LM338 Voltage and Power Questions

[Electromotive]

Hi:
I am building two LM338 variable regulated voltage supplies for my O-Scale 2-Rail Train Layout. Existing transformers Pull 10-13Volts (full on) at an estimated 1.5 Amps. Due to the circuitry, the existing power supplies cut off at 1.0 Amps, so they run but are a little anemic, more so with a double header (2 engines ).

I chose LM338 (From Schematics off this web-site due to the higher 5 Amp capacity fed by 24V (4A) transformers.

My calculations show I will end up with about a 23Volt maximum potential, which is way above the normal running voltage of between 7-13 volts. I am using Motorola LM338 in the TO-220 package due to the lack of availability of the TO-3 package. I have a relatively large heat sink (approx 5"x4") and could run a fan off the 12v trsfr ctr tap.

My Questions are:
1 - I am assuming the heat buildup will occur at lower train voltages and reduce at the higher running voltage. Is this correct?

2 - Assuming I can add a fan to the heat sink, will the lower running voltage cause me issues with excessive heat? (Calc. 23volts - 7volts = 16volts).

3 - How much can I reduce this top-end voltage by adjusting R1 and R2? It doesn't appear that there is a lot of adjusting that can be done with the 120/240 Ohms as suggested in the literature.

You all have been very helpful in the past and I would appreciate any comments on this issue.

Thank you... Tom

 

[Hunter64]

Don't waste your time with linear regulators and their power dissipation. It's quite easy to use a switching regulator.

 

[Electromotive]

Don't waste your time with linear regulators and their power dissipation. It's quite easy to use a switching regulator.

Interesting, but I'm afraid this doesn't help me much, as I have purchasedAll the components for two power supplies including the 24v transformers from MPJ and have the voltage and amp meters and cases.

 

[Electromotive]

Would it make any sence to mount a TO-220 regulator directly to a heat sink but instead of the mica insulator, insulate the entire heat sink for improved heat conductivity?

 

[bertus]

Hello,

Keep in mind that the tab is connected to the output voltage when mounting it directly to the heatsink:



Bertus

 

[Electromotive]

Right, but much more conductivity. Any further comments on if a relatively large heat sink can dissipate 16 to maybe 25 watts?

 

[dave9]

^ Yes of course, a typical CPU heatsink could easily handle that with a fan, but it's quite lossy.

What I would do is (use switchmode PSU as already suggested but keeping closer to your agenda...) try the circuit using the transformer center tap for 12V. Remember that is AC, so once rectified to DC and smoothed, it's going to float up closer to 16V, then once you subtract the dropout voltage of the linear regulator (per the datasheet, a bit under 2.5V @ 4A, and less at lower current), you'll going to get fairly near 13V. If you want an extra ~half a volt on top of that, use schottky diodes instead of silicon for the bridge rectifier.

I assume this is the transformer you're working with?

https://www.mpja.com/24V-4A-Center-Tapped-12-0-12-Transformer/productinfo/27845 TR/

 

[Electromotive]

Right on. I had thought about this but will lose half of my 4 amp potential. However, I don't need the 5amp capability of the 338. I also don't realy need the 4 amps from the transformer either. My existing power supply needs a little more than 1 amp, so this may very well reduce my potential heat issue.

I purchased 10 lm338's in the to-220 package to avoid the chinese knock-off issues. The to-220's are made be Texas Instruments. Do you feel these will handle the heat dissipation, which is now (thanks to your suggestion) is 16v-2.5 or 13.5v - 7volts normal running or 6.5vx1.5 amps for an approx. 9.75 watts?

I have not investigated the lm138 to see if the to-3 package availability is better than the obsolete lm338.

I can add a fan to the heat sink if required.

Thank you so much for this assistance. I will keep the forum updated, as that may help others in a similar manner.

 

[dave9]

There is no current reduction doing that, AFAIK. You'd still have 4A @12VAC. However, I overlooked something. Since you bought from MPJA, I assume you're in the US, and therefore using 110VAC mains?

If so, take a look at the wiring diagram on MPJA (included below). You might be able to wire it as if you had 220VAC, then the output would be 12V (and 6V at the center tap).

This is where someone else more knowledgeable about transformers could help, as I don't know what effect using the 220VAC wiring (primaries in series) would have on current. Would it result in 1/4th as much so 1A?

If it doesn't quarter the current then it also seems useful to power a 12V fan from 6V (closer to 7VDC) for less wear and noise, if the fan is large enough for that. If as you stated you have a 4"x5" heatsink (assuming it has fins), an 80mm x25mm typical computer fan, spinning at all, should provide enough airflow for ~10W.

It could possibly stay within thermal limits without a fan at all if the heatsink is a good design... you only mentioned 4"x5" and not the height. 4"x5" piece of plate aluminum isn't (usually) enough for 10W but 4"x5"x2" tall with fins, might be. I'd still plan for a fan then go from there deciding if it's necessary unless you have the C/W spec for the heatsink.

 

[Bluejets]

Trying to decipher just what you are running from this supply.

Is it just 2 locos or is there other aux equipment driven from the same output?

Possible for you to provide a sketch of the circuit involved including any existing equipment that will be reused.

When you say "variable supply" I suspect this drives the locos. If so, there are much better ways for example pwm.

 

[Bluejets]

I don't know what effect using the 220VAC wiring (primaries in series) would have on current.

Transformers are fairly robust devices and series running of the primaries will do no harm.
For 20 odd years I have used a 40VA on a dimmer (normally considered a no-no) for foam cutting with a range of tools which require both end of the available power output.

Output reflects the input (along with any core losses) and visa versa regardless. Drawing 1A @ 12v secondary will result in a 12VA primary power draw.

 

[dave9]

^ I'm not suggesting it would do harm but rather, for 110VAC it runs the two primaries in parallel to get 4A output. If using 110VAC with the two primaries in series, shouldn't that result in 1/4, (or 1/8th?) the current? I don't know. This wouldn't be a hard thing to measure and change.

Drawing 1A @ 12v secondary will result in a 12VA primary power draw.

I'm skeptical about this, won't it be in the range of 60% efficent so power in > out? Plus I don't know how changing the primaries from parallel to series would change efficiency. I suspect it'd go up but could be wrong.

 

[Electromotive]

Instead of lowering the transformer output (starting to get a bit confusing), what if we go back to basics from the standard TI lm138/LM338 literature and work with the original 24volt output at 4 amps.

From: Vout=1.25(1+R2/R1) + Iadj×R2
Where: R1=240 R2=2500potentiometer.

Most suggested circuits list 5000 for the pot and even higher. Using 5000 gives voltage maximums north of 27 - 30 volts (therfore our heat generation problems.

What if we make the pot. 2500 ohms?

This gives us 14.2volts +Iadj × R2..... I have no idea what Iadj for a value is?

Yes, my heat sink has 2" fins.
I can run a small fan off 12v ctr tap

Anything incorrect with this logic.

Thank you all.... Tom

 

[dave9]

It's not confusing really, if you want it simple then all you have to do is use the center tap to get 12V if you don't want to consider the other wiring scheme I mentioned. That still seems a better option to not waste over half the power as heat, compared to taking the full 24VAC into the regulator.

Different potentiometer resistance values don't affect this (beyond a trivial level), as it's just a voltage divider that sets the output while the fact that it's a linear regulator means it dumps all excess voltage as heat, so the ideal input voltage is just 2.5V above your peak output voltage target.

 

[Electromotive]

Thank you Dave9:

I will continue to build and test per this 12VAC recommendation. Will advise of the results. Thank you for continuing on this.

 

[Bluejets]

^
I'm skeptical about this, won't it be in the range of 60% efficent so power in > out? Plus I don't know how changing the primaries from parallel to series would change efficiency. I suspect it'd go up but could be wrong.

Transformers are 1:1 devices as far as VA is concerned.
If you want to get technical ( really unnecessary) it's power out is equal to power in , less any losses.
Transformer losses are in this instance, negligible and of zero concern.

 

[Bluejets]

^ I'm not suggesting it would do harm but rather, for 110VAC it runs the two primaries in parallel to get 4A output. If using 110VAC with the two primaries in series, shouldn't that result in 1/4, (or 1/8th?) the current? I don't know. This wouldn't be a hard thing to measure and change.

With reference to the supplied diagrams and assuming they are the correct type.

If one looks at it in this manner, ......with one primary winding only connected to 110v, output would be 1/2.
So 2 primaries connected in series would be a further 1/2, so to answer your query..... 1/4.

But it all irrelevant really.
All the op has to do is connect one winding as you suggested and he should be fine.

 

[Electromotive]

With reference to the supplied diagrams and assuming they are the correct type.

If one looks at it in this manner, ......with one primary winding only connected to 110v, output would be 1/2.
So 2 primaries connected in series would be a further 1/2, so to answer your query..... 1/4.

But it all irrelevant really.
All the op has to do is connect one winding as you suggested and he should be fine.

Yes, the 110volt schematic of the MPJ transformer is correct. If power in equals power out, can I assume that the 4 amps on the 24volt secondary will remain as 4 amps on the output if I use the 12volt center tap?

 

[Electromotive]

Gentlemen:

Finally, completed my two Model Railroad Regulated Power Supplies. I owe a great deal of thanks to all those on this forum that assisted me in completing them. As you might recall, my 1 Amp power supplies always seamed a little anemic. I needed to go to the LM338 for up to 5A of output.

Please find the below pictures and final schematic I used:





Here are the list of problems this forum assisted me with:
1 - Chinese LM338 had issues. Finally Settled on US TI TO-220 Package. Bought 10 so I would have spares.
2 - Discovered working (multiple) schematics and eventually massaged them for my particular application.
3 - Changed from 24V 4Amp Transformer to using the center tap at 12V at 4 Amps.

Here are a list of self inflicted problems during construction:
1 - Wired Pot. backward and had to reverse end wiring.
2 - One worked well, the second had a faulty or damaged LM-338. Changed it out.
3 - NEVER LET ANY SMOKE OUT !!!!!

Here is a list of the application results.
1 - Engines previously running poorly now run well and instead of their current draw of 1 Amp max., the current has increased to
1.5 Amps.
2 - Double header engines (two) run well together, basically identical to one running but have increased pulling power.
3 - Holding the shaft of a running DC motor, the supply is capable of supplying 4A.
4 - No appreciable heat generated to date., ie. the heat sinks are big enough, especially due to lowering the input voltage from 24
to 12.

Again, thank you all that have assisted. This is a great forum because it is manned by generous and patient people willing to help.

Thank you... Tom

 

[Bluejets]

Love the old style gauges.

Remember Dad had similar looking in his auto workshop switchboard which came out of a WW2 bomber of some description.

I'd have thought the T03 style cases would have a higher dissipation rate.

I still prefer the much more efficient switching regulators.