# LM317 heat during operation

#### HLab

Joined Feb 16, 2020
3
I constructed a power supply that steps down a 12v dc to ~3 to 3.2v dc.

Resistors are 1/2 watt

Vin = 12.7v (full charged lead acid battery)
Vout = 3.2v

Measured temp during the time period of 1 hour and 30 mins max's out around 63 degrees celsius and usually goes down to around 50+ degrees celsius then goes up again at max. Temp at 0 hour is 31.2 degrees celsius. Measurements are without heat sink.

Operation requirements:
It is required to operate 7 hours daily

I have already read the datasheet and slightly understood it. But I still need real world experiences and recommendations.

Question:
1. Is the degree of heat still reasonable specially for the required operation daily?
1a. If yes, can it possibly cause fire in the long run?
1b. If no, what are my other options? Given that 12v is the only source voltage.

I tried to drop the voltage before entering the LM317 Vin using a voltage divider from 12v to 9v. It works without load, when I attach the load the Vin and Vout voltage drops significantly rendering it useless.

Question:
1. Why is this happening? This is a part I am confused.
2. Is there another option to lower down the the source before it goes to Vin? I think 9v is the optimal voltage to drop, lower than that won't activate the LM317 (as I understand in the datasheet)

#### crutschow

Joined Mar 14, 2008
34,075
1. Is the degree of heat still reasonable specially for the required operation daily?
60°C is no problem.
The LM317 can go to a junction temperature of over 100°C without failure.
If it gets too hot it will automatically reduce the current to limit any further heat build-up.
1a. If yes, can it possibly cause fire in the long run?
No, not unless there's a serious failure.
You might want to consider a fuse (say 0.5 to 1A) at the battery output terminal to protect against any accidental shorts, since a lead-acid battery can deliver enough current to start a fire otherwise.
1. Why is this happening? This is a part I am confused.
A voltage divider has an output resistance equal to the equivalent of the two resistors in parallel.
It's the load current through that resistance that lowers the divider voltage.
That's why voltage dividers are normally used only with a very small load current.
2. Is there another option to lower down the the source before it goes to Vin? I think 9v is the optimal voltage to drop, lower than that won't activate the LM317 (as I understand in the datasheet)
A resistor in series with the LM317 input will reduce the voltage, but I don't see a particular need for that here.
For example, a 30Ω, 2W resistor in series will reduce the voltage by 5.7V at a 0.19A load current.

The LM317 needs an input voltage about 3V greater than the output voltage, so for 3.3V output you need about 6.3V minimum input.

Last edited:

#### MrAl

Joined Jun 17, 2014
11,280
I constructed a power supply that steps down a 12v dc to ~3 to 3.2v dc.
View attachment 199267

Resistors are 1/2 watt

Vin = 12.7v (full charged lead acid battery)
Vout = 3.2v

Measured temp during the time period of 1 hour and 30 mins max's out around 63 degrees celsius and usually goes down to around 50+ degrees celsius then goes up again at max. Temp at 0 hour is 31.2 degrees celsius. Measurements are without heat sink.

Operation requirements:
It is required to operate 7 hours daily

I have already read the datasheet and slightly understood it. But I still need real world experiences and recommendations.

Question:
1. Is the degree of heat still reasonable specially for the required operation daily?
1a. If yes, can it possibly cause fire in the long run?
1b. If no, what are my other options? Given that 12v is the only source voltage.

I tried to drop the voltage before entering the LM317 Vin using a voltage divider from 12v to 9v. It works without load, when I attach the load the Vin and Vout voltage drops significantly rendering it useless.

Question:
1. Why is this happening? This is a part I am confused.
2. Is there another option to lower down the the source before it goes to Vin? I think 9v is the optimal voltage to drop, lower than that won't activate the LM317 (as I understand in the datasheet)
Hi,

With 12v input and approximately 3v output you have a drop of 9 volts. 9 volts at 0.2 amps means you have 1.8 watts being dissipated inside the LM317. My guess is that without a heat sink the die temperature is higher than 63 degrees C and could be much higher, given that input of 12v. I see you have a diode in series with the LM317 however so that reduces the power down a little maybe closer to 1.6 watts but that's still a little high for that package assuming the normal package with the mounting tab not the TO3 package which is a little better but more rare to see around these days.
Really the max power for that package without a heat sink should probably be limited to 1 watt. That means add 4 more diodes in series with the LM317. Of course then testing is in order.

But you do have to realize that you lose oodles of run time doing it this way with or without the diodes, and that is really the main point to be made here. The reason is because there is a very large power loss due to the way the LM317 operates, not to mention that fact that the LM317 temperature change means the output voltage varies more than you might gander from the data sheet.

So really this application is crying for a buck converter. A buck converter does a true power conversion of the power source voltage to a lower source voltage losing less energy and therefore dissipating less heat in the process.
Since an LM317 draws about 0.2 amps at about 3v in the application that means the 12v battery is being drained by 0.2 amps continuously. Using a buck since the input is 12v and the output is 3v the idealized current would only be 0.050 amps, four times less. That means the battery can provide four times more run time than when using the LM317 in the idealized situation.
In the real life situation there is still some loss in the buck converter, but it is not nearly as much. With 3.2v and 70 percent efficiency you might see only 0.076 amps drawn from the 12v battery. That means a run time of over two times for the 12v battery. The heating effects are much less too so nothing gets really hot.

So this is something to think about. Stay with the LM317 and put up with the heat and reduced run time, or go with a good buck converter (cheap in places like Amazon) and get more run time and much less heating.

Joined Feb 20, 2016
4,428

Switch to one off these, adjusted to 3V out.
It will possibly triple the battery run time as well as reducing the heat.

#### LesJones

Joined Jan 8, 2017
4,174
Another solution using a switch mode regulator is 5 volt USB power supplies made for cars. (Such as this.) All the ones I have seen use an MC34063 regulator IC. They can be easily modified for 3 volts output bay changing a resistor. Most of the £1 shop chains also sell these.

Les.

#### Marley

Joined Apr 4, 2016
501
Although running the LM317 with no heatsink and a case temperature of 60°C is within specification, it is good design practice to keep operating temperatures as low as possible. A small heatsink will make a big difference. High operating temperatures reduce long-term reliability of the product especially if there are electrolytic capacitors nearby.

My "rule of thumb" is that if you can hold your finger on it for some time without pain, then it's OK. If you can't, then it's to hot for long-term reliability. Don't do this test if it's got high voltages on it, obviously.

For your application where the output voltage is a quarter of the input voltage then a switching dropper often makes more sense. Switchers make noise though so have disadvantages for sensitive analog applications.

#### HLab

Joined Feb 16, 2020
3
I have forgot to mention that it will be put in an enclosed acrylic enclosure for water protection specially during rain and washing. It is supposed to be integrated to a motorbike system on its 12v park lights.

You might want to consider a fuse (say 0.5 to 1A) at the battery output terminal to protect against any accidental shorts, since a lead-acid battery can deliver enough current to start a fire otherwise.
A voltage divider has an output resistance equal to the equivalent of the two resistors in parallel.
It's the load current through that resistance that lowers the divider voltage.
That's why voltage dividers are normally used only with a very small load current.
A resistor in series with the LM317 input will reduce the voltage, but I don't see a particular need for that here.
For example, a 30Ω, 2W resistor in series will reduce the voltage by 5.7V at a 0.19A load current.
The LM317 needs an input voltage about 3V greater than the output voltage, so for 3.3V output you need about 6.3V minimum input.
About the fuse, this modified circuit will be attached to a motorbike connected to its 12v park lights. I think it has a fuse already on the battery side. I will just confirm it to be sure.

So if I were to use a resistor in series to Vin, it will lower the heat dissipation of the LM317? But the series resistor will heat up right?

Hi,
I see you have a diode in series with the LM317 however so that reduces the power down a little maybe closer to 1.6 watts but that's still a little high for that package assuming the normal package with the mounting tab not the TO3 package which is a little better but more rare to see around these days.
Really the max power for that package without a heat sink should probably be limited to 1 watt. That means add 4 more diodes in series with the LM317. Of course then testing is in order.

But you do have to realize that you lose oodles of run time doing it this way with or without the diodes, and that is really the main point to be made here. The reason is because there is a very large power loss due to the way the LM317 operates, not to mention that fact that the LM317 temperature change means the output voltage varies more than you might gander from the data sheet.
So this is something to think about. Stay with the LM317 and put up with the heat and reduced run time, or go with a good buck converter (cheap in places like Amazon) and get more run time and much less heating.
Yes, I used a diode in series in the Vin for reverse polarity protection. It was supposed to be a 1N5817 but it's not locally available, yes I can order online but the costs goes higher and I'm trying to lower the cost because I'm trying to make an income making this converters. Because these EL lights that my supplier have runs on 2 double A batteries hence converting them so it can be directly wired up to the motorbike system.

Based on my circuit though, is it correct?

I might have to consider to buy the ready made ones. Though it's much more fun making our own ones.

View attachment 199277
Switch to one off these, adjusted to 3V out.
It will possibly triple the battery run time as well as reducing the heat.
Have you used these before? How is it's heat dissipation? I might have to try this.

Another solution using a switch mode regulator is 5 volt USB power supplies made for cars. (Such as this.) All the ones I have seen use an MC34063 regulator IC. They can be easily modified for 3 volts output bay changing a resistor. Most of the £1 shop chains also sell these.
Les.
This is a good idea also. There are water proof ones available but I have to open this to change some components inside, I might compromise its water proofing.

Although running the LM317 with no heatsink and a case temperature of 60°C is within specification, it is good design practice to keep operating temperatures as low as possible. A small heatsink will make a big difference. High operating temperatures reduce long-term reliability of the product especially if there are electrolytic capacitors nearby.
My "rule of thumb" is that if you can hold your finger on it for some time without pain, then it's OK. If you can't, then it's to hot for long-term reliability. Don't do this test if it's got high voltages on it, obviously.
For your application where the output voltage is a quarter of the input voltage then a switching dropper often makes more sense. Switchers make noise though so have disadvantages for sensitive analog applications.
Actually it is too hot to be touched. I will be adding a heat sink soon and test the circuit again. If a heat sink is installed, theoretically the heat should be the same from the LM317 and the heat sink or one is still hotter than the other?

Sorry but what is a switcher?

#### SamR

Joined Mar 19, 2019
5,020
Switching power supply voltage control instead of linear regulation. Basically PWM voltage control.

#### HLab

Joined Feb 16, 2020
3
Switching power supply voltage control instead of linear regulation. Basically PWM voltage control.
Is it the same one used on a DC motor for speed control? Because I have one currently installed in my DIY grass trimmer.

If yes, I can replace the POT into a resistor right? to make in a fixed voltage. Though size comparison, it will be a bit bigger than the one recommended above.

Joined Feb 20, 2016
4,428
Is it the same one used on a DC motor for speed control? Because I have one currently installed in my DIY grass trimmer.
No. They are similar in some ways but you will need a power supply, not just a PWM controller.

#### MrAl

Joined Jun 17, 2014
11,280
I have forgot to mention that it will be put in an enclosed acrylic enclosure for water protection specially during rain and washing. It is supposed to be integrated to a motorbike system on its 12v park lights.

About the fuse, this modified circuit will be attached to a motorbike connected to its 12v park lights. I think it has a fuse already on the battery side. I will just confirm it to be sure.

So if I were to use a resistor in series to Vin, it will lower the heat dissipation of the LM317? But the series resistor will heat up right?

Yes, I used a diode in series in the Vin for reverse polarity protection. It was supposed to be a 1N5817 but it's not locally available, yes I can order online but the costs goes higher and I'm trying to lower the cost because I'm trying to make an income making this converters. Because these EL lights that my supplier have runs on 2 double A batteries hence converting them so it can be directly wired up to the motorbike system.

Based on my circuit though, is it correct?

I might have to consider to buy the ready made ones. Though it's much more fun making our own ones.

Have you used these before? How is it's heat dissipation? I might have to try this.

This is a good idea also. There are water proof ones available but I have to open this to change some components inside, I might compromise its water proofing.

Actually it is too hot to be touched. I will be adding a heat sink soon and test the circuit again. If a heat sink is installed, theoretically the heat should be the same from the LM317 and the heat sink or one is still hotter than the other?

Sorry but what is a switcher?

Hi,

Ok this just changed. When you say it has to be put into a box of some type then it does not matter as much if you add a resistor or more diodes. You must use a buck converter (also called a switcher but of the buck type).
That is because heat builds up which means temperature rises even more when a heat source is enclosed in a chamber that does not allow much heat to escape. Yeah, it could work if things are such that the heat does not build up too much but that is a matter of testing as it is hard to predict exactly what temperature will be reached inside. It's better to use a buck though as the heat will be much much less so the temperature will be much much lower.