LM317AHVT and Heatsink

[mike_mike]

I replaced the potentiometer and now it seems to work good.
I reused the old potentiometer in another power supply, made using the same schematic and the problem did not appeared.
I checked with the scope on the ouptut for oscillations please have a look at reply #45. Are the results good ?
I measured the voltages while the voltage at the output was low (about 17.87V) and I found that the voltage on the 100R resistor was 1.25V while the voltage on the potentiometer was lower in comparation with the moment when the voltage at the output was good (19.87V).

How can I check to see what component is causing those problems ?

[floobydust]

In electronics, you don't want to fix a problem unless you really know first what went wrong. So you go after it.
My math, for 19.87V out RV2 was 1,483 ohms, and 17.87V is 1,324 ohms, a drop of 159 ohms (with R6=100R).
But the potentiomenter works fine in another build. So is it the pot? It's not making sense...
If the pot has a noisy spot, you can test it with a multimeter. Or connect it backwards (reverse rotation) so the wiper is somewhere else to get the same ohms but not on the noisy spot.

Try to get it (PSU) to misbehave, go back to the old potentiometer or use a 1.5k resistor in place of RV2, connect a scope to the output and look for oscillations to rule them out.

At these currents you will need to pay attention to wiring and single-point grounding. The end of R6 sampling output voltage should be as close as possible to the output terminals/binding posts. "regulator to minimize line drops which effectively appear in series with the reference, thereby degrading regulation."

The J2 output ground should connect to the input big filter caps' ground after J1, as well as the pot RV2. Not at the end of a long trace/wire that runs to all the little parts. All this is to get the best regulation and stability possible- you already got 10mV at 6.17A which is very good so your layout must be fine, but a pic helps. It is possible to wire a PSU like this so that it starts to oscillate under heavy load.

Those scope measurements look fine. You are basically looking at noise at 2mV/div. Is the raw DC from a mains transformer and rectifier, or is it from a SMPS?

[mike_mike]

Those scope measurements look fine. You are basically looking at noise at 2mV/div. Is the raw DC from a mains transformer and rectifier, or is it from a SMPS?

I used a 50Hz transformer, with 30V secondary and 10A current and a 35A rectifier.
I checked again, with another PCB and new components and now, the voltage does not drop. And I used the same potentiometer which I used when the voltage dropped.

[Zero999]

Going from what you've said, I suspect it was the solder joint to the potentiometer. It will probably work if you put the original potentiometer back. As I said before, it's likely it was a connection problem.

[mike_mike]

Going from what you've said, I suspect it was the solder joint to the potentiometer. It will probably work if you put the original potentiometer back. As I said before, it's likely it was a connection problem.

Before I change the PCB, I tested with the wires of the potentiometer soldered on the PCB, and the result was that the voltage still dropped from about 19.87V to about 17.87V.
In this case, does it still was a connection problem ?

[Zero999]

Going from what you've said, I suspect it was the solder joint to the potentiometer. It will probably work if you put the original potentiometer back. As I said before, it's likely it was a connection problem.

Before I change the PCB, I tested with the wires of the potentiometer soldered on the PCB, and the result was that the voltage still dropped from about 19.87V to about 17.87V.
In this case, does it still was a connection problem ?

Then it's more likely there's lose connection inside the potentiometer. Intermittent connections can be difficult to spot.

My advice is don't use that suspect potentiometer in another project. Junk it. Using parts which could be faulty will just cause frustration and time wasting later on.

[mike_mike]

It can be a lose connection inside the pot even if the pot is a multiturn pot ?
I forgot to mention that I used a multiturn pot...

How can I check for sure to see what component is making that problem ? I want to have a good power supply. I already made a few supplies in the last few years and they were oscillating... so I don't know what should I do... I am also very maniac in electronics and I want that those power supplies to be made by me.

[Zero999]

It's not uncommon for any potentiometer to develop an intermittent failt, multi-turn or not. The fact it works now suggests it was a problem with the pot. It's possible it was something else of course.

[mike_mike]

Could those voltage drop problems be a result of oscillations ?
If yes, then how can I find the oscillations ?
I already tested with the scope on the output (please have a look at the previous posts).

[mike_mike]

Hello,
I am working at a new project. The schematic is attached in this post. The project uses an LM317T and a PNP pass transistor (TIP36C). I marked on the schematic the pins for LM317 and TIP36. LM317 and TIP36 will be connected to the board using wires.
I made the layout but I am not sure if it is correct.
I read a few documents about layout guidelines and I tried to respect some rules in this layout: I divided the schematic into low current and high current zones, I added copper planes for GND, I put the unpolarised capacitors near LM358, I put the bridge rectifier and 0.1R resistor in the lower part of the layout - because of the heat generated, I tried to make the traces as short as I could.

The DRC does not return any error, but I want to know if the layout guidelines are respected ?
If the layout guidelines are not ok, could you help me to correct the problems ?

[mike_mike]

Hello,
I am working at a new project. The schematic is attached in this post. The project uses an LM317T and a PNP pass transistor (TIP36C). I marked on the schematic the pins for LM317 and TIP36. LM317 and TIP36 will be connected to the board using wires.
I made the layout but I am not sure if it is correct.
I read a few documents about layout guidelines and I tried to respect some rules in this layout: I divided the schematic into low current and high current zones, I added copper planes for GND, I put the unpolarised capacitors near LM358, I put the bridge rectifier and 0.1R resistor in the lower part of the layout - because of the heat generated, I tried to make the traces as short as I could.

The DRC does not return any error, but I want to know if the layout guidelines are respected ?
If the layout guidelines are not ok, could you help me to correct the problems ?

I am attaching updated version of the layout.

[MrAl]

Hello,

33v at 0.7 amps is more than 23 watts which is a lot.  You need a good heat sink.
But that's a lot of power too.  The efficiency is going to be quite bad.

Consider using a buck circuit which will not waste as much power.

Dont know if you need it, but sometimes a small transistor is added for current limit too.

[mike_mike]

Thank you for the replies.
I built the circuit and I found that the minimum output current set by the 50k pot is about 0.45A.
The designer of this circuit said that the minimum output current will be 0.15A.
1. Which is the correct value ?
2. How the operational amplifier commands the NPN transistor ?
3. This transistor works in the linear region ?

[MrAl]

Thank you for the replies.
I built the circuit and I found that the minimum output current set by the 50k pot is about 0.45A.
The designer of this circuit said that the minimum output current will be 0.15A.
1. Which is the correct value ?
2. How the operational amplifier commands the NPN transistor ?
3. This transistor works in the linear region ?

Hello,

I get around 0.12 to 0.14 amps minimum, however that is theoretical.  When you use real components that could be off.
However, if you are getting 0.45 amps then that could be because of several reasons.
For one, what is your load?  Remember that there will be a minimum output from the LM317 regardless of the state of the NPN transistor unless you also provide a negative voltage to the emitter so it becomes capable of reducing the output to an absolute zero.
Another reason could be because of oscillation that makes a DC measurement look higher.  You could check for that with a scope.

Another issue though is that the LM358 probably is not fast enough for a good power supply.  You could look for a faster version op amp unless you could put up with the slower speed.  Slower speed means it takes longer to react to an overload.

If the circuit is working properly the NPN works in the linear mode sometimes and saturated for larger current overloads.
The op amp output drives the NPN, so when the current gets high enough and the op amp gain is such that it puts out roughly 0.62 volts, the transistor starts to turn on and enter the linear mode.  With higher current it will turn on fully and so it will be saturated.
The slew rate of the LM358 is a little slow though so it may not be fast enough for a good power supply.

[mike_mike]

Hello,
The load that I used was a 1R resistor at the beginning of the tests. I also used as load a 40W/24V light bulb. In both cases the output current was 0.45A.

[MrAl]

Hello,
The load that I used was a 1R resistor at the beginning of the tests. I also used as load a 40W/24V light bulb. In both cases the output current was 0.45A.

Hi,

Ok well the minimum output from an LM317 without a negative power supply is about 1.25 volts, and with a 1 ohm load, that means 1.25 amps, but because the load is so low it could be competing with other resistances in the circuit so you see 0.45 amps.  It could also be something wrong with the circuit itself like wiring error or something.
Check it with other loads to make sure it works as you think it should without overload.

Also, there is usually some input offset voltage on common op amps and that would mean an error in the measured current.  You should also try changing the 0.1 ohm resistor.  Try 0.2 ohms and see what you get.  It should agree with the 0.1 ohm resistor except be about 1/2 the current.  So 0.45 should drop to 0.225 unless there is a very bad overload and then it may be 0.45 again.

[mike_mike]

Hello, I have got some trouble regarding the attached schematic. When I unplug the transformer (24V/100VA) from the wall socket (230V) then the output voltage from the power supply is rising for a short amount of time. Can this problem be simply solved or it needs a more complicated solution ?

I will also do more experiments in the next days to see more exactly what is happening.

[iMo]

A diode from output to input?

[mike_mike]

I drew a more detailed schematic. Please have a look.
I did not understood completely the idea behind the place were I should install the diode. Could you please be more specific ?

[MarkF]

Diodes D4 and D5 are there if you connect something stupid to the outputs.
They are basically protection in the case you connect something that has a voltage on it.
  - D4 is in case you connect something with a higher voltage to the output then is being output.
  - D5 is in case you connect a reverse polarity voltage to the output.

Capacitor C3 (1000uF) on the output is extremely high.  The LM317 response to transients on the load will be slowed down because of it.  I would not put anything higher than 100uF.  It may be causing your voltage rise when you unplug the AC.

Also, you typically do not see diodes (D1 and D2) across R2.  Although, they do limit the amount of bypass current.

[mike_mike]

Can someone please recommend a schematic that sink current from the power supply ?
I need it for establishing the minimum current from this power supply to about 15mA. The current sink circuit will be used as minimum load.
The problem is that I did not found any current sink circuit that works starting from 1.25V ...

[MarkF]

The current source that I can remember is the LM334.  It is 1 to 40V and 10mA max.

The minimum load for the LM317 is 10mA which the LM334 can satisfy.
You do not need to have 15mA.

[Zero999]

Can someone please recommend a schematic that sink current from the power supply ?
I need it for establishing the minimum current from this power supply to about 15mA. The current sink circuit will be used as minimum load.
The problem is that I did not found any current sink circuit that works starting from 1.25V ...

There is no need for a current sink with this circuit, because R4 draws a steady 12.5mA from the LM317AHVT's output, which is above the worst case minimum load requirement of 12mA, stated in the data sheet.
https://www.onsemi.com/pub/Collateral/LM317AHV-D.pdf

[mike_mike]

I tested the attached schematic to see if the output voltage rise when the transformer is unplugged from the mains.
The result was that the voltage does not rise when powering a load that sink a current which is below the maximum output current.
If the load sink more current than the maximum output current then when unplugging the transformer from the mains then the output voltage rise.
I made the above mentioned tests using LM317 AHVT from On Semiconductor. Will the behavior of the circuit remain the same if I will use LM317T instead of LM317AHVT ?
 

[not1xor1]

I tested the attached schematic to see if the output voltage rise when the transformer is unplugged from the mains.
The result was that the voltage does not rise when powering a load that sink a current which is below the maximum output current.
If the load sink more current than the maximum output current then when unplugging the transformer from the mains then the output voltage rise.
I made the above mentioned tests using LM317 AHVT from On Semiconductor. Will the behavior of the circuit remain the same if I will use LM317T instead of LM317AHVT ?

You might put a 1-2k 2W resistor in parallel to C2. That might solve the problem.