EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: monkeysuncle on January 26, 2016, 08:08:22 pm
-
I'm working on a design for a simple signal generator to run off of a 9v battery. A bunch of op amps and the obligatory 555. I want a 0v reference to sit sedately between the two supply rails, so I thought I would just use a resistive divider and an op amp voltage follower to buffer it.
Then it occurred to me there was something I really didn't understand. When I build a linear power supply, I can start with an LM317 or 7805 or whatever, and I hang a whole bunch of capacitance on its output lead to improve its response to load transients. But with an op amp, they always tell me "never put a lot of capacitance on the output or it will oscillate." So can somebody tell me why a capacitor stabilizes a linear voltage regulator like an LM317, but destabilizes common op amps like a TL072?
--Dan Menes
-
The open-loop output impedance (before they connected the feedback internally) of a non-LDO regulator is relatively low: the output is an emitter follower with a relatively large transistor.
The open-loop output impedance of a normal op amp is considerably higher: TI's data sheet has an electrical diagram with 64 + 128 = 192 ohms at the output. The pole with a substantial load capacitor will affect the phase margin.
-
For op-amps, I'd recommend the "stability 101" video series on youtube by analog devices
https://www.youtube.com/watch?v=QHIninAg4Is (https://www.youtube.com/watch?v=QHIninAg4Is)
That video is relevant to the op-amp portion of the question, see other the other videos in the series for more explanation about the A and B terms and some of the other things he's talking about. In short, though, the op-amp output resistance forms a RC network with the output cap, which adds a phase shift, and if this is too big before the gain is below zero, you can get oscillations. I'd imagine the same applies to linear regulators which have a specification for minimum output cap ESR, its just that linear regulators have a much much lower output resistance than an op-amp, so can handle more capacitance.
For linear regs with no minimum output ESR, I'll defer to those more knowledgeable to analog shenanigans than me.
-
Thank you. Does that mean if I buffer the output of my TL072 (insert favorite "jellybean" op amp here) with an emitter follower, or perhaps a push-pull complementary pair, then I don't have to sweat the output capacitance?
-
Usually
Amplifiers for driving un-matched, heavy and / or reactive loads at frequencies higher than a couple dozen kHz with good linearity are art & science. For most applications the simple schemes are sufficient and will be able to drive a couple nF up to a few hundred kHz.
-
Thank you, all. You've told me what I need to know.
--Dan
-
Thank you. Does that mean if I buffer the output of my TL072 (insert favorite "jellybean" op amp here) with an emitter follower, or perhaps a push-pull complementary pair, then I don't have to sweat the output capacitance?
Hi
Maybe yes, maybe no.
Your TL072 op amp cares about load impedance up around 5 to 10 MHz. Get it's output phase messed up in that region and you will have a stability issue. If the output buffer is inside the feedback loop, it's phase will add to the TL072's. Either way, there is no guarantee. A free circuit analysis package like LT Spice is your friend in this case.
Bob