13
\$\begingroup\$

Working on a project where we're using a couple of OP470 Quad Op amps. There are 2 unused op amps, and I need to buffer a signal coming from a sensor (this is according to the sensor's data-sheet). I'd like to use one of the extra op-amps. I know that theoretically you can buffer a signal using the negative feedback of the op amp as shown below:

schematic

simulate this circuit – Schematic created using CircuitLab

However, I also vaguely remember something about the dangers of self-oscillation and op amp stability. The OP470 is unity gain stable. Here is the datasheet:

http://www.analog.com/static/imported-files/data_sheets/OP470.pdf

My question is, is it safe to use an op amp in this configuration as a buffer without worrying about self-oscillation? Is there anything else I need to take into consideration?

\$\endgroup\$
  • 1
    \$\begingroup\$ don't accept the answer too quickly , let other's also to share their answers with you. Not telling that accepted answer is bad. But there may be more good alternative solutions than that. Just wait for some time. \$\endgroup\$ – Standard Sandun Jun 26 '13 at 16:06
  • 1
    \$\begingroup\$ (1) Don't forget it needs power supplies (2) does the signal from your sensor fit nicely within the power rails of the buffer (3) how far do you want to send the buffered output (4) what are you sending it to (5) Does there need to be any signal input offset or gain correction (6) how long is the input feed connection (7) what bandwidth is the signal you are inputting (8) should it be filtered to suit what might be downstream? \$\endgroup\$ – Andy aka Jun 26 '13 at 16:17
16
\$\begingroup\$

If the datasheet says it is unity gain stable, then yes. Unity gain stable means precisely that the op-amp will be stable when used as you describe.

Do be sure the feedback path is short. If you make it long, then its inductance is increased, and maybe weird things will happen. No need to be extremely paranoid on this point; just don't go routing it 10 inches around the board and you should be fine.

\$\endgroup\$
  • 8
    \$\begingroup\$ Usually the output and negative input pins are adjacent, so keeping this feedback connection short is easy. This is one of the things to route first as a high priority. Also, don't forget the bypass cap. Without one the amp can do weird stuff. \$\endgroup\$ – Olin Lathrop Jun 26 '13 at 14:51
  • 1
    \$\begingroup\$ "the bypass cap" is the first mention of a capacitor in this post. I'm assuming you mean on the power supply for the op-amp? \$\endgroup\$ – NickHalden Jun 26 '13 at 16:08
  • 2
    \$\begingroup\$ @NickHalden A bypass cap is also known as a power-supply decoupling capacitor. See What is a decoupling capacitor and how do I know if I need one? \$\endgroup\$ – Phil Frost Jun 26 '13 at 16:11
  • \$\begingroup\$ @Phil Frost: Should there be a Low Pass filter on the positive input of the op amp? \$\endgroup\$ – Dor Jun 26 '13 at 20:59
  • \$\begingroup\$ @Dor only if you want it for some other reason. It's not required for stable operation. \$\endgroup\$ – Phil Frost Jun 27 '13 at 2:56
8
\$\begingroup\$

Using the OP470 for a unity gain amplifier or buffer is not a bad choice. What's important will be the phase margin where the open loop gain crosses 0dB. It is quite common for opamps to have 45 degrees of phase margin at their open loop crossover frequency, which would result in an overshoot of 1.3 times Vin followed by ringing. That is not the case with the OP470, if you look at TPC (Typical Performance Characteristics) 16 in the datasheet (page 7), you will see that the phase margin at crossover is 58 degrees. Now, optimal pulse response you would want a phase margin of 60 degrees, so 58 degrees is pretty good.

Small signal pulse response of the OP470 as a unity gain buffer is shown in TPC 27 of the datasheet. You will see there that the pulse response has hardly any overshoot and no ringing.

Possible concerns:

  • Output loading. You don't show what the load for the buffer is, but capacitive loads can degrade phase margin and cause poor response. Output impedance of the OP470 rises sharply as it approaches crossover, so capacitive loads over 100 or 200 pF would be expected to degrade phase margin. There is a section of the datasheet (page 12) that addresses driving capacitive loads up to 1000pF.

  • Feedback resistor from the output to the negative input. As Phil Frost says, you will want to keep the feedback path relatively short. But, sometimes for testing or serviceability a feedback resistor will be used. If you do this you will want to put the resistor close to the negative input, and you will want to keep the value less than 500 Ohms and greater than 100 Ohms for best response. The datasheet covers this in the section about unity-gain buffers (page 12).

  • OP470 is BJT technology, so no rail to rail operation. You need to make sure the input voltages don't get closer to the bias rails than about 4 volts, and you have to be ok with the output voltage being no closer to the rails than about 3 volts. Only you know your system well enough to say if this matters.

\$\endgroup\$
2
\$\begingroup\$

The output of that op amp is probably driving a capacitive load. I ran into a similar issue with a high slew rate op amp. The solution described in the datasheet is to put a resistor of about 47 ohms after the negative feedback. This solution immediately solved my oscillations.

The quote:

The LT1632/LT1633 are wideband amplifiers that can drive capacitive loads up to 200pF on ±15V supplies in a unity-gain configuration. On a 3V supply, the capacitive load should be kept to less than 100pF. When there is a need to drive larger capacitive loads, a resistor of 20Ω to 50Ω should be connected between the output and the capacitive load. The feedback should still be taken from the output so that the resistor isolates the capacitive load to ensure stability.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.