current limiting for a linear power supply

[gehan_s]

hi all,

i am designing a linear power supply and i want to add a current limiting option to it. i am using a LM317 and driving its adj pin via a filtered, amplified PWM signal from a PIC to set the voltage.

for current limiting i have 2 options.

1.
i get the limiting current value to the pic (as a voltage) set by a pot (using A/D converter of the PIC). i get the current flowing in the circuit via a 1ohm sense resistor and compare the two. if the current limit has exceed i will reduce the PWM on time until the current drops to the maximum level set.

2. second one is without the PIC handling the current limiting i will compare the two currents using a comparator and if the current limit has exceeded, the comparator will switch on a transistor which will drag the adj pin of the LM317 down.

my question is which one is better ???
if both are bad can you suggest a better method???
in the second option will the system try to oscillate ????

thanks in advance !!!!!!!!!

 

[crutschow]

If you aren't concerned about a short overshoot in the current before the limit occurs, then the first method should be OK. If you want to limit the overshoot, then the second method is faster. Of course the LM317 doesn't need a current limit since it is internally protected for both over-current and over-temperature.

Yes, the second method may oscillate as the current may sawtooth around the set point.

 

[gehan_s]

crutschow, thank you very much for your reply !!!!!!!!

i have another question. without using the comparator can i use some other circuit (probably with negative feedback) that will reduce the adj pin voltage just the right amount so that current limiting can be done smoothly ???? if so can you please suggest a circuit for that.

thank you !!!!!!!

 

[crutschow]

An op amp (with a diode on the output to avoid it affecting the voltage when below the current limit) would work, but you might have to put a small capacitor from the op amp negative input to the output, to reduce any tendency to oscillate.

Note that you can only reduce the voltage to about 1.2V minimum by connecting the adj pin to ground, thus it will not limit the current for a short circuit on the output (other than the inherit internal limit of the LM317), but I don't see that as a particular problem.