I have seen some images in which two linear regulators of the same type (basically the same part number) have their outputs tied together. I believe this is being done to increase the total current drive capacity since the current will be shared between the two voltage regulators.
Can any off the shelf linear regulator IC be connected in this way? Are there any drawbacks to this approach?
I believe this is being done to increase the total current drive capacity since the current will be shared between the two voltage regulators.
Theoretically yes, but practically no, or not always. As others have already stated, the circuits, especially the power transistors, should be well matched or else the sharing will be poor or even impossible.
I prepared an LTspice simulation to show what could possibly happen in practice.
I designed a 5V linear regulator with discrete components, and applied dynamic load (you'll see pulsating currents) to test its dynamic response.
See the regulators in dashed boxes in the image below:
The regulators are basically identical but there are some slight differences: Power transistors (formed as Darlington) and the BJTs forming stable ref voltage at the input section are slightly different. But the BJTs forming the differential amplifier are identical. Also the feedback resistors have a very small difference (2k28 vs 2k25, and 2k23 vs 2k2) but the output voltage difference is still within the tolerance (1%).
Both regulators are able to source 1 A current individually, with almost the same performance. See the output voltage waveform when they are not in parallel:
Green and grey lines are the output voltages. You can see the very small, negligible difference.
When the regulators are connected in parallel and the same load is applied, the "combined" output voltage has almost the same waveform:
Now see the output currents (pulsating) of each regulator:
Ignore the spikes for now. As you can see, the regulator at the bottom sources all of the current (green waveform - output current reaches to 1 A) whilst the top one sources only a few milliamps.
Note that if the feedback resistors were identical while the other BJT differences mentioned above were kept the same, the sharing would have been possible – may or may not be 50-50.
Summary
It sure depends on the design but sharing will not always be possible.
Can any off the shelf linear regulator IC be connected in this way?
Generally no.
I believe this is being done to increase the total current drive capacity since the current will be shared between the two voltage regulators.
The current will be shared in some way between the two regulators, but without regulators designed for that purpose, or steps taken to adjust the regulators to share, the current will not be shared equally. One regulator will tend to hog all the current, with the other being idle. Even a few mV offset in the set voltage of the two regulators will cause a large imbalance in their output currents.
The simplest way to try to get two arbitrary regulators to share nicely is to put a small resistor in series with each output. Now that 'few mV' difference in output voltage will be swamped by 'a few hundred mV' drop across the resistors. The resistor value is a compromise, large enough to share current nicely, small enough to meet your load regulation and excess heating requirements.
A better but more complicated way is to put the resistors on the regulator inputs, and use the measured current to control the voltage output of all of the regulators but one so that they draw the same current.
Regulators designed for direct parallel connection usually have this sort of sharing control built in.
