advertisement


LM317 Regulator

Isobarik

pfm Member
Is it necessary to use a heat sink on the LM317?

I ask this as I built a power supply for my Naim 42.5 years ago

I based the power supply on this: I built this about 20 years ago and it's still going strong, but I can't remember if I used a heat sink on the regs.
I must thank Roy K Riches for this, a lovely and very helpful guy!
 
If it's in a Hicap like application it certainly needs nothing like as much as Naim give it. Their heatsink is total overkill.

It depends on how much voltage it's dropping and how much current is drawn at that reduced voltage but beyond that the data sheet is fairly unhelpful.

I've certainly implemented them with just a bolt on to220 heatsink for preamps and they don't seem to get very warm at all. Expo drop 6V on ones epoxy'd to the pcb without any heatsink with no issues.

 
This site contains affiliate links for which pink fish media may be compensated.
Whether a regulator needs a heatsink is dependent on the volts dropped across it and the current drawn through it.

37V supply will drop 13V across the regulator and a 42.5 draws about 150ma.

So the regulator needs to dissipate 2W ish, no heatsink required
 
If it's in a Hicap like application it certainly needs nothing like as much as Naim give it. Their heatsink is total overkill.

It depends on how much voltage it's dropping and how much current is drawn at that reduced voltage but beyond that the data sheet is fairly unhelpful.

I've certainly implemented them with just a bolt on to220 heatsink for preamps and they don't seem to get very warm at all. Expo drop 6V on ones epoxy'd to the pcb without any heatsink with no issues.

Hi colablue,
Thank you so much for the info.
I'll inspect it when I return home, I've got plenty of thick aluminium plate, I'll mount them on two pieces of that with some compound.
 
This site contains affiliate links for which pink fish media may be compensated.
Whether a regulator needs a heatsink is dependent on the volts dropped across it and the current drawn through it.

37V supply will drop 13V across the regulator and a 42.5 draws about 150ma.

So the regulator needs to dissipate 2W ish, no heatsink required
Right! thanks for that David.
I'll leave it be then.
 
Sorry - I'm going to dissent slightly, there:

The 317/337 is a 'TO220' Package, and about the most you can dissipate in it, without heatsink - is 1-1.2W, This is in the regulator datasheet.

So - my preference would be do add a small heatsink; it might only be tiny/cheap such as @colasblue suggests.

You might well get-along well, utterly without: but on a warm day, under full load suggested - the 317 will ultimtely fold-back its current output to preserve itself, and , perhaps, just perhaps, that'd be perceptible.

NB
The LM317/337 are the 'Horseshoe crab' of of voltage regulators - they'll survive, regardless.
 
I recon there is some form of a heat sink used in the great Exposure 13 phono preamp.
NFPGpUl.jpg
 
Sorry - I'm going to dissent slightly, there:

The 317/337 is a 'TO220' Package, and about the most you can dissipate in it, without heatsink - is 1-1.2W, This is in the regulator datasheet.

So - my preference would be do add a small heatsink; it might only be tiny/cheap such as @colasblue suggests.

You might well get-along well, utterly without: but on a warm day, under full load suggested - the 317 will ultimtely fold-back its current output to preserve itself, and , perhaps, just perhaps, that'd be perceptible.

NB
The LM317/337 are the 'Horseshoe crab' of of voltage regulators - they'll survive, regardless.
Thanks Martin, I'm going to implement your recommendation. Better safe than sorry! as there is no ventilation in the enclosure and it's pretty tight in there.
 
I seem to remember the datasheet for the regulators had a device temperature Vs noise plot with lower temps being better. The better brands (not the LM series) had a lot more graphs on the datasheet. I can't remember the manufacturer now but the part suffix was not LM but something else.
 
The LM317 was developed by National Semiconductor in the first place. It would be unusual for a second source to provide more graphs than the original, but it is likely that TI have removed info since their take over.
 
If it's in a Hicap like application it certainly needs nothing like as much as Naim give it. Their heatsink is total overkill.

It depends on how much voltage it's dropping and how much current is drawn at that reduced voltage but beyond that the data sheet is fairly unhelpful.

I've certainly implemented them with just a bolt on to220 heatsink for preamps and they don't seem to get very warm at all. Expo drop 6V on ones epoxy'd to the pcb without any heatsink with no issues.

Can anyone put a number on how much actual heat-sinking those tiny things do? I mean, it must be something, and it's got to be better than a bare TO-220, but how much, exactly?

EDIT; it took 20 minutes of searching, but I found a vendor who stated a figure (the product, or its clones, is common, specs not so much):


Material: Aluminium
Size: 21(h)x15(w)x11(d) (+/-1mm)
Thermal Resistance: ~20°C/W
 
Last edited:
This site contains affiliate links for which pink fish media may be compensated.
Good question.

I've used a few of these (https://www.rapidonline.com/trusemi-bk-t220-0061-to220-bolt-on-heatsink-36-0150) and they are rated at 17degC/watt. Work nicely too, in a use like the OP has.

So the previous tiny profile extrusion might be 20-25degC/watt; but that still means that at 2w out, the reg is running 70degC or less, and that's the difference between a long life, and edge of current-foldback, for such a part, in the OP's requirement. Much better than nothing - tho' I'd prefer fitting 'larger'.


Edit to add:

NB for reference, the T0-220 package (widely used for Vregs like this, and also medium power transistors, even small power amps like the ancient TDA2030 etc) is rated for 1.2w dissipation without a heatsink, and max 20w with.

In practice call that 1w unaided in cool air, and about 15heatsunk (big sink!) both, abs.max for long life - the package tab size , and esp junction:tab thermal resistance is a firm limit.
 
PS @Isobarik

- while you are playing - stick 100uF of simple good-quality electrolytic across the input to the reg (the green cap: even, in place of the green cap .)

These simple regs, usually work better like that, because the output load-induced ripple appears on the input pin as it draws on the very-local supply, and the 'gain' of the internal error amp is a bit limited to suppress that as a negative effect. So giving it a bit more, very-local reservoir, can help.

The above isn't in the datasheet, but documented elsewhere. The 'mini 0.1uF' input cap recommendation in the datasheet, is only to suppress the reg oscillating if it's more than ~150mm from the nearest reservoir cap! Your reg sensibly uses larger, but neither value is enough for the 'better/best behavior' you can get from such a jellybean part. And, that's also why (say, Naim) start with 10uF in that position for every implementation.
 


advertisement


Back
Top