I would like to have a clean output from a linear regulator (variable 1-5V) but also a small heatsink on it, so I thought about placing a ready-to-use step-down module before it, and to wire it so that the voltage drop on the linear regulator is almost constant.

Since the step-down (buck) are non-isolated, in practice they also have three pins only.

My idea is to have a circuit like this:


simulate this circuit – Schematic created using CircuitLab

In other words, the potentiometer defines the final voltage at the output. The DC-DC regulator is set to have a constant 4V voltage on its output, that corresponds to the drop between input and output of the linear regulator.

Provided I use capacitors (ceramic and electrolytic) and ferrites between step-down and linear regulator, plus a ferrite also after the linear regulator, would it work? I have no other idea about how to set the same thing without using a double potentiometer.


I wrongly used the ADJ pin of the step-down as if it were the ADJ pin of the linear regulator. It was wrong, because the step-down is a complete module, where I use the GND line. I updated the schematics.

  • 1
    \$\begingroup\$ Nice idea. But... I think a switching regulator ground pin is not just a sense pin; real current flows there. Unlike an LM317, the switcher has power components, filter capacitors etc, connected to ground. You need a switching regulator with separate GND and Sense/Set pins. \$\endgroup\$ – tomnexus Aug 5 '15 at 9:49
  • \$\begingroup\$ Yes, it is absolutely possible to create such a thing. I know them as hybrid power supplies and they offer excellent low noise figures while being in efficiency between linear and smps. However I think there is much more to it than just slabbing a linear regulator after an smps \$\endgroup\$ – PlasmaHH Aug 5 '15 at 9:54
  • \$\begingroup\$ @Dzarda my schematics was incorrect, since I mentioned using a ready-to-use step-down module. I updated the schematics but i see it won't work, as you mentioned current flows back into the GND rail and that means something is wrong. \$\endgroup\$ – FarO Aug 5 '15 at 10:13
  • \$\begingroup\$ @tomnexus I didn't think about using the Sense pin of the buck regulator. Maybe you should articulate more and post it as answer \$\endgroup\$ – FarO Aug 5 '15 at 10:14
  • \$\begingroup\$ @PlasmaHH I know the concept is feasible, it's the implementation using ready-made step-down modules that I'm interested in. \$\endgroup\$ – FarO Aug 5 '15 at 10:14

In the LM317 datasheet you can find and interesting scheme showing how to use two LM317s as a linear regulator with preregulation. The first LM317 maintains the dropout voltage of the second at about 5V.

If you can find a step-down IC with a sense input which works the same as the LM317, you could do the same trick:

enter image description here

The important thing to note is that the LM317 works by maintaining a fixed reference voltage (~1.25V) between OUT and ADJ terminals. Find a step-down chip that keeps a fixed reference voltage between its sense input and its output and you are done (the sense input must draw negligible current).

In that schematics the calculations are done in this way: the current in R1 is set by the reference voltage across ADJ and OUT:

$$ I_{R1} = \dfrac {V_{ref}} {R_1} $$

neglecting the ~40μA absorbed by ADJ \$I_{R1}\$ flows through R2, causing a voltage drop:

$$ V_{R2} = R_2 \cdot I_{R1} = V_{ref} \dfrac {R_2} {R_1} $$

Add the reference voltage across R1, i.e. Vref, and you get the dropout voltage of the second regulator:

$$ V_{DO} = V_{ref} + V_{R2} = V_{ref} \left(1 + \dfrac {R_2} {R_1} \right) = 1.25V \left(1 + \dfrac {720 \Omega} {240 \Omega} \right) = 5V $$

EDIT (to report more findings)

You can get some ideas from the following sources.

This sites sports a circuit that does what you want: LM317 Adjustable 20V, 1.5A Supply with Simple SMPS Tracking Preregulator (it warns the circuit was only simulated and not built).

Some application notes on the subject that can be relevant or interesting:

enter image description here

The EEVblog video #260 talks about simulating a switching preregulator in the context of Dave Jones μSupply project (there are other videos in that series).

  • \$\begingroup\$ What about using one of these and the GND as sense? pololu.com/product/2851 (it's a complete step-down module, non-isolated meaning that V-_in is directly connected to V-_out) \$\endgroup\$ – FarO Aug 6 '15 at 11:47

Not a real answer

Tried to redraw your circuit so it would make a bit more sense to myself and maybe someone else also:


simulate this circuit – Schematic created using CircuitLab

Things to note:

  • The LDO has only a high-side regulation, meaning it won't regulate down voltage fed in via external means. R3 and R4 provide such path.
  • R3 and R4 are implementation-defined, really. In your updated schematic you have omitted them.
  • The Buck's output has to be maintained circa 2V above the LDO's output, due to its voltage drop.

I would try to simulate this thing. Seems like a perfect job for LTSpice.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.