LM338 Dissipated Heat

This is the place for any magazine-related discussions that don't fit in any of the column discussion boards below.
[email protected]
Posts: 9
Joined: Sun Oct 02, 2005 1:01 am
Contact:

LM338 Dissipated Heat

Post by [email protected] »

I have a general question about the heat produced by the LM338 TO220 package voltage regulator. I am regulating 18 Vdc down to 6 volts to power the servos of a robotic arm. After running the arm for about 5 minutes the heat sink attached is nearly too hot to touch. I currently do not have any cooling air being blown across the heat sink, but I still feel this is probably too hot for such a short run time. There is approximately 1.5 Amps of current draw on the regulator.<p>Is this abnormally hot?
User avatar
philba
Posts: 2050
Joined: Tue Nov 30, 2004 1:01 am
Location: Seattle
Contact:

Re: LM338 Dissipated Heat

Post by philba »

abnormaly hot for how you are using it? no. is it a problem? yes.<p>You've got 12 volts of dissipation going on (18-6). 12*1.5A is 18W of wasted energy. I'd expect that poor lil to220 to get hot enough to fry an egg or an errant finger. <p>Can you find a lower voltage source? I'd try to get the input voltage as low as possible. From the datasheet, it looks like the min input is about 2.6V over the output voltage at 3A. I'd go with 3 so 9V would be a good target number for the input V to minimize the waste heat generation.<p>Also, you didn't say how big your heatsink is but bigger is definitely better (especially surface area). A fan could help but I think you can solve the vast majority of your problem by simply reducing the input voltage. It will never run cool but you can cut your heat in half but going to 9V input.<p>[ October 02, 2005: Message edited by: philba ]</p>
User avatar
Chris Smith
Posts: 4325
Joined: Tue Dec 04, 2001 1:01 am
Location: Bieber Ca.

Re: LM338 Dissipated Heat

Post by Chris Smith »

The Lm 338 is capable of operating temperatures of 150 C or 302 degrees F. Way past finger HOT. <p>It has a seven amp maximum current rating with five amps as the standard. <p>Unless your trying to conserve power, its running just fine and wont cause you any problems.<p> Make sure you have heat sinks, and fans if possible, and again, unless your trying to conserve on power, ITS doing its job as intended.
Robert Reed
Posts: 2277
Joined: Wed Nov 24, 2004 1:01 am
Location: ASHTABULA,OHIO
Contact:

Re: LM338 Dissipated Heat

Post by Robert Reed »

Jamey
Your setup is dissipating 18 watts across the TO-220 package. I dont know what size heat sink you are using, but a good rule of thumb is this--press your finger hard on the chip after it reaches max. operating temperature (5 minutes is long enogh). If you can hold your finger on it, you are probably OK. The normal threshold of pain (heat) is about 140 degrees farenheit.Remember that the thermal resistannce from the actual chips junction to the grease seal to the actual heat sink is about 2.5 degrees per watt. this equates to a 45 degree (C) rise of the junction temperature before we ever get to heat sinking. So the finger temperature you feel is still somewhat cooler than the actual chip juntion. If you want to cut down package dissapation and you are committed to your 18 volt supply, install a series resistor ahead of the chips input. Three 12 ohm @ 5 watt resistors in parralel would drop 6 volts, therby leaving 6 olts acoss the regulator, and consequently cutting its dissapation in half (9 watts). You can reduce this even further by increasing the series resistor, but calculate the wattage you will need for any change here. By making this change even a relatively small heat sink should suffice for the chip. In either case, this is a terrible waste of power, but you gotta do what you gotta do.<p>[ October 02, 2005: Message edited by: ROBERT REED ]</p>
rshayes
Posts: 1286
Joined: Tue Mar 04, 2003 1:01 am
Contact:

Re: LM338 Dissipated Heat

Post by rshayes »

The LM338 is specified with a maximum junction temperature of 125 C. The voltage drop across the device is 12 volts. A load current of 1.5 amps will cause the LM338 to dissipate 18 watts. The junction to case thermal resistance is given as 4 C/watt for the TO-220. The temperature difference between the junction and case should be less than 72 C. Working backwards from the maximum junction temperature would indicate that the heat sink should remain below 53 C (127 F) to maintain the junction below its maximum rating. It does sound like your heat sink is hotter than this.<p>One solution to this is to use a dropping resistor in series with the regulator input. This transfers some of the heat dissipation from the regulator to the resistor.<p>The dropout voltage of the regulator is less that 3 volts. This means that the input voltage to the regulator should be kept above 9 volts for a 6 volt output. The resistor should be below 6 ohms (9 volts drop divided by 1.5 amps). At 1.5 amps load, it will dissipate 13.5 watts.<p>The maximum dissipation in the regulator will occur when half of the voltage drop is across the regulator and the other half is across the resistor. This will occur at a load current of 1 amp. At this current, the regulator will be dissipating 6 watts. This is quite a difference.<p>Using 6 watts, the temperature difference between the junction and case would be 24 C. This would put the allowable heat sink temperature at 101 C or about boiling water.<p>Of course, you might have to heat sink the resistor.
User avatar
philba
Posts: 2050
Joined: Tue Nov 30, 2004 1:01 am
Location: Seattle
Contact:

Re: LM338 Dissipated Heat

Post by philba »

<blockquote><font size="1" face="Verdana, Helvetica, sans-serif">quote:</font><hr>
The Lm 338 is capable of operating temperatures of 150 C or 302 degrees F. Way past finger HOT. <p>It has a seven amp maximum current rating with five amps as the standard. <p>Unless your trying to conserve power, its running just fine and wont cause you any problems.<p> Make sure you have heat sinks, and fans if possible, and again, unless your trying to conserve on power, ITS doing its job as intended.<hr></blockquote><p>According the National datasheet, the LM138 goes up to 150C and the LM338 (which he has) to 125C.<p>Still, I don't think its a good idea to push a device near the spec'd maximums unless you are doing destructive testing. It might work, it might not but all it would take is a jammed motor or accidental short somewhere to push it over the edge. I'd work at reducing the power dissipated in the lm338 - several decent suggestions have been given though I'm a fan of the start with a lower voltage approach.<p>Phil
[email protected]
Posts: 9
Joined: Sun Oct 02, 2005 1:01 am
Contact:

Re: LM338 Dissipated Heat

Post by [email protected] »

Thanks for the replies. I will change the input from my transformer to the center tap voltage. This should cut the voltage down to 9 and hopefully reduce the heat considerably. Thanks again!
[email protected]
Posts: 9
Joined: Sun Oct 02, 2005 1:01 am
Contact:

Re: LM338 Dissipated Heat

Post by [email protected] »

The center tap does not provide a voltage high enough. When I activate the relay that energizes the power supply I can not get enough power for the supply to stay on. I have gone as far as setting the power supply to 1.25 volts and slowly bringing it up to achieve the desired 6 volts. As soon as it is loaded down it the supply shuts off. I have tested the circuit with as little as 8 volts with a bench top supply and bypassing the transformer of the circuit and it seems to work fine, but from the center tap NoGo.<p>If I decide to put LM338s back to back to make small steps down how many should I use? Can I step down from 18 to 14 to 10 to 6 using a total of three regulators or would two be sufficient. I am really needing this supply circuit to run cool.
User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: LM338 Dissipated Heat

Post by MrAl »

Hi there,<p>It could be that the ripple is too high so that
when you lower the ac voltage the dips go too
low and cause the problem. You should try adding
more filter caps first at the lower voltage.<p>If that doesnt work and you cant get another
transformer, you could also consider using
a string of diodes in series to drop some
voltage. The diodes will still get quite warm,
but 1N5400 series will work out ok.<p>Also, you might consider going to one of
National's "Simple Switchers" ic line. These
things connect almost the same as your linear
regulator with only a few extra parts.
Normally, to drop from 18 volts to 6 volts you
would use a switcher if any power is involved
because the switcher doesnt have to drop the
whole 12 volts. The pass transistor inside
turns completely on or completely off, while
in the linear regulator it's partially on all
the time. The partially on transistor in the
linear uses lots of power and gets hot, while
the transistor in the switcher uses much less
power and only gets a little warm.
Inductors for the switchers can be found on the
web at places like Digikey, along with filter
caps.<p>If you'd like to try reducing the voltage again
first, try adding filter caps (1000uf or more)
to help stabilize the dc into the regulator.
As much as 10,000uf may be required. Adding
filter caps not only filters the dc better, this
also slightly increases the average dc output of
a rectifier circuit by storing more charge during
ac peaks. You thus end up with smoother dc that
is slightly higher and might be enough.
Of course if you expect low line conditions at
any time you'll have to test the circuit to make
sure it still works during those line condition
periods.<p>
Take care,
Al
LEDs vs Bulbs, LEDs are winning.
User avatar
philba
Posts: 2050
Joined: Tue Nov 30, 2004 1:01 am
Location: Seattle
Contact:

Re: LM338 Dissipated Heat

Post by philba »

Reading your post, I realized we don't know your exact circuit. How are you taking the center tap? What is the actual voltage after the rectifier? Can you post your schematic?<p>Based on everything you wrote, 9V into your regulator should be just fine. Even if you have significant ripple, say 1V, I think the 338 shouldn't shut down but rather pass the ripple on. A shut down indicates something more significant. Are you sure you are only pulling 1.5A?<p>Phil
[email protected]
Posts: 9
Joined: Sun Oct 02, 2005 1:01 am
Contact:

Re: LM338 Dissipated Heat

Post by [email protected] »

I have created a screen shot of the eagle PCB for my project. The supply getting hot is the upper supply. The circuitry that is additional is for slow turn on and a LED.<p>As for where I am getting the center tap. The transformer has three output lines. The outer two were the 12.6 AC and taking a measurement from one outer line and the center line yielded 6.3 AC So that is where I tapped off for the supply board. After the rectifier and input filter I had 18 volts with the 12.6 AC coming in and just under 8 with the center tap. If I use a bench top power supply it take just of 8 volts to keep the circuit up and running.<p>I can probably make a better schematic using PSpice if that would help with exact values.<p>Is it possible for a transformer to have the three wires on the output side and have a voltage present, but not be considered a center tap transformer. I ask this because I read a voltage from the center wire but the part number does reflect the transformer being center tapped.<p>Image
User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: LM338 Dissipated Heat

Post by MrAl »

Hi Jamey,<p>
I dont think 6.3vac will be high enough to run
your regulator ic chip. Even with 50,000uf
under a load of 1.5 amps the average dc is
just under 6 volts, which isnt enough. The
rectifier diodes drop voltage too under load.<p>Using the 12.6v output, you get something like
13vdc under load which is still too high, but
adding three diodes brings this down to about
10v or so. Using 5 diodes in series might
get you to the right voltage, but again this
will be wasting a bit of energy in the extra
diodes.<p>What would be nice is if you could find a 9vac
transformer that can handle 1.5 amps. Perhaps
simply purchase another one?<p>
I had this same problem when my scanner power
supply blew out about a year ago. It used
a 15v off-line switcher. I considered using
a linear, but no matter how you cut the cake
you end up using extra power (which means it
gets very warm or hot) in normal use, partly
because you have to design for low line.
I ended up using one of National Semiconductors
"Simple Switchers" and it worked out perfectly.
The ac voltage coming in could be as high as
24vac so there was no worries about the circuit
cutting out at low input line (108 volts on
a 120vac line normally, but around here it's more
like 90 volts low line!) Put out a constant
15vdc up to 3 amps max output, and the parts
involved were relatively cheap.
I was lucky however in that i had a 50uH choke
laying around i had purchased many years before
that, and it worked fine in that app, but Digikey
and other online parts houses have chokes too.<p>Seriously, you should consider going to a
switcher of some type. The benefits add up
quickly for the cost of a few extra parts.
Think of how nice it would be to be able to use
a tiny heatsink maybe 1 inch square!<p>
Take care,
Al
LEDs vs Bulbs, LEDs are winning.
User avatar
philba
Posts: 2050
Joined: Tue Nov 30, 2004 1:01 am
Location: Seattle
Contact:

Re: LM338 Dissipated Heat

Post by philba »

It's hard to tell from the picture but I think your diagram shows both regulators as having the input voltage going to the adjust pin, not the input. It looks like you are using them backwards. Your pin labels agree with the LM338 datasheet. Is this just a typo on your schemetic or maybe you are inserting them backwards to how your schematic has them (two wrongs make a right here)? <p>With 12.6VAC, you should be seeing something like 17.8 - 2 diode drops or about 16.4VDC. With 6.3VAC, it should 8.9 - 2 drops or about 7.5V which is below the lower input limit for 6V. Still, I don't think you should be seeing shutdown, just something like 5.5V (but load dependent as al said).<p>What is "variable" in your diagram? a potentiometer I would guess<p>by the way, since you are using eagle, did you not use their schematic editor?
User avatar
philba
Posts: 2050
Joined: Tue Nov 30, 2004 1:01 am
Location: Seattle
Contact:

Re: LM338 Dissipated Heat

Post by philba »

OK, I've looked at your circuit enough to deduce that its basically the "slow turn-on" circuit from page 9 of the datasheet. Is this correct?<p>The diode is wired in backwards from the datasheet diagram. Try fixing that.<p>If that doesn't do it, I would strip out the slow turn-on part of the circuit and see what you get. Go with just the regulator, the 2 voltage ref resistors and the .1 uF cap (see below). Its possible that's where the problem lies.<p>I dont see the .1 uF capacitor - this can help control ringing and excessive heat generation. it should go between Vin and ground very close to the regulator (one each for both of them).<p>finally, just a very minor quibble but you don't need a voltage divider for the LED, just a dropping resistor - probably in the 330-470 ohm range though a standard 20mA 2V led should have a 200 ohm resistor). I bet yours never lit up with those resistor values.<p>good luck.
Phil
[email protected]
Posts: 9
Joined: Sun Oct 02, 2005 1:01 am
Contact:

Re: LM338 Dissipated Heat

Post by [email protected] »

The slow turn on portion of the circuit is actually from the Nuts and Volts magazine from a few months back. The values in the eagle schematic aren't correct at all. I more or less uploaded the schematic so you could see the interconnects. I placed the parts on the design for spacing and such. This is the first time I have created anything from scratch like this.<p>I may try and upload a PSPice version some time this weekend if I have time to create it. I appreciate all the comments on my circuit and you all have given me much to think about.<p>Thanks again. I will keep you posted on the final results.
Post Reply

Who is online

Users browsing this forum: Ahrefs [Bot] and 35 guests