LM317 + MJ2955 = overheating at 5v - 3A

Thread Starter

Dimitris1986

Joined Dec 13, 2023
3
Hello there. I'm trying to build a linear power supply for my dj equipment. Controller Pioneer DDJ SX3, rated at 5V - 3A.
Psu.jpg

As you can see i'm using that circuit to drain 3A at 5.15vDC.
Output choke is from PC PSU's output, all 100nf are too close to lm317, zobel also.
The Problem here is power supply overheating even draw 2.5 - 3A with controller connected. Bridge rectifier has 80-85 celcius, lm317+2955 75-80 celcius. isnt normal i believe.
My heatsink is from old sony integrated amplifier rated 2x30watt rms. i think is enough without forced cooling.
Without controller connected, temps are about -20 celcius with constant load at 4A (resistors and halogen lamp for load).
what going wrong here? Is too many 40000uF at rectifiering? is too many power dissipation? Need lower voltage trafo? Circuit have errors?
Thanks in advance
 

Papabravo

Joined Feb 24, 2006
21,225
What is going wrong is that you are asking your parts to dissipate over 30 watts and most TO-3 packages have a θjc of 1.52 °C/W. That means from case to junction there will be a temperature rise of approximately 46 °C. which is almost 115 °F. Assuming the case is at room temp of 68 °F, add 115 °F to that giving 183 °F. I would not put my lips on that part – would you?
 

dl324

Joined Mar 30, 2015
16,898
As you can see i'm using that circuit to drain 3A at 5.15vDC
That's not what I calculate. I get 4.4V.

\( V_{OUT} = 1.25V(1+\frac{680\Omega}{270\Omega})=4.4V\)

The current set resistor you've chosen isn't sufficient to guarantee that the LM317 will regulate. The regulator requires a minimum load current of 10mA. Some will work with a smaller load, but it isn't guaranteed. I've also come across one that wouldn't regulate unless the minimum load was more than 10mA.
The Problem here is power supply overheating even draw 2.5 - 3A with controller connected. Bridge rectifier has 80-85 celcius, lm317+2955 75-80 celcius. isnt normal i believe.
Have you used this data to calculate junction temperature?

You're operating the LM317 at a tenth of it's rated current. I'd use a current divider on the input and operate the regulator near its maximum current with your highest expected load. Put the transistor and regulator on the same heatsink and use thermal compound.
My heatsink is from old sony integrated amplifier rated 2x30watt rms. i think is enough without forced cooling.
What you think doesn't matter. What is the thermal resistance of the heatsink? Are you using thermal compound/washers?
 

KeithWalker

Joined Jul 10, 2017
3,089
The power that is being dissipated as heat is the product of the difference in the input and output voltages (16 - 5.5 = 10.5) times the output current. e.g. 10.5V x 3A = 31.5 Watts. That is quite a lot of power to dissipate without a cooling fan. You certainly will see the temperature go up on the heat sink. A lower voltage power source would certainly help to cool things down.
 

Thread Starter

Dimitris1986

Joined Dec 13, 2023
3
i'm sorry, divider is 220ohm+680ohm. So i need lower voltage power source.
317+2955 are in same heatsink. i'm using thermal compound/washers and thermal paste also.
i dont know thermal resistance of the heatsink.
Ok i'm understand that, but bridge rectifier why is getting so hot? is behind active parts of circuit. can increase to 80 celcius with 3A load?
 

Audioguru again

Joined Oct 21, 2019
6,687
11.5VAC x 1.414= 16.26V peak minus 1.8V for the rectifier bridge= 14.46V.
The output of the LM317 is 1.25V/270 ohms= 4.63mA x 680 ohms= 3.15V + 1.25V= 4.4V.
14.46v - 4.4V= 10.1V x 3A=30.3W.

Does anybody play as Sony stereo with a full blast tone continuously? No. Its heatsink is too small for 30.3W continuously.

The 14.5VDC rectified voltage is too high for a 4.4V output.
 
What about a switching regulator? It will be way more efficent and wont heat up almost at all. You can use LM2596 buck regulator IC for this. LM2596-5 is pre built to regulate the output voltage to 5V. If you need a very precise output voltage, you can use LM2596-ADJ that can output an adjustable output voltage that you can set with a simple resistive divider feedback. This circuit is more efficent and it requires less components.

You can consider making one of those.

Have a look at the datasheet of LM2596, it is easy to understand in my opinion, and it contains example schematics that might be what you need already.

If a switching regulator causes too much noise for your application you can add LC filters to the output of the buck converter. Use ceramic capacitors in parallel to electrolythic ones for better high frequency noice filtering. I think this should suffice.
 

AnalogKid

Joined Aug 1, 2013
11,027
Current sharing between the 317 and 2955 is determined by the 2955's base-emitter resistor. In your case, assuming Vbe is 0.7 V at low currents, that lets 70 mA through the 317 before the 2955 starts to bypass it. That should be enough for stable operation, but more current through the 317 wouldn't hurt.

It looks like there is around 8.8 V across the 317. At 70 mA, that is a little over 0.6 W dissipated in the 317. Since a large heatsink is available, you could double the 317 current.

Please repost your schematic with the correct resistor values. Also, make sure it has a unique reference designator (R1, C3, etc.) for each component.

ak
 

Papabravo

Joined Feb 24, 2006
21,225
Was this supply actually designed or were parts just thrown at a chassis to see if it would work? Hoping for the former, but suspecting the latter.
 
Top