Can LM337L be Used in Preregulator Mode

I have been thinking the same question, LM317 as pre-regulator, tracking regulator or cascading LM317s to compete against shunt regulator.

Are you concerned about line or load rejection primarily? I doubt you'll beat a shunt for the latter. LM337 sucks fairly badly for HF line rejection, even in comparison with LM317. I have a shunt design on my blog which gives 10mohm Zount below 10kHz, in conjunction with an LM317 CCS feeding it I suspect it'll give 120dB+ line rejection (though I've not measured it).

Mark Johnson

Paid Member
2011-05-27 3:27 pm
Silicon Valley
If your (Vin - Vout) voltage difference is large enough to support both a regulator and a pre-regulator, I suggest it's a good idea to use a discrete (rather than an IC) as the pre-regulator. Especially so when you're regulating the negative rail using an LM337 as standiy is doing.

The LM337 uses a darlington NPN series pass transistor, which isn't so great for HF line rejection in a negative voltage regulator, as abraxalito remarks. You'll get far far better results in a negative v.r. using a PNP (or Pchannel MOS) series pass transistor in the pre-regulator. The downstream LM337 provides short circuit protection and overcurrent protection.

You can offset the PNP's base voltage from either (1) the final output, which topology goes by the names "tracking" pre-regulator, or "bootstrapped" pre-regulator, or else (2) from ground. I prefer option (2) since ground is by definition the lowest noise point in circuit, and since option (2) allows you to install massive bypass caps at the pre-regulator without affecting the feedback loop and phase margin of the final regulator.

So: Vin_raw -> PNP pass transistor configured as a capacitance multiplier -> LM337 -> Vout.
Last edited:


2009-06-24 10:49 pm
There's also (3) lowpass filtering of the supply to the regulator. This can work well for the 3x7L variants when used with headroom and low frequency (e.g. audio) loads as chip inductors are cheap. Pay attention to bypass ladder selection; good layout and MLCC selection to parallel the 'lytics can deepen the stopband by 20dB.
I built a bench supply using the Acoustica tracking preregulator circuit. The positive side works perfect. The negative side will not hold the voltage under load. For instance, with no load the voltage is -13.8v, with 100ohms the value moves to -12v. The positive side does not move. I tested a standard lm337 circuit and it also holds fine. I adjusted the lm337 preregulator circuit for the different pinouts and reversed the LED. I am using a center tapped transformer and a full wave rectifier to feed it. The input voltage is 40v. for each side.
I am stumped on this one. Anyone have an idea what is going on.