I am trying to design a regulator for 2.5V at 0.5A with 4.5V input which at switch on will go to 4.5V (ish) briefly and then settle to the 2.5V level. I had the idea of using an LT1117 in an LTspice simulation (I will actually be using LD1117 but the data sheets look the same). I added a capacitor, C1, across the feedback resistor but the output voltage oscillates so I tried it without C1 but there is still oscillation though at a lower level.

What is going on, please?

 

hi Albert,
Right click the LT1117 model an Select show the Model Test, run that model, its got the spikes.!!
The model is suspect.
E

Update:
Removing the pulsed load on the Model Test circuit removes the spikes.

 

I took the test circuit and modified the circuit to use the same vales as in #1 and it doesn't oscillate. Presumably the model used in the test circuit is different from what you get loading from the component selector?

[Edited to add the files]

 

Give C2 in the first sim a bit of ESR and the oscillation disappears.

 

hi A,
Tried , same results as yours.
Alec's ESR also kills the oscillation.
Suspect the oscillation is an artefact of the model.???
E
@Bordodynov

 

If I leave the ESR at zero but extend the stop time to 5s the oscillation disappears!
OMG

 

If I leave the ESR at zero but extend the stop time to 5s the oscillation disappears!

.... but not if you set the Maximum Timestep to something like 10uS .
It's looking like the model is quirky.

 

? some mods

 

Many DC stabilizers do not work well (not sustainably) with low serial resistances of capacitors. In fact, they are designed for use with oxide or tantalum output capacitors. For some stabilizers, it is explicitly stated that small ESR capacitors can be used at the output. If not specified, it is dangerous to use a capacitor with a small ESR.

 

Connecting C1 between IN and ADJ should do what you want, but C2 still needs a bit of ESR for stability.

 

In any case that circuit needs large values of capacitor to get anywhere near 1 second delay. I have come up with this discrete version instead (R5 is a pot to adjust the output volatge).
My guess (anybody agree or disagree?) is that the 'delay' capacitor here will encourage stability.

 

See

 

See
View attachment 194518

But these two both oscillate and neither does the high voltage at switch on.

 

hi,
Your discrete version works OK with a stepped load.
E

 

hi,
Your discrete version works OK with a stepped load.
E

Thanks. I think I will go with that.

 

But these two both oscillate and neither does the high voltage at switch on.

I applied 100 mV interference with a frequency of 50 kHz to check the quality of stabilization at a time interval of 1 ms. And my circuitry has no power-on emissions. This is not an oscillation.

 

Re : #12 , Bordodynov 2019.12.14 at 5:38 PM
R3 in your upper circuit depends on input (it may address "positively" to voltage regulation but contrary it'll cut peak-power draw at high input voltages - just it has a fuzzy function - should be sourced from regulated side - me thinks) . . . Woops! - i missed it's at a "Cut-OFF" shoulder - so it makes it even more dangerous as allowing more current at higher input ...

besides it takes some (minimum) tests to get (at least) a far hint about the voltage regulator's behaviour (the steady state gives it's "resistance value" only . . .)

the circuit has nothing more to do with this thread than that it's been under revision/debugging

__________________
other -- the schem. allows to adjust the output voltage with some modifications

 

Compare the two circuits. In the beginning I applied constant voltage. Then there was a disturbance, and at the end I started to change the current consumption (imitation of impulse load operation). I forgot to add that my scheme is thermostable.

 

See

 

? donno what the '1117 will think of such . . .
... the R2 value needs to be \( R_2=\frac1{\frac1{120}-\frac1{312}}=195 \)
the C4 needs an emptying chain . . . sometnig reverse diode in parallel with the R4 . . . ?and a path to GND blocked @/by +4.5V ON

/// the mosfet acts as (sharp as) a darlington . . . at that output current it can be used as a switching element (a power switch) -- but i don't think the completed schematic for Mos-Fet.. (withiout '1117 ← has higher FW voltage drop than MOSFET) ..would be more simple than the "mess" below . . .

. . .