microcontroller power supply overheated

[karthickiw]

hi friends,

I'm using LM317T regulator and its connected resistors value are below

R1= 680 ohm
R2 = 220 ohm
Vin = 12 volt from battery
vout = 5 Volt to PIC microcontroller

My PIC microcontroller board is take maximum 100mA current. For this low current it's produce the more heat on regulator body. Give any idea to reduce the temperature in regulator without using heat sinks.

[dan king]

Hi,

Take a look at the LM2592 parts from National Semiconductor. They are switcher type buck regulators and therefore run very cool when "dropping" 12 volts down to 5, they are even rated for 2 amps. I've used these and am very happy with the performance.

They do require a couple more components (a schottky diode and an inductor) but well worth the advantages of lower power and less heat.

[SherpaDoug]

Your TO220 package should be able to handle the 700mW without too much effort. How is it mounted? Can you solder it down to the PCB so the fiberglass can help dissipate the heat? Can you heatsink it to ANYTHING, even plastic is much better than air. Can you get any airflow?

Another option is to put a 1W 35 Ohm resistor between the 12V supply and the LM317T and its input cap. This would drop 3.5V in the resistor and 3.5V in the regulator splitting the heat between the two.

Changing to a switcher chip as Dan King suggested may be the best answer if the cost, noise and complexity are OK. It will make for a "Greener" product in the end.
_________________
The search for better is endless. Instead simply find very good and get the job done.

[Ttelmah]

There are a lot of things to consider here.
First, just how hot is the package?. A regulator, can work typically with a junction temperature up at something like 125C, and even being 'careful', and keping this down to something like 100C, still leaves you with a painfully hot package, before any problems arise....
Next question, is what the environment is like?. The performance of the TO220, in terms of it's ability to get rid of heat, will vary massively with features like the airflow, what else is nearby, etc. etc..
Third question, are you sure you are working off '12v'?. You have to remember that a car battery (for example), will typically sit at about 14.5v when the car is running, and a little unregulated '12v' wall-wart, may well develop something like 17v, if the total load is only 100mA. Makes a big difference to the heat involved...
Now, the TO220, mounted on it's own, without any heatsink, is normally quoted as having a junction to free air temperature rise figure, of below 60C/W. If your supply is '12v', you have to dissipate 0.7W, and your junction would then be about 42C above the ambient temperature. The case will be at about 5C below this, giving a case temperature for a 25C environment of about 62C. This is entirely 'acceptable' in most situations - but hot. If though the supply was really at something like 17V, and current drawn was actually 110mA, the junction temperature would be up at 104C, and the case temperature at about 96C, and this would be getting pretty nasty.....
Another poster has mentioned bolting the transistor to the board, and with a board having a 'power plane' in particulr, this really can help, easily halving the temperature rises involved.
Using a switcher, is also good, and for this low current, something smaller than the LM2592 can be used. I'd actually suggest the LM22671, which is available in 8pin packages, and gives 500mA, using tiny ceramic capacitors. However you do have the question of noise. I commonly use a 'composite' regulator, when running off a higher supply voltage (24v in industrial systems), having a switcher, giving a seven to 8v rail, and then using a linear regulator from this, for analog circuitry. For most digital work, this is unneccessary. Difference would be an expected dissipation around 50mW, with the right inductor, _and careful board layout_. This is _vital_ round switchers.

Best Wishes

[karthickiw]