Temp. Compensation for linear power supply

[smartsarath2003]

Hello,
I built a small power supply with LM338 regulator. The problem I am facing is the output voltage going down slowly when i leave the load ON. Is this happening because of temp. I connected a load of 220R at 24V output, so it not too much current actually. Does anyone have anyidea why is this happening and how can we prevent this?  :
I checked the LM338 datasheet and its given that the Vref , Iadj goes up by temp. From this can I conclude that the variation in Iadj is more than variation in Vref and thats why the regulator is pulling its output down to maintain 1.25V. Looks li bit confusing ???
Any Help Please.........................
thanks

 

[audioguru2]

Hi Smart,
Your resistor from the adjust pin to ground is probably getting hot (nearly 1/4W) and decreasing in value which reduces the output voltage. Use a bigger resistor that won't heat as much and will be more temp-stable. If it is a pot then replace it with a stable cermet pot.  ;D

 

[MP1]

Smartsarath,
I have used this chip. There should not be much current across this resistor. 
You ask about temperature problems. Are your components getting hot? This is where you want to start. There are also comments in the data sheet regarding output impedance. What is your load impedance?

You did not post your schematic so it is hard to say what might be wrong. There are several examples in the data sheet. Which one did you build?

MP

 

[audioguru2]

I have used this chip. There should not be much current across this resistor.

The datasheet shows an LM338 circuit for up to a 25V output. With its 5K pot adjusted about halfway for a 25V output, it dissipates 238mW which is very close to its max if it has a 1/2W rating to be dissipated along its entire resistive element. It is also very close to the max rating of a 1/4W resistor.
Carbon pots and resistors change value at the limit of their ratings. View attachment 36784

 

[MP1]

No heat coming off the resistor or pot on mine. Again, I would like to see smartsarath's schematic. If I cannot see the schematic, my first guess is going to be ripple.

MP

 

[audioguru2]

Maybe you used an LM138 and its 240 ohm resistor (half the current).

 

[smartsarath2003]

Hai,
I used a 10uf cap between ADj and GND. I still have the last digt of my meter gowing down, but is const after few seconds. its actually going down after conncting load then going up a bit and then slowly going down. The first bit is some thing to do with regulation, But didn't understand the second and third bits.
Also I loaded it with 1A const. current load. And exactly same thing that I discribed above is happening again. In addition the regulator enters thermal shutdown ??? and the voltage drops to 0V and will be back to 24V only after power off and power on back.
Any solutions to tackle this problem...
I used approximately 10x5x0.2cm heat sink
Thanks

 

[audioguru2]

Hi Smart,
Thermal shutdown!
There's another problem and voltage stability is one symptom of it.
Maybe your input voltage is too high. How much?
Maybe your heatsink isn't completely flat where the IC mounts, no thermal grease was used or the heatsink doesn't have large fins. What kind of insulator did you use between the IC (which package?) and the heatsink?
Thermal shutdown certainly indicates a heatsink problem.

 

[Shahriar]

Dear smartsarath2003

Put an ameter in series with your load to check how much current U draw from your regulator.
putting a 220Ohm load in 24V draws just 110mA and in this case know heat sink is needed.
I am sure that you have connect sth. wrong. specialy the pins of lm338

HTH - Shahriar

 

[Kevin Weddle]

If yours is like mine, you probably have a regulator that is not performing well. Mine did that just before the voltage went completely down after few minutes. I have since replaced it and it works fine.