Problem with voltage regulator LM338

Hello,

I have problem with a voltage regulator LM338. This is first time when im using a regulated voltage stabiliser. I want to get 6V with up to 5A, im using a 15V 5A voltage supply (some laptop AC/DC adapter).

The problem im observing is very large voltage drop under load . Example, the voltage without load is 6.2V and after applying load of ~1.2A it drops as low as 5.2V, basing on that i can forsee that with greater load the voltage would drop drasticaly lower which would mean i would not get suffiscent voltage for maximum load and if i would want to to have enough voltage for maximum load it would be too high for minimum load.

Is it normal? The circuit im using is from datasheet:

D1, D2 1N4007, R2 is 1k Ohm, C2 as in schematic, C1 is 100uF. LM338 has quite big radiator three times bigger than the IC itself.

Is there is something with schematic, maybe the power source, or the LM338 itself is flawed (bad part or its old IC)? Do you suggest other voltage regulators 6V with up to 5A?

What is the actual input voltage when this problem happens ?

15.18v without load, drops to 15.12v under that 1 .2A load. Also the voltage regulator circuit is assembled on breadboard, dont know if its relevant.

How hot is the LM338at 1.2A? Can you hold it in your hand hot?
How far away is the source of the 15V? What input capacitor?

The temperature of LM338 rises to about 40 degree celcius, there is about 1.5 half meter cable from 220v socket to AC-DC 15V 5A adapter, then there is another 1.5m cable with 15V to the LM338, and then the load is just centimeters away from it like 10 max.

LM338 has quite big radiator three times bigger than the IC itself.

Click to expand...

My guess is your heatsink is nowhere near big enough and the IC is self-protecting itself because it is cooking. At 5A the IC would have to dissipate (15-6)*5 = 45W ! A bypass transistor could be used to pass some of the 5A, relieving the reg of some of the burden.
I very much doubt your breadboard is designed for 5A currents.

This comes from the data sheet. I think you need a capacitor across the 15V close to the regulator. 1.5m is a long distance. The data sheet might be confusing so 'any cap is better than no cap'.
40C is fine.
You might have a grounding problem (ground loop). If the input cap makes no difference then we should look at your grounds. Can you send a picture?
_________________________________________________________________________
EXTERNAL CAPACITORS
An input bypass capacitor is recommended. A 0.1 μF disc or 1 μF solid tantalum on the input is suitable input
bypassing for almost all applications. The device is more sensitive to the absence of input bypassiing when
adjustment or output capacitors are used but the above values will eliminate the possiblity of problems.
The adjustment terminal can be bypassed to ground on the LM138 to improve ripple rejection. This bypass
capacitor prevents ripple from being amplified as the output voltage is increased. With a 10 μF bypass capacitor
75 dB ripple rejection is obtainable at any output level. Increases over 20 μF do not appreciably improve the
ripple rejection at frequencies above 120 Hz. If the bypass capacitor is used, it is sometimes necessary to
include protection diodes to prevent the capacitor from discharging through internal low current paths and
damaging the device.
In general, the best type of capacitors to use are solid tantalum. Solid tantalum capacitors have low impedance
even at high frequencies. Depending upon capacitor construction, it takes about 25 μF in aluminum electrolytic to
equal 1 μF solid tantalum at high frequencies. Ceramic capacitors are also good at high frequencies; but some
types have a large decrease in capacitance at frequencies around 0.5 MHz. For this reason, 0.01 μF disc may
seem to work better than a 0.1 μF disc as a bypass.
Although the LM138 is stable with no output capacitors, like any feedback circuit, certain values of external
capacitance can cause excessive ringing. This occurs with values between 500 pF and 5000 pF. A 1 μF solid
tantalum (or 25 μF aluminum electrolytic) on the output swamps this effect and insures stability.

Thats the radiator im using 3.5 x 2.5 x 1cm dimensions, silicone dystanser and then LM338 screwed to it. At 1.2A it would be like 11 W .

I have one capacitor between regulated + and - 100uF, here comes the pics:

The black white cable is 15V from AC-DC adapter, the brown/red is connection to load. The yellow one goes to ADJ, red is OUT and brown is IN. The cap in top right corner between GND and 6V from regulator is 100uF 25V electrolityc

As above - stick another 100uf and 100nf on the 15v input to ground and see if that helps at all.

I drew a green line where "ground" flows. These proto boards have too much resistance. Please to change your wiring so ground is a small area not flowing in this big loop.
EDITED: Measure the voltage drop from ground going into the board and ground leaving the board. The ground to ground voltage will be some voltage. (0.5V)

Just placing capacitors didnt help but i rearranged parts placement to prevent ground loop, i made it so that ground is only in upper left quadrant of board (and yes, there was voltage difference between ground in and out, it was something 0.2 - 0.3 V)


Now voltage still drops but only half of what it was before, from 6v to 5.7v @ 1.2A, for now cant test it under bigger load. But it seems that the board is at fault, i guess that making some real testboard PCB awaits me

The datasheet shows that poor wiring (which is common on a breadboard) causes poor voltage regulation.
Get rid of the breadboard, reduce the input voltage and use a much larger heatsink.

Thank you everyone for input , i shall come back soon with a PCB design

btw. i really enjoy how forum have changed, especially drag'n'drop images to upload

I agree that most of your problem is from using a breadboard. The little spring contacts inside the plastic block are wayyyy too small to handle any kind of power currents. The internal resistance is too high.

I don't trust them with anything more than a few tens of mAmps.

Also be aware that once you have pushed a big pin into a hole, its ability to make good contact with a small pin on your next project is degraded.

Thanks for help, as i have previously announced im back with PCB design. First, updated schematic (added input capacitors), then PCB design,


The yellow rectangle is board outline, the cyan thick rectangle is bigger radiator i plan to use 70mm x 25mm x 18mm height (image of it attached above), is it big enough to help dissipate power (45W max)?

Also, do you have any suggestions? The PCB is not final, i want to add 2-3 more Vout/Gnd pairs so i can attach to it directly some devices, also maybe a 5A fuse on Vout line just after the last C1 capacitor ( i have 5A 35V poly fuse). Traces are 1mm thick, enough or should be made bigger to reduce voltage drops?

I also have idea to flip the LM338 chip to the other side of board so one side of the board will have traces and all parts while on the other not used side there will be only LM338 and its radiator (so the board will close in the shape of ~ 25x 70mm rectangle).

Copper is free! Use it.
I used paint to exit you picture.
>increase the copper on the high current traces. Vin and Vout.
>I moved R2 and C2 up to increase the size of the copper in your ground. The connection between gnd(in) and gnd(out) should be real good. This is why your bread board had problems. I would move D1,2, C1 up to double the copper in the ground. Maybe face D1 to the right and at the top edge of the board so R2 can move up more.

Updated PCB,

Added 3 extra GND/VCC pairs i was talking about, adjusted elements position to maximise the minimum width of area between GND in and GND out.
Also about the fuse i think its not needed, the LM338 is temperature and current limited, am I right?

Updated PCB again, mostly because the footprint of diode had cathode and anode assigned to wrong pins (in schematic it was connected right but on PCB design it was showing up in wrong direction).

R1 to C2 is not connected yet. Good wide copper.

ahhh, dont know how i overlooked that, updated PCB in post above, now im fighting to convert my PCB image to PDF -.- (program im using is allowing export only to gerber format, and the current program i used to convert gerber -> PDF (gerbV 2.6) sometimes is not working as it should (missing layers/parts after export, some of the images instead of being smooth vectorised are pixelated in zoomb), do you know any program that can help me getting nice PDF from gerber file? (i was also using GerbMagic but it was hard to use and out of nowhere it started telling me that it has expired and tells me to buy it ._.)

ps. im attaching multisim and ultiboard projects of this circuit as well as exported gerbers for future use