Output impedance of the LM317 at 1Mhz?

The performamce of most active regulators will start to degrade above a mere 1KHz. As bwaslo alludes to, it is the impedance of the passive capacitors placed in shunt which will primarily determine the 1MHz regulator source impedance as seen by the load.

If having low source impedance supply at high frequencies is important then you will need to place high-frequency application supply bypass capacitors as close to the load as is possible.
At that frequency, it probably will be the ESR of the capacitor used at its output...

Perhaps but the one thing I know for sure is that this will happen if you get the ESR of the caps on the output too low.

The picture below is actually of an LM337 and it's a bit more sensitive to oscillation than the LM317 which did not exhibit this behaviour with the same set of caps on the output. Cured with a small resistor (~ 0R2) in series with the output pin.



2006-09-08 2:04 pm
Theory is a very nice thing....
until you confront it to the reality that is:

That the 317 could play the role of a perfect gyrator at such a frequency seemed very fishy, and I made a quick check.

My doubts were confirmed: the impedance I measured in the conditions of the graph was |6.6Ω|, less than 10° inductive. Almost purely resistive, and if you account for the inductor isolating the load from the output, the net result is purely resistive.

Here is the test circuit (LM317 is from ST):


  • Z317b.png
    8 KB · Views: 189
It is not Z rising with frequency which makes it inductive, but Z at a perpendicular angle.

Resonance with C is inconsistent with the resistive impedance which Elvee measured. Perhaps he could give more details of his measurement?

Yes that's a point. But Elvee measured it as voltage phase shift at the test generator end. I would expect no phase shift there as the generator output voltage is imposed. What should be measured is the phase shift of the current into that Zout.
Did you measure that Elvee?



2006-09-08 2:04 pm
The historic analogue process isn't known for breeding race horses.

Here is how a 2N3055 looks: up to ~250KHz, the impedance is inductive and rising, but then there is a reversal, and eventually for much higher frequencies it becomes capacitive


  • Z3055.png
    30.8 KB · Views: 127