The datasheet for the OP37 boasts that the device is a low-noise, high-speed, precision op-amp. However, one reason why one may choose to not use the OP37 is that it is not unity gain stable. That is, one cannot use the OP37 in a voltage-follower configuration.

This feature of the OP37 invites the question of why a unity gain un-stable op-amp would be implemented in the first place. One reasonable supposition is that it provides a higher speed than any other comparably priced op-amps with otherwise similar features. Are there any other reasons why a manufacturer would choose to implement a unity gain unstable op-amp? Or is speed-for-price the only reason?

For reference, here is the Bode plot for the op-amp.

enter image description here

  • \$\begingroup\$ Short answer: sacrificing stability allows more GBW (gain bandwidth product). \$\endgroup\$
    – glen_geek
    Commented May 19 at 18:01
  • 1
    \$\begingroup\$ Math, just another thought. The very worst possible case for stability is when you have maximum feedback to the (-) input, which is when you use a wire to connect the output to the (-) input and get a non-inverting gain of exactly 1 from (+) input to output. (You can't get a non-inverting gain less than 1.) \$\endgroup\$ Commented May 19 at 18:31
  • \$\begingroup\$ RF amplifiers very often are not unity gain stable. It has no use if you want gain anyway, and it cannot give the best performance. \$\endgroup\$ Commented May 21 at 12:05

5 Answers 5


A tale of two op amps....

The OP37 and OP27 are very similar internally, but their high-frequency compensation differs, making OP27 unity-gain stable, while OP37 is not. Simply by modifying an internal compensation capacitor from 120pf [OP27] to 150 pf [OP37], OP37 has an attractive gain @ 1 MHz of 33 dB while OP27 has only 17 dB.OP27 is unity-gain stable, while OP37 is not

Note that this particular OP27 data sheet shows the frequency scale in error - units of MHz. The scales of phase are somewhat twisted too - a clockwise/counterclockwise problem depending on which side of a clock you face.Analog devices OP27 data sheet

In addition, OP37 slews faster than OP27:

  • OP37 slew rate 17 V/microsecond (typical)
  • OP27 slew rate 2.8 V/microsecond (typical)
  • \$\begingroup\$ Interesting. The minimum GBW for the OP27 is 5 MHz. The minimum GBW for the OP37 is 45 MHz, but the minimum gain is 5, meaning that one is realistically limited to 9 MHz. So although the bandwidth of the OP37 is somewhat higher, the gain at high frequencies is significantly higher. Is this merely accomplished by changing the Miller feedback? \$\endgroup\$ Commented May 19 at 23:27
  • \$\begingroup\$ Seems so. The Texas Instruments 1982 Linear Circuits data book combines data sheets for OP27, OP37. All DC specs and noise specs are the same. There are 4 internal capacitors...as mentioned, only one capacitor differs in value between OP27 and OP37 variants. Their RC stabilizing network seems a bit more complex than other opamps. \$\endgroup\$
    – glen_geek
    Commented May 20 at 0:19

It's one of a handful of things you can sacrifice to get increased bandwidth. Decompensated op amps like this can have significantly higher gain-bandwidth product than unity-gain-stable ones. Not even just speed-for-price, but you can get better speed out of high-speed decompensated amps than you can out of any unity-stable standard op amp on the market.

They typically can't match the bandwidth of current-feedback amps, but they don't sacrifice precision to the extent that CFAs do; decompensated op amps are typically just as precise as their unity-stable counterparts.


An op-amp is a multi-stage amplifier, to which the user applies negative feedback to get the desired gain with high linearity.

Due to parasitic resistance and capacitance an amplifier inevitably ends up acting as a low-pass filter. This means that in addition to attenuation there is a phase shift. The more stages an amplifier has the more potential there is for phase shifts.

The frequency response of an amplifier with two or more stages (i.e. pretty much all op-amps) will contain multiple break frequencies. Arround each break frequency the phase shift increases. After the first break frequency there is about 90 degrees of phase shift, after the second break frequency there is about 180 degrees of phase shift (and so-on but we really only care about the first two).

A 180 degree phase shift turns negative feedback into positive feedback. That's a problem. If the "loop gain" of the feedback path at that point is one or more than the amplifier will oscillate. To avoid this the designer must ensure that the "loop gain" drops to less than 1, before the second break frequency is reached.

Generally ensuring this means artificially reducing the bandwidth of one of the stages of the amplifier by adding "compensation" capacitance.

A system with higher closed loop gain has lower "loop gain" in the feedback path, so needs less "compensation" capacitance to achieve stability. Different op-amps tackle this in different ways.

  • Some op-amps are compensated from the factory for unity gain operation. Sacrificing bandwidth for flexibility.
  • Some op-amps have a minimum gain specified by the manufacturer. Sacrificing flexibility for bandwidth.
  • Some op-amps have a pin to allow fitting compensation capacitance externally. This provides the flexibility to accommodate low-gain applications without sacrificing bandwidth in high gain applications. However bringing a signal that would otherwise be internal to the outside of the package is likely to bring design compromises of it's own.

All the other answers gave good explanation about what you can do when you use these amplifiers. Here's a very specific real scenario where one of these is useful.

When you need a good loop gain at frequencies relatively close to the GBW of the op-amp, you're probably looking to improve your out-of-band loop gain (i.e., improving the linearity of signals that fall out of your band of interest) that cannot be entirely suppressed by any sort of passive filtering preceding the amplifier.

The Real Scenario

Imagine you have 2 interferers at 1MHz and 1.1MHz, but your band of interest stops at 100kHz. If you have no loop gain to speak of using the unity-gain stable op-amps, there is nothing the amplifier can do to suppress the error signal for these interferers, thus we have a large error signal between the op-amp's inputs and the output produces horrible distortion; you're entirely dependent on your external filtering to dictate the amount of error signal at the input of your amplifier.

If there's still loop gain, and with some help of the up front filtering, the distortion produced by these out-of-band interferers can be reduced, hopefully enough to your specs.

In an IC design I did once, I had to give up some stability margins in order to meet the very tough out-of-band distortion requirements on my custom amplifier. The system designer made it explicit no change in the filter made out of discrete passives up front was possible.


This feature of the OP37 invites the question of why a unity gain un-stable op-amp would be implemented in the first place.

The simplest answer is: when the goal is to have gain more than 1 at high frequency.

Nowadays we have pretty fast unity gain stable opamps. But the same opamp without internal compensation would have higher gain at high frequencies. If you are not going to use unity gain anyway, you don't need to care whether it is stable at it.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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