Any good examples of using a switching regulator as a preregulator for an LDO?

[slburris]

[stepping out of my comfort zone of digital electronics into analog]

I'm trying to educate myself on designs of using a switcher as a preregulator to just above the drop out voltage of an LDO.  Does anyone have examples of this they could point me to?

The essential problem as I understand it is that linear regulators are really good at rejecting 120hz noise, but suck at rejecting higher frequency ripple, such as you would get from the output of a switcher.  Presumably you would want to put an LC filter or the like between the switcher and LDO?

I'm aware of power supplies like the Hameg HM7044 which do the switcher - linear thing, but looking at the schematic it looks like the switcher output is just sent through a pass transistor and they call it a day.

Hence I'm looking for more examples.

Scott

[mamalala]

I think the main reason for that arrangement is efficiency, and thus thermal management. Using a linear regulator to get, lets say, 3.3 volts from a 12 volts source means a lot of wasted power. Using a switcher to first generate something like 5 volts, and then a linear regulator to generate the final 3.3 volts is more efficient and produces less heat.

Also, with a suitably designed switcher there isn't much higher frequency noise to handle. Using good filter caps, and a decent linear reg should generate a rather clean and stable output.

Greetings,

Chris

[Bored@Work]

I'm trying to educate myself on designs of using a switcher as a preregulator to just above the drop out voltage of an LDO.  Does anyone have examples of this they could point me to?

Analog guru Jim Williams wrote several app notes about this

http://www.linear.com/docs/4099
http://www.linear.com/docs/4120
http://www.linear.com/docs/4126
http://www.linear.com/docs/11877

The pre-regulators he uses in these app notes are a bit crude, he must have had some fun building regulators from scratch instead of using switching ICs.

Dave's uSupply project also uses one, a circuit from the LT3080 datasheet. (however, the LT3080 has another problem, so I wouldn't use that as LDO, but the general concept is the same).

[T3sl4co1l]

There are also some old Unitrode (now TI) appnotes, which also cover magnetic amplifier uses.  Example: you can have a switching regulator with one secondary winding -- good for say 13.0V raw, to be LDO'd to 12.0V.  Then using mag amps, tap the AC down for lower voltages, like 5.5-6V for a 5V LDO'd output, and so on.  Modern computer PSUs (ATX+?) do this for generating the 3.3V outputs off the 5V winding.

Tim

[slburris]

These are perfect.  I hadn't thought of Jim Williams but I should have.  If he didn't address something then surely Bob Pease did.  That last app note, Minimizing Switching Regulator Residue in Linear Regulator Outputs, tackles this right on.


I looked at the LT3080-1 data sheet and the circuit on page 20 is fascinating.  Using an adjustable linear regulator and having its output drivie a MOSFET in the feedback loop of the switcher is clever.  It automatically adjusts the switching voltage for a small differential across the linear regulator even as the linear regulator voltage is changed.

I'm surprised there isn't an opportunity for oscillation though.  And it relies entirely on the ripple rejection of the 3080 to filter noise from the switcher, I think. 

More to study, thanks!

Scott

Analog guru Jim Williams wrote several app notes about this

http://www.linear.com/docs/4099
http://www.linear.com/docs/4120
http://www.linear.com/docs/4126
http://www.linear.com/docs/11877

The pre-regulators he uses in these app notes are a bit crude, he must have had some fun building regulators from scratch instead of using switching ICs.

Dave's uSupply project also uses one, a circuit from the LT3080 datasheet. (however, the LT3080 has another problem, so I wouldn't use that as LDO, but the general concept is the same).

[geekbot]

Check ianscottjohnston on youtube. He designed his linear power supply with a switching pre regulator and it's output has only 4mV RMS Noise! Isnt that great?

[nsayer]

I do this in my current J1772 Hydra design. I have a DC-DC converter module that goes from +12 to +/-15. That relatively unregulated output goes to the supply rails of an analog switch IC, but then a 7812/7912 pair drop the +/-15 down to +/-12, which are what's actually fed into the switch inputs to generate the required +/- 12vdc square wave for the J1772 pilot.

I went with over-voltage on the switch supply rails to insure that the switch was not being asked for rail-to-rail excursions

It turns out that that design is overly conservative and pricey. The next iteration is going to use Sziklai pairs (kinda) and a +/-12v converter module instead.