PSU design help

[robimarko]

Hi,I am designing and making my own power supply as my first real electronics project because I am just a beginner with just over 6 months of experience in electronics and 2 years of studying electronics engineering in school.
I am using LM338 as core of the circuit and of course 24V transformer with 4.16 amps output.
In multisim this circuit works like a beauty and outputs voltages between 1.25V and 30V.
Now i want to add current limiting part of the circuit cause it is very good feature to have.
Does anybody have some ideas on that part?

[Sam__]

Look up the standard lm317 current limiting circuit. Shout be a good place to start.

[robimarko]

Not a lot of helpfull info when i google that

[Asim]

You can find info about voltage regulators as current limiters all over the place,check the lm317 datasheet as an expample.

Here is a schematic of the lm117 as a current limiter http://www.ae5d.com/images/Lim.png.

You need to put on mind that R1 will dissipate a lot of power, so choose a high wattage resistor ( P=I^2 * R)

[mariush]

First of all, do the math and figure out how much power is wasted as heat in the regulator at various output voltages and currents. 
If you think you're going to do 3v @ 4-5A with  30-40v input, you're dreaming.

See my post here which goes through some calculations to determine input capacitor and other tips: https://www.eevblog.com/forum/beginners/beginer-new-power-supply/msg474863/#msg474863

The basic current limiting is done with just a resistor. That's more or less designed as a fixed current limit, not adjustable current limit. 

For a proper adjustable current limit, you would use a current shunt (0.1ohm - 1ohm is common) and measure the voltage drop on the current shunt.
For a 1ohm resistor, 1 volt drop on resistor equals 1A going into a device you have connected to your power supply. You can this voltage drop with a circuit that will dinamically adjust the output voltage down (or turn off the regulator completely) when the current limit is reached, to protect the device connected to the power supply.
Opamps are typically used for this.

See for example this application note: http://www.micrel.com/_PDF/App-Notes/an-19.pdf

or see the schematic of a power supply with adjustable current limit attached below  (you're interested in the bottom right section , you see there R57 current shunt of 0.1 ohm and the IS point going into IC1A , a LM358 opamp.

[tautech]

OP your circuit is exactly the PSU I have used for years, but your transformer is underrated for the LM338.
The 338 can supply 5 amps or more.
I would suggest you use LM317 to gain the benefit of the device's thermal and short circuit protection.

[robimarko]

Yeah I pretty much used various parts of already done psu-s.LM317 is too weak for me because it can output just 1.5 A but I think it can be increased with some power transistors.Will have to look into it

[rdl]

The LM338 must still dissipate power the same as an LM317. The size of the heat sink required to get 4 amps out of an LM338 will likely be more of a problem than the transformer only being able to supply 4 amps.

[cellularmitosis]

Hi robimarko!

This is a great first project.  My first project was the same as yours: a dual LM317 supply (one for current limiting, another for voltage limiting).

Dave covers the "classic" version of this circuit briefly in his power supply design video series, before moving on to more sophisticated designs:





I designed something very similar and had some PCB's made.  I'm on revision 2 of the board now, and am probably going to put it up on tindie.com for sale soon.

https://github.com/pepaslabs/DualLM317BenchSupply

The series resistor in the CC portion of the circuit has to burn a lot of power, so you can't use a potentiometer; you'd have to use a large rheostat to have truly variable CC.  Instead, I decided to keep it simple by instead offering 6 fixed-current limits.  I figured the most rudimentary of beginners (whom this supply is aimed at) most likely just need protection against letting the magic smoke out, rather than fine current adjustment (I am also working on an op amp / pass transistor supply which will have fine current limiting).

Schematic:

https://github.com/pepaslabs/DualLM317BenchSupply/blob/master/DualLM317BenchSupply_schematic.pdf?raw=true

The PCB:



Assembled board:





The board next to its copper-clad prototype:



If you know how to use KiCAD, I've open-sourced all of the CAD files.  Please feel free to use them!

The total BOM costs end up being about $30 to assemble a board, and that assumes you can find a wall-wart on clearance for $3, so the fully assembled version of this board is a bit expensive (compare to an entry level bench supply, the CSI1802SX for $50 http://www.circuitspecialists.com/bench-power-supply-csi1802x.html ).

For that reason, I think I'm going to just sell the PCB's, rather than a full kit, as that would add needless markup and shipping to a bunch of components which people can simply order themselves, otherwise I'd have to charge the cost of a CSI1802X, which would be ridiculous.

[robimarko]

Thanks for really good reply.
How much will boards cost?

[cellularmitosis]

At seeedstudio.com, a batch of 5 boards sized 10cm x 10cm, 2 layer, with lead-free HASL pads is $35.51 shipped.  That's $7.10 per board.  If I order more than 5 boards, the cost goes down.

I'd probably start the boards at $10, and if the first 5 sell, I'll order the next highest quantity for the next batch, and so on.  Thus, if it becomes popular, eventually I will make some profit.

[void_error]

Hi,I am designing and making my own power supply as my first real electronics project because I am just a beginner with just over 6 months of experience in electronics and 2 years of studying electronics engineering in school.
I am using LM338 as core of the circuit and of course 24V transformer with 4.16 amps output.
In multisim this circuit works like a beauty and outputs voltages between 1.25V and 30V.
Now i want to add current limiting part of the circuit cause it is very good feature to have.
Does anybody have some ideas on that part?

You could take a look at my PSU design thread (https://www.eevblog.com/forum/projects/bench-power-supply-design/15/), schematics are on page 2. It can be simplified to suit your design (It can easily be changed to not use the digital part). Haven't posted the updated schematics yet but will do in a few hours probably.

Also, it's very unlikely that you'd be able to output 30V regulated at 4A because of the losses in the transformer causing the secondary voltage to drop.

I have a similar transformer rated at 24V 4.17A and I did some measurements. It's internal resistance turned out to be about 0.5 ohms.

That means that for 4A into a resistive load the secondary voltage will drop by 0.5ohm x 4A = 2V but the load is a rectifier with a capacitor so peak currents will be higher. I'd suggest you go for a 20V maximum output voltage to ensure regulation, otherwise you'll have to use a 36V secondary transformer. Aso, for a 4A output current you'll have to calculate the maximum power dissipation on your series pass element (output voltage set to minimmum). I suspect the LM338 won't handle it.

[robimarko]

Thanks for reply it is really and interesting design.
I am waiting for latest schematics, I really like PIC controll

[mij59]

Hi,

Please check the data sheet of the LM338, the output current is limited by the voltage across the regulator. 

[void_error]

Here's an idea to start from. I'll leave the details and choice of components to you.

VCC is the input from the rectifier + cap.

The INA138 can be replaced with an opamp configured as a differential amplifier and the shunt resistor (R2 || R3) recalculated for an output of 1V/1A.

Q1 should be capable of dissipating at least 100W, you could use the classic 2N3055 and a big heatsink or it can be replaced with 2 or 3 smaller transistors with series emitter resistors, paralleled.

If you want overcurrent indication you could use a comparator (or an opamp) with a LED and resistor on the output, with the inputs tied to the outputs of U2.1 and U2.2 respectively.

Sorry about the quick and dirty reply but I just got my PSU digital control working  (at least most of it) and I'm onto optimizing the 700+ lines of C code I have so far.

Hi,

Please check the data sheet of the LM338, the output current is limited by the voltage across the regulator. 

That's what I think is called the SOA limiting circuit, it will limit the output current for large input-output voltages. Noticed it when goofing around with LM317s. The output current available will be lower.

[robimarko]

Thanks again,i really like your digital control with PIC MCU.
I am going on a vacation in few days so after vacation I will finally start some real planning and designing