), leading to a pole at 37 kHz and zero at 122 kHz.

What I'd still want to know, is how well this works in practice? Is there a practical reason (other than reducing the low-pass filter frequency range) why one would use a bigger resistor than that? Is 130 Ω (150 Ω) with 10 nF sufficient to prevent ringing on the Teensy 3.2 ADC inputs when used with an LMV358A?