# Help for Voltage regulator LM338 with current limitation of 5A

#### romain514

Joined Dec 16, 2015
18
Hi all,
I want to do a voltage regulator with a transformer that gives me 24VAC@6A and a LM338. But since the LM338 can support 5A I would like to limit the current to 5A. I did a little circuit here and placed the resistor R4 (0.24Ohm) where i think it would limit the current. I also have a potentiometer to regulate the voltage from 1.25V to 30V.
Can anybody help me and tell me if my circuit is good and if it would function as it is?

By the way: I'm a noob in electronics so please be patient we me!

Thanks all

#### dl324

Joined Mar 30, 2015
12,871
Putting a current sense resistor there will affect voltage regulation. Better to put it up stream of the regulator and clamp the output to 1.2V for over current.

Or just let the regulator protect itself.

#### Dodgydave

Joined Jun 22, 2012
9,956
No it wont work,

Your transformer is drawn wrong, also the series resistor wont limit the current it will just drop the voltage.

#### alfacliff

Joined Dec 13, 2013
2,458
the transformer rating of 6 amps is no problem, the voltage regulator has internal regulation. and current limiting. most problems are caused by insuficient heatsink or too high input voltage causing overheating. also, the 24 volt transformer should be changed to a 24 volt center tapped transformer with the center tap grounded instead of the four diode bridge shown, the voltage to the regulator will be less and the heat generated will be less too.

Last edited:

#### romain514

Joined Dec 16, 2015
18
@ DogyDave:

I know that the drawing for the transformer is not good, as i didnt find the good part in eagle for it. But my transformer is a Triad Magnetics F-260U. Here you can see it on Digikey: https://www.digikey.ca/product-detail/en/F-260U/237-1950-ND/5032185

Can somebody tell me how should I do this circuit? Or like you guys say the LM338 as an internal current limiter? I'm not sure of that but I just want to be sure that the current will never be more than 5A.

On the LM338 Datasheet it says that it can peek to 7A so thats why im asking questions.

Thanks

#### AnalogKid

Joined Aug 1, 2013
9,260
A single 338 can be configured as a constant voltage regulator or a constant current regulator, but not both. To do what you want, have one 338 set up as a traditional adjustable voltage regulator. Then, in front of that, put a current limiter circuit. This can be another 338 configured differently, or a power transistor with some parts around it.

ak

#### ScottWang

Joined Aug 23, 2012
7,025
Choosing an AC 0-12V-20V transformer, using a switch to set two ranges of power source:
1. Vo_Lo = 12Vac*1.414 = 17 Vdc, used for output voltages 0~12V
2. Vo_Hi = 20V*1.414 = 28.3 Vdc, used for output voltages 0~24V

#### romain514

Joined Dec 16, 2015
18
will that work?

@AnalogKid
I will try to make a circuit with what you said and put it here to ask you again if its good. Thanks!

#### #12

Joined Nov 30, 2010
18,222
AnalogKid hit where I was aiming. You can't regulate voltage and current at the same time, with only one active component. They will fight. You must choose. Is current limiting so almighty important that you can't trust the chip to be accurate enough? Usually, it isn't. Besides that, the current regulator in the chip is sensitive to temperature so most people run into problems with heat. That is why ScottWang suggested a tapped transformer. Any LM338 that can have 30 volts across it can not pass 5 amps because it will overheat and shut down on internal safety.

Good start kid, but you came here for fine tuning, and it's coming at you. Probably harder than you would expect, but you showed enough quality to get good, solid answers. We've seen so many amateurs completely flummoxed that we wouldn't treat them that way. We would be tippy-toeing around some VERY basic concepts because that's as fast as they could learn. You scored about 98%. Excellent for a first try!

#### #12

Joined Nov 30, 2010
18,222
VR1 R5 R2
You forgot to connect your control loop back to the adjust pin.
The control loop is dominant. Q2 only comes into the equation when you hit maximum current.
Only then does Q2 dump the signal from the control loop.

#### AnalogKid

Joined Aug 1, 2013
9,260
Actually, I like Dave's circuit in post #6. The only advantage to having two power devices as I described is that the peak regulator power dissipation that happens when current limiting kicks in is spread out across two parts.

ak

#### romain514

Joined Dec 16, 2015
18

@AnalogKid here is my new design following what you've said
can you tell me if that works?
thanks
View attachment 96686

#### MikeML

Joined Oct 2, 2009
5,444
@romain514, have you done a heatsink calculation. Unless you lower the input voltage to the regulator by closely matching the transformer secondary voltage to the drop-out voltage of the regulator, your LM338 will be dissipating some where between 25 and 100W, which requires either a huge heatsink and/or forced air cooling with a blower/fan!

The current limiting intrinsic inside the LM338 is described on the Data Sheet: here.

#### #12

Joined Nov 30, 2010
18,222
Add another D2 across LM1 by connecting from C2 to C5 and change VR1 to 2200 ohms.
Somebody is going to have to do the transformer section for you because you look lost.

Edit: MikeML did it better in the next post.

Last edited:

#### MikeML

Joined Oct 2, 2009
5,444
The two cascaded LM338 (one for current limiting, other for voltage regulation) is a bad way to go. It more than doubles the power dissipation because each regulator must have enough head room not to drop-out, plus 1.25V drop across the current-sensing resistor.

The better way is right off the TI LM338 datasheet, which shows the 2N2222 reducing the V(adj) as a function of the current in the 0.2Ω sensing resistor. Obviously, its value would be adjusted to get 5A. Rs = 0.6/5 = 0.12Ω. Ps = 0.6*5 = 3W

This sensing resistor only drops ~0.6V, so wastes less power, and the input voltage, at the minimum of the filter capacitor ripple only has to be 24V + V(do) (Vdropout), which is about 2.7V @ 5A.

Last edited:

#### romain514

Joined Dec 16, 2015
18
@MikeML
is there anything to add on that circuit? can you verify it for me please? thank you very much. And by the way how can i will regulate the voltage here?

Anybody else has anything to add or suggest? thank you everyone!

#### dl324

Joined Mar 30, 2015
12,871
Ground needs to be between R3 and R4.

R3 needs to be a pot.

You may need a small additional load, minimum current for LM317 is 10mA and LM338 may also have one.

#### AnalogKid

Joined Aug 1, 2013
9,260
The two cascaded LM338 (one for current limiting, other for voltage regulation) is a bad way to go. It more than doubles the power dissipation.
Don't think so. For a given transformer output voltage, load voltage, and load current, the power dissipated in the entire regulator system is a constant no matter how it is divided up (pass devices, sense resistors, whatever). This assumes that there is enough headroom for two series devices, but that was pretty clear in post #1.

The better way is right off the TI LM338 datasheet, which shows the 2N2222 reducing the V(adj) as a function of the current in the 0.2Ω sensing resistor.
That's essentially the same as Dave's circuit in #6.

ak

#### #12

Joined Nov 30, 2010
18,222
And here we have 3 ways to do the transformer. Notice the different wattage rating of the two diode design. The four diode design uses the transformer more efficiently and that translates into .

There is about 1/2 volt difference in the voltage at the first capacitor. Well within safe range.

#### Attachments

• 21.7 KB Views: 27