I've build the following linear power supply:

enter image description here

Before building it I tried to model it in LTspice. I couldn't find an LTspice model of L4941BV LDO (datasheet [PDF]) or a suitable replacement. Thus I used the following "naive" model:

enter image description here

R1, R2, L1 and L2 are measured resistance and inductance of a transformer windings. V1 is 50Hz, amplitude: 310 V. Here is Vout:

enter image description here

My idea was following. Let's say that all extra voltage turns into heat, and also account for ~10 mA quiescent current of LDO itself. In other words assume that this circuit's output power is ~0.95W.

Naturally this model turns out to be inadequate. It predicts PF 0.442 and efficiency 32%. The real device can produce up to 217 mA before a noticeable ripple (60 mV) will be observed. Under these circumstances it's measured PF is 0.925, consumed power - 2.75 W and efficiency - about 39.5%.

I tried to play with values used in the model but didn't come even close to observed PF and efficiency. Thus it looks like a less naive model is required.

The problem is I have no clue how to make an accurate model of an LDO. Could you please explain how it can be done or at least suggest some literature on the subject.

  • 3
    \$\begingroup\$ I'm quite confused here, what you have built is not even near an LDO, it's not regulating anything at all, just a plain old unregulated AC-DC converter. Did you forget to show your "naive" version? \$\endgroup\$
    – pipe
    Jun 25 '20 at 11:39
  • \$\begingroup\$ Maybe replace the current sink with a ~30 ohm resistor and see if you can get the PF and voltages to agree with reality. Speaking of reality, in reality you would want a lot more voltage at the input of the regulator to account for mains variations. That particular regulator is a lateral PNP type that has an Iq spike near dropout, though not as severe as some, so it’s not as simple to accurately model as, say, a modern CMOS type. \$\endgroup\$ Jun 25 '20 at 11:44
  • 1
    \$\begingroup\$ You can use "NCP1117ST50T3G" spice model.Its output voltage 5V fixed like L4941BV and 1A. you can find model in the link .mouser.com.tr/ProductDetail/ON-Semiconductor/… \$\endgroup\$
    – GuneyBoss
    Jun 25 '20 at 11:52
  • \$\begingroup\$ @GuneyBoss The LM1117 types have much higher (like 5x higher) dropout voltage than the OP's regulator. They're really semi-LDO types. \$\endgroup\$ Jun 25 '20 at 12:14
  • 2
    \$\begingroup\$ @GuneyBoss At OP's 200mA operating current, the typical dropout of the NCP1117 is about 950mV compared to 150mV for the L4941. \$\endgroup\$ Jun 25 '20 at 14:03

If you want to understand how LDOs work and how to simulate them i suggest much more reading. https://www.analog.com/en/analog-dialogue/articles/low-dropout-regulators.html is a decent start and I'm certain you could find a lot more.

Alternatively you could use a SPICE model of a LDO IC that is close to the one you have. You can find some at : https://www.pspice.com/models/rohm

  • \$\begingroup\$ I guess I didn't explain the situation well enough. LDO was used next to the circuit shown, on OUT. I edited the text accordingly. \$\endgroup\$ Jun 25 '20 at 15:03
  • \$\begingroup\$ Ok, I understand. Can you show what you used as an LDO then ? \$\endgroup\$ Jun 25 '20 at 15:05
  • 1
    \$\begingroup\$ I will upload a complete schematic shortly. \$\endgroup\$ Jun 25 '20 at 15:06
  • \$\begingroup\$ The question was updated. \$\endgroup\$ Jun 25 '20 at 15:27

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.