wierd voltage divider issue

[aruna1]

Hi
I'm working on a power supply based on a smps and a linear regulator. idea is to use buck converter to drop high voltage to a value 3 volts above the required value, and use a linear regulator to get the output. I'm using LM2596 as buck converter and LM317 as linear regulator.
I wanted to measure the output voltage using a PC16F877A and display on a LCD. so I used a voltage divider at LM317's out put to reduce output voltage ito safe value.

now comes the question. power supply is designed to give variable voltage and 21.24V is the maximum. circuit is fine and voltages are regulating correctly. I used a dummy load to test power supplies ability to with stand current requirements. so far I tried up to 1A and regulator works fine. how ever there is a problem at voltage divider. I'm using 10k trimpot as divide to set divide point voltage. so I have tuned it so that when power supply give 21.25v divide point output is 5V.
under no load condition divide points voltage is correct and varies as i change the output voltage. but when I load the circuit and begin to draw current voltage at divide point increases. actually when I load the regulator, regulator's output drops slightly. I can see this on multimeter. but voltage of dividing point increases rather than decreasing. its funny, just like ohms law got inversed. that is as voltage across trimpot drops voltage on divide point increases. (it should be decreasing right?).

got any idea what the heck is going on?
i have attached the linear regulator circuit here.

PS. I tried changing the trimot but results were same

[Soeren]

Hi,

now comes the question. [...] under no load condition divide points voltage is correct and varies as i change the output voltage. but when I load the circuit and begin to draw current voltage at divide point increases. actually when I load the regulator, regulator's output drops slightly. I can see this on multimeter. but voltage of dividing point increases rather than decreasing. its funny, just like ohms law got inversed. that is as voltage across trimpot drops voltage on divide point increases. (it should be decreasing right?).

got any idea what the heck is going on?

Given that it works OK unloaded, it's a safe bet that the load (how large?) alters the functionality of the circuit - the most likely possibilities...

Bad layout, with eg. the 10k trimmer close to the regulator and some wire resistance to where the load is, but you didn't mention how much it rise, when the voltage at the load drops how much, which makes it a bit hard to see if that's a realistic scenario.

Oscillations - 'scope the output carefully ('scope on AC setting). If that's the problem, bad layout is the most probable cause, but do test the caps, especially the output cap for capacity and ESR.
What did the 'scope show?

What does the writing on the regulator say (verbatim)?
If you want a better answer, provide better info - you are the only one knowing what exactly is in front of you (and personally, I'm much more inclined to solving electronics problems, than guessing on what you didn't care to mention or drag it out of you bit by bit).


You can remove D1 and D2, they are not needed in your circuit.
C1/C3 should be mounted as close to the input/output terminal as possible.
Connecting the free end of the 5k potentiometer to the wiper/ground will ensure a measure of control when the potentiometer ages and the wiper might lift from the carbon(?) track when adjusted. With the free end unconnected, it will spike up to a few volts below the input voltage (and that's an annoying thing if you feed eg. a microcontroller board from it).

[aruna1]

Given that it works OK unloaded, it's a safe bet that the load (how large?) alters the functionality of the circuit - the most likely possibilities...

load was 0 to 1A, it is variable dummy load

Bad layout, with eg. the 10k trimmer close to the regulator and some wire resistance to where the load is, but you didn't mention how much it rise, when the voltage at the load drops how much, which makes it a bit hard to see if that's a realistic scenario.

I use thick wires but problem begin to arises when i draw about 100 mA. I don't think wire resistance is the cause. I uses AC current carrying wire used in house wiring. voltage at load drops about 10-20 mV. voltage rise at divider is few milivolts and increasing as load increases.

What does the writing on the regulator say (verbatim)?
If you want a better answer, provide better info - you are the only one knowing what exactly is in front of you (and personally, I'm much more inclined to solving electronics problems, than guessing on what you didn't care to mention or drag it out of you bit by bit).

regulator says:
LM317T
WK00P0017
china

and has ST logo.

You can remove D1 and D2, they are not needed in your circuit.

D1, D2 is recommended in datasheet as protection against capacitor discharge through regulator

C1/C3 should be mounted as close to the input/output terminal as possible.

they are mounted close to regulator

Connecting the free end of the 5k potentiometer to the wiper/ground will ensure a measure of control when the potentiometer ages and the wiper might lift from the carbon(?) track when adjusted. With the free end unconnected, it will spike up to a few volts below the input voltage (and that's an annoying thing if you feed eg. a microcontroller board from it).

5K potentiometer is temporary just for testing purposes. I intend to use variable voltage (using a DAC) to replace potentiometer

[aruna1]

it seems problem is with the resistance of the current paths. I added parallel paths and seems problem is gone

[aruna1]

funny, problem is back, I have added parallel paths to all current carrying paths, but problem is still there

[Soeren]

Hi,

I use thick wires but problem begin to arises when i draw about 100 mA. I don't think wire resistance is the cause. I uses AC current carrying wire used in house wiring. voltage at load drops about 10-20 mV. voltage rise at divider is few milivolts and increasing as load increases.

10..20mV at 100 mA sounds a bit on the large side, if this is a measured drop from the regulator output to the load.
How long are the wires you use?

Inductance of connecting/test wires may come into play. It's very easy to use the LM317 as a switch mode controller (not a super effective one though).

You really need to break out the 'scope and check if you have oscillations and if you do, check the caps carefuly.

regulator says:
LM317T
WK00P0017
china

and has ST logo.

Good, the "T" postfix means that it can handle up to 1.5A (with adequate cooling).
Did you try replacing it?

You can remove D1 and D2, they are not needed in your circuit.

D1, D2 is recommended in datasheet as protection against capacitor discharge through regulator

Under some conditions they are, but not in your circuit...

Figure 18 shows the LM317 with the recommended protection diodes for output voltages in excess of 25 V or high capacitance values (CO > 25 µF, CAdj > 10 µF). Diode D1 prevents CO from discharging thru the IC during an input short circuit. Diode D2 protects against capacitor CAdj discharging through the IC during an output short circuit. The combination of diodes D1 and D2 prevents CAdj from discharging through the IC during an input short circuit.

Even if you want them in the finished circuit, loose them while finding the problem, just to be sure they're not implicated.

5K potentiometer is temporary just for testing purposes. I intend to use variable voltage (using a DAC) to replace potentiometer

OK. You want C_Adj to be much smaller in that case (and since it's nor needed for the circuit to function, pop it out while trouble shooting).

When it's working, make sure you get the D/A-C output stable and low impedance (through buffers), as any ripple present will be transferred directly to the output.

But... 'scope the circuit to find out what's happening first.

[aruna1]

sory for the late reply. I paused troubleshooting linear regulator part and started working on SMPS. I have some troubles with eliminating the noise. I'll post a new thread for that. After SMPS is finished I'm thinking of building linear regulator part again in a new board and see what happens