Hi,

Am a total newbie. Trying to use a voltage regulator, LM338 to reduce voltage on a power supply from 35-37 down to 24v. I need a regulator that can handle 5A at those voltages.
I went to an online calculator on the 317 regulator and got the following figures for the standard, basic regulator circuit:

Input 35 v
Output 24 v
R3 120 ohm
R4 2.2K ohm
Vout 24.17

Questions - do these apply to the LM338 and are they correct?

Do I need a heat sink for this application with a TO-3 package of the 338?

Should I incorporate diodes in the basic circuit for protection?

Do I need capacitors and if so, what value?

What wattage resistors should I use?

TIA

 

Yes those calculations will work for the LM338. Diode protection isn't normally necessary but it wont hurt. any standard wattage resister will work they aren't passing much current so 1/4 watt will be fine. as for capacitors normally its 0.1uF on the input 1uF on the output. and as for heat sinking you wasting 11v at 5 amp. that is some SERIOUS heat i don't know how you are going to keep that cool. might be worth looking at a switch mode regulator

 

Oh btw Watts = Amps x Volts so your wasting 55 watts of power as heat. that pakage would need a heat sink after about 1.5/2 watts.

 

Thanks so much, Ctenom.

I did not give you all the information.

The actual current draw is slightly over 3 A.

There is not a continual load, only intermittent.

It is being used to operate a solenoid momentarily (maybe 2-3 seconds) about 8 times in a 2-3 minute sequence and then may be inoperative for at least 30 minutes so the regulator would have a chance to return to ambient temperature.

However, this is not a constant, it could be operative sooner or later in the sequence so I will add a heat sink for insurance.

 

No problem. if your driving a solenoid use diode protection across the regulator as they produce alot of back EMF.

 

Thanks again, Ctenom,

I have a couple of 1N4001 diodes - will they work or should I get 1N4002 or higher?

I can find ceramic for the 0.1uF capacitor and tantalum for the 1.0uF here locally - are these suitable for this application?

 

The data sheet for the LM338 recommends the 1N4002 but the only differences is the 1N4001 is rated for 50v and the 1N4002 100v. seems as your only outputting 25v id say the 1N4001 should be fine. and yes those capacitors should be fine the data sheet actual recommends a ceramic for the input. and you can go up to 25uF on the output if you want to but you must stay above 1uF

 

Oh btw Watts = Amps x Volts so your wasting 55 watts of power as heat. that pakage would need a heat sink after about 1.5/2 watts.

That design is non feasible. 55W is too much to dissipate in a single package. The thermal resistance from junction to heatsink will be about 1.5C/W plus heat sink resistance means you would need a heatsink of thermal resistance less than about 0.5C/W to keep the junction temp under about 140C. That is a massive heatsink.

The way to do this is use multiple pass transistors and spread the power.

 

I was as concerned as you at first but he has gone on to say that he is only drawing 33W of power for 2/3 seconds every couple of minutes. the data sheet says it can dissipate up to 50W of heat with a appropriate heat sink.

 

I was as concerned as you at first but he has gone on to say that he is only drawing 33W of power for 2/3 seconds every couple of minutes. the data sheet says it can dissipate up to 50W of heat with a appropriate heat sink.

Yeah, that rating is actually with an infinite heatsink. We used to laugh a lot about those specs.