# Best topology for active attenuator

What are the advantages and drawbacks of each op-amp attenuator topology ? Which one would you use ?

Resistor divider followed by a buffer :

Inverting op amp attenuator :

• "Which one would you use?" This depends on a lot of things - it's very broad... Do you have any more context? Commented Feb 19, 2015 at 23:07
• The best one is the one that meets your requirements. Do you need high input impedance? Do you need low output impedance? Do you want to invert the signal or not? Commented Feb 19, 2015 at 23:16

As the comments say, it depends on what you want to do. But I want to point out a dangerous thing about your second circuit: if you run with a gain of less than unity, the op-amp may oscillate.

Many op-amps are specified as "unity-gain stable" but they will indeed oscillate if your gain is less than unity. TL074 / 084 is an example.

Other op-amps such as the NE5534 are stable only with gain greater than 5. If you run a NE5534 with gain less than 5, it will oscillate.

There is a cure for this oscillation problem - you have to add noise gain to the circuit. You do that simply by adding a resistor from the op-amp (-) input to ground such that the gain of that resistor and feedback resistor is at least unity. Then you can have a larger input resistor (R2) for gain less than unity.

I ran into this exact problem almost 30 years ago and either EDN or Electronic Design did a good article on the problem some time after that. By then, I had already figured out what the problem was and fixed it but the magazine article went into a lot of detail that I hadn't thought about.

• I think the "dangerous" situation is the same as for each (positive) unity gain amplifier. And - yes - this additional resistor improves the phase margin. But I dont understand the R2 argument. Why should R2 be larger?
– LvW
Commented Feb 20, 2015 at 8:21
• I don’t think decreasing the gain below unity makes unity-gain stable amps oscillate. In the limit of large attenuation, R2=0, so to noise the circuit is basically a follower. If the opamp is unity-gain stable, no oscillations. Commented Aug 10, 2021 at 11:00

Jim Williams advice* was to "Always invert, unless you can't."
That said, I made a switched attenuator stage and found the non-inverting was easier.

*From one of his "Analog Circuit Design" compendiums.