Effect of large battery on loop response

[gnurf]

I'm not sure I'm providing enough information in order to get a meaningful reply, but I'll try anyway:

Lets say you have a regulator of some sort, with a LC filter on the output, and some sort of feedback loop with an unknown transient response (i.e. you lack info on the system's phase margin, unity gain, etc).

Now if you hook a large battery (with very low internal resistance, if that changes anything) onto the output so the regulator works both as a charger and a power provider to the loads, is it possible to say anything a priori about how this affects the system stability? I understand that if the battery can be seen as a large capacitor the LC cutoff frequency drops, but what I'm looking for is an argument along the lines of "since there's a large battery on the output, stability is guaranteed...".

Does such an argument exist? Is it possible to say anything general about a situation like the above?

 

[berkeman]

I'm not sure I'm providing enough information in order to get a meaningful reply, but I'll try anyway:

Lets say you have a regulator of some sort, with a LC filter on the output, and some sort of feedback loop with an unknown transient response (i.e. you lack info on the system's phase margin, unity gain, etc).

Now if you hook a large battery (with very low internal resistance, if that changes anything) onto the output so the regulator works both as a charger and a power provider to the loads, is it possible to say anything a priori about how this affects the system stability? I understand that if the battery can be seen as a large capacitor the LC cutoff frequency drops, but what I'm looking for is an argument along the lines of "since there's a large battery on the output, stability is guaranteed...".

Does such an argument exist? Is it possible to say anything general about a situation like the above?

In general, positive voltage regulators can only "pull up" -- they cannot pull an output voltage down if something else pulls the output voltage up. They are not like an amplifier, which is able to hold an output voltage steady whether you try to pull it down (with a load) or pull it up (with a higher battery voltage, for example).

So if you put a battery on the output, and the battery voltage matches the output voltage, then the regulator can only go unstable in the pull-up direction, and if the battery has more current capability than the regulator, it should hold the output pretty stable.

If the battery has a lower voltage than the regulator, then the regulator will likely go into its current limit behavior, whatever that is (foldback, burp, straight droop, whatever).

What is the relationship between the battery voltage and the normal output voltage of the regulator?

 

[Mike_In_Plano]

Back in the days before current mode control had caught on, voltage mode control loops were difficult to stabilize over a wide control range. The switching supply had phase lag, and the output capacitance represented lag, and load interacted with the output cap to change the lag... SO, in the absolute sense, no.. There are some systems that have to be tweeked to their load.
Fortunately, most modern systems use current mode control. Because of this, there is less lag in the switching supply / control area, unless the designer put it in for a reason. For most of these, extra "capacitance" on the output is okay.
However, many designers want an integrator in the output to reduce the steady state error. Unless your battery has some impedance, or your supply is protected from over current, the integrator can attempt to drive the supply to a higher voltage than the battery's ready for (at that moment).
So, no, there isn't a universal solution. But then again, many supplies don't function as well as you would expect. There are typically "sweet spots" within their specification, that can strain or even destroy a supply. I went through about 5 manufacturers once to find one that met every single one of his specs without getting damaged.

 

[gnurf]

Thanks for the replies. I'll try to answer your question with a more concrete example:

Lets say the regulator in question is a buck converter similar to this one: http://upload.wikimedia.org/wikipedia/commons/f/f0/Buck_conventions.svg, and that the input is a current-limited voltage source of 1A/6V and the R represents a battery with an operating range between 3-4 volts, and another load in parallel (I'll get back to it).

In an attempt to also trick Mike_In_Plano (Plano?? I've always read "Mike in Piano"!) to respond again, the control scheme is current-mode control where the current is sensed directly after the inductor and processed in a digital control loop whose sole purpose is to maximize the current through the sense resistor. I.e the control loop doesn't care what happens beyond that point.

Now, if the regulator is at a working point where it is sourcing the maximum amount of current that the current-limited source can handle, and thus we're charging the battery at 1A, when suddenly (as in a sharp transient) the other load (say a RF transceiver) draws 2A of current.

Without the battery, I'm thinking a transient like that asks big questions of the regulator and its response, but with a battery present, am I correct to assume it will "buffer" such a current draw and make the regulator's (which may or may not have a slow loop with a lot of processing and what not) transient response less critical?

EDIT: Assume that the large battery easily can handle 2A and has a very low internal resistance. Also, if it's possible to say something about how a voltage-mode control would change the situation, I'd be interested in that too. Thanks.