Capacitively Loaded Op-Amps - John Dunn, Consultant, Ambertec, P.E., P.C.

Op-amp stability or instability when driving capacitive loads is sometimes an area of difficulty.

The test fixture sketched below was used to examine the transient responses of several common op-amps when feeding a moderate (380 pF) load capacitance.  

 

The test results for a small variety of devices constitute a cautionary tale:

TLE2062

A widely used device, but apparently unstable in this circumstance, especially on the positive going output edge where non-linear overshoot is approximately 400% and therefore beyond the range of this waveform presentation.  

 

 

OPA2234P

 

Beautifully stable, although slower than TLE2062 ( note the 2 uSec / Div. horizontal scale ). This part's speed is adequate however for speech audio bandwidth to 3 KHz.  

 

LF412CN

Nominally stable, but negative going edge response was deceptive. When only another 150 pF was paralleled with the existing 380 pF, this device went into vigorous oscillation. 

 

LF442CN

Stable and not apparently sensitive to additional capacitance. The above 150 pF affected the transient response waveforms only slightly. This device requires three volts of rail voltage headroom into a 10K load.    

 

TL072CN

A venerable item used in many designs, but its very poor stability is almost as bad as TLE2062 on positive going edge and worse than TLE2062 on negative going edge.

 

Comments

[John Stoughton]

And a 22 ohm resistor between opamp output and feedback tap might make all of the damped oscillations look a lot better (if you have headroom on the rails), no?

[John Dunn]

Maybe and maybe not. Please see this related item:

http://licn.typepad.com/my_weblog/2011/09/beware-of-loop-gain-john-dunn-consultant-ambertec-pe-pc.html

[Victor Koren]

Look for single stage OpAmps like AD817, they have lower open loop gain but are very fast and are stable with large capacitive load. I have used it extensively in the past.

[Andy Fierman]

A good article on the subject can be found here:

http://www.analog.com/library/analogDialogue/archives/31-2/appleng.html