I made a simple voltage regulator with lm317. It consists of 12V 1A transformer, bridge, filter cap, and two lm317's in parallel, one for 5V/3.3V, the other for 1.25-13.75V.

It worked fine for a bit, and then suddenly the max output voltage of the second one was not 13.75 anymore, but 11.5V. Also, on transformer output the voltage was not 13V without load anymore, but about 10.5V.

Where should I search for the problem?

 

And also, voltage on 3.3V/5V output was a bit higher than those without the load in the beginning, and now it's exactly 3.3V and 5V.

 

If the voltage from the transformer has gone down then its faulty.

 

I disconnected everything and the voltage drop is still present, so I guess it's transformer...

But I don't get it, how could have that happen? If the wire is burned, shorted or something, shouldn't it be broken completely?

I mean, it still works very fine with new voltage, voltage is regulated and stuff.

 

Please post a drawing of your circuit. Include resistor values, filter cap values and load.

 

I disconnected everything and the voltage drop is still present, so I guess it's transformer...

But I don't get it, how could have that happen? If the wire is burned, shorted or something, shouldn't it be broken completely?

I mean, it still works very fine with new voltage, voltage is regulated and stuff.

When a transformer is said to be shorted, then the short could be to the casing, or there is a path around some of the coils. Since there is a voltage drop in the output, I would say there is a short in the secondary coils. Fewer coils means lower voltage.
The lower voltage regulators will work fine, but the maximum voltage of the higher regulator has been lowered. Also, the transformer may start to heat because of the short. This could lead to more coils being shorted and the transformer becoming useless or may even catch fire. Replace it to save you some headaches later.

 

Is it proper to use two different output regulators in parallel like that ??

On the other side of that question... I have my 0 - 12 volt variable " battery replacer " supply set up with 4 LM317's in parallel on a substantial heat sink, to coax more amperage out of the unit... ignoring the idea that due to discrete differences in device parameters.
Theory states, one will be forced to work harder than the rest... I just let Ohms' Law deal with the difference... Current splits evenly across parallel devices...

It will put out 6 amps intermittantly for up to 3 minutes at a time to repeatedly cycle a photo-flash unit, hardly breaking a sweat, and has done its job flawlessly for 20+ years...
For what it's worth, when charging from zero, larger consumer flash units can draw in the neighborhood of 10 amps initially, which tapers off as the main capacitor attains full charge.

 

Here is the schematic:

http://i210.photobucket.com/albums/bb204/n00blord1/09062013083_zpsbe9c944b.jpg

And how it looks:

http://i210.photobucket.com/albums/bb204/n00blord1/09062013081_zpsbb7e363f.jpg

(beauty comes from outside in this case)

But I guess it doesn't matter cause if it's the transformer...

Will it continue to work if it's shorted? it's never hot, and never was...

 

Tell me if I am wrong but, the caps on the output side appear to be 2200 uF and 100uF on the input side. If true, the caps should be swapped. 2200 on the input and 100 on the output side.

Also, the control resistor from output to adjust appears to be 400 ohms (possibly 100 ohms), it is too dark to read. If it is 400 ohms, you need something smaller because the lm317 needs a minimum current flow to sustain regulated voltage (10 mA if I remember correctly). That gives 1.2 volts/120 ohms = 10 mA.

 

Couple of suggestions. In your schematic you are using a 100Ω resistor for R1 on the upper wide voltage range regulator. Thats a bit too low IMO for your input voltage... I would suggest moving it up to a 120Ω or higher resistor. Next I would solder the wire going to the trim pot directly onto the adj pin of the of the regulator and run the other wire directly to the ground terminal of your primary filter capacitor to avoid ground loops and help increase transient response. Same goes for the resistors on the fixed voltage regulator. If you really want to get anal, you can take a SMD resistor and solder it directly on the output and adj pins near the PCB.

Also the regulator that is on the heatsink touching the bottom of the case, that is poor thermal design. Linear regulators get HOT depending on how much current they are outputting and the voltage they are having to drop... ask too much of the regulator without adequate heatsinking and it will go into thermal protection mode. Heat rises, and in your design you have insulated almost half of your heatsink by mounting it in that position. Its easy to tell when a regulator goes into thermal protection mode... your output voltage will suddenly drop like a rock.

Speaking of dropped voltages, all regulators have a dropout voltage. LM317s have a guaranteed dropout voltage of 2.5V IIRC. In other words your power supply needs to be 2.5V greater than the target output voltage. And since your using a linear transformer the harder you load them the more their output voltage drops. Taking a wild guess as to how much power that transformer could supply, I'm going to say 500mA rated tops. And from my experience with similar wall warts, if they are rated at 500mA @12V it should be putting out about ~15V under no load.

In addition, I would suggest relocating power status LED to the output of the 3.3v regulator instead of the unregulated supply rail. The upside to this is that you can use a lower Ω current limiting resistor while simultaneously providing a small load for the regulator. All linear regulators should have a small minimum load to maintain stability IMO and an LED is nearly perfect for that application.

There are some other things I noticed missing that you should have. While regulators will work without them, all linear regulators should have a .1uF ceramic bypass cap placed as close to the regulators input pins as possible for maximum stability. And where are your fuses? Call me anal retentive but it appears as if the transformer is wired directly to the plug connection. There should be a fuse there if for nothing more than safety reasons. The power switch should also be connected on the high voltage side as well so that when the switch is off the power supply is completely de-energized. You also might want to consider adding some protection diodes to the regulators so that when you shut off the power you don't have to worry about any capacitive loads on the output backfeeding into the regulator.

Lastly, I would think about swapping capacitors. If you happen to have one laying around, I would put in a 2200uF cap for your primary. I'm taking it by the size of them your using 22uF caps on the output. Nothing wrong with that, in fact that is pretty much what the datasheet calls for. But from what I hear if you put a solid 10uf low impedance tantalum on the adj pin (lower the ESR the better from my understanding) it can make the output much smoother ripple wise. However the ESR on the output filtering caps must be within a certain range or it can cause the regulator to oscillate... but that is a whole other thread.

Hope this helps.

 

Is it possible that your bridge rectifier has failed in a way that you only have 1/2 wave rectification? Can you put a scope on the filtered dc on the input capacitor. Also on the output of the higher voltage regulator to see if it is dropping out each cycle.

 

Also the regulator that is on the heatsink touching the bottom of the case, that is poor thermal design. Linear regulators get HOT depending on how much current they are outputting and the voltage they are having to drop... ask too much of the regulator without adequate heatsinking and it will go into thermal protection mode. Heat rises, and in your design you have insulated almost half of your heatsink by mounting it in that position. Its easy to tell when a regulator goes into thermal protection mode... your output voltage will suddenly drop like a rock.

This PSU is intended for my summer holidays, I need small portable PSU so I could learn microprocessors (my modified ATX is too big to carry around and I have no time to design small SMPS and any type of PCB). So all I need to power are leds, 7-segment displays and LCD displays. How much could those heat up the LM317? My guess is not much...

And from my experience with similar wall warts, if they are rated at 500mA @12V it should be putting out about ~15V under no load.

This one is rated @1A. With no load, after rectifying and filtering i get 17VDC.

Switch is on the primary side.

I guess I made a poor drawing of my circuit, hope this one's better:

http://i210.photobucket.com/albums/bb204/n00blord1/ajdsag_zpsd52bba39.png

On stripboard, the caps are very close to the inputs/outputs.

 

Get a proper heat sink and place it on the outside of your power supply. That way you will have something that will work during your whole summer The way you have done it now is not very efficient.

 

What you also can look at is some old cell-phone charger. Even if they are quite small many can output around 5 volt and 500-1000mA. They should be labeled with both output voltage and current. Put a 1000uF and 100nF cap in parallel and you are ready to go