LM317 heat sink connected to the ground

[symbianas]

Hi, I bought such down-step voltage converter:



Can I mount it with the same scree to the frame where the ground is? Or does the heat sink need to be separated from the ground?

[bobbydazzler]

You can't connect it to ground the metal tab is the output pin on the lm317.

[Zero999]

You can't connect it to ground the metal tab is the output pin on the lm317.

That doesn't always prevent the tab from being grounded. If the rest of the circuit is isolated, i.e. power supply for the LM317 and the circuit it's powering are both isolated from ground, then the tab can be grounded. On the other hand, if any other part of the circuit is grounded, then it could cause a short circuit. Either way, it's better to have it floating.

[Ian.M]

That's not a step down voltage converter, its a linear regulator.  The difference is it throws away the voltage drop across it multiplied by the load current as heat.  A step down (switching) voltage converter converts input power (Vin * Iin) to output power (Vout * Iout), trading voltage for current, usually with reasonably high efficiency so typically has far less heat to dissipate for the same voltage drop and load current.

A tiny little heatsink like that is likely to have a thermal resistance to ambient of about 15°C/W in free air, which it hasn't got due to the obstructions of the capacitor,  nearest terminal block and the preset, and due to the close proximity of the capacitor, you probably shouldn't run the heatsink much over 85°C.   Also, I don't see any signs of heatsink compound, so at minimum you should remove the heatsink, apply some and refit it.

The poor heatsinking is going to seriously limit the load current that module can safely supply, especially if the input voltage is significantly higher than the minimum. Thankfully its the thick tab version of the TO-220 LM317, so R_θJC(bottom) is negligible.  It may even be worth removing the heatsink, desoldering the LM317, and resoldering it fitted under the PCB with a right angle bend in its legs so it can be mounted flat to a significantly bigger heatsink, with a silpad and top hat washer for insulation so you can ground the heatsink, and with a pair of plastic spacers so you can screw down the PCB to the new heatsink by its mounting holes.

[Zero999]

Yes, it's a linear regulator, rather than a step-down converter. I wouldn't bother improving the heatsinking, because it I wanted more power, I'd use a proper step-down converter. This sort of module is only good for dissipation a couple of Watts, at most, so should only be used for low power devices. For example, if I wanted to power a 5V circuit, from a 12V supply, I wouldn't uses it to power anything which draws more than a couple of hundred mA, at most.

[Ian.M]

A couple of hundred mA dropping 12V to 5V is overoptimistic.   Dropping 7V @200mA + 10mA quiescent current in the Adj divider it will be dissipating nearly 1.5W which with that crappy heatsink and partial obstructed airflow, it likely to result in more than 30°C temperature rise.   If its in an enclosure with limited airflow, on a hot summer's day that would cook that probably 85°C capacitor right next to the heatsink.   IMHO, 1W dissipation would be about as hard as I'd care to push it without forced airflow.

[bobbydazzler]

No protection diodes and bad capacitor choice(assuming the big electrolyte is only 85c rated, and assuming the adjust pin doesn't have 10uf+) is the biggest problem with this design. 
1uf x7r mlcc on input and output pins with 10uf mlcc on the adjust pin would be the best choice for performance and max power output, I don't think there would be any problem running the lm317 at 100c+ if you didn't need to worry about protecting the capacitors.  A bigger heatsink would be nice but it would also increase the volume of the device and maybe you need it to be that small(though there's probably better ways to make the volume smaller too).