Heatsink selection

Hi

I had a question regarding heatsink selection. I'm building an single channel of API312. My power transformer is a Antek 10VA, dual coil 15V. To make the 48v, I'll use the voltage tripler of my psu, which would get me around 64V, in the case of. I got about 16V to drop across an LM317HV. For the bipolar rails, about 6.5v of drop across LM317/337.  Do I really need an heatsink on each regulator? Let's say the 48V rail would get warm, is a little TO-220 heatsink would do the trick or I would be better of to go with one big ass heatsink? Thanks

What is the current? What is the voltage drop?

Ass-uming 16V (at high wall voltage?) and 10mA for one Phantom, 2mA regulator current, it is shy of 0.2W. A TO-220 in free air is good for nearly a whole Watt. You seem to be fine with a naked reg.

Once you know the power being lost in the regulator (as heat), which will be roughly (the voltage drop across the regulator) X (the current through the regulator), you can calculate the temperature rise of the case from the data sheet information.

There should be a theta(j-c) and a theta(c-a) figure. Add these to get the theta(j-a) figure, which is the thermal resistance from the junction to the air. This is specified in degC/W.

Multiply the theta(j-a) number by the number of watts the device has to dissipate and this will give you the temperature rise of the regulator case. Add the expected maximum operating temperature inside the enclosure (ambient max) to get a temperature for the regulator case. Is this too hot? Will it likely start to char the PCB over time? If so then work out what temperature ridse you want and add a heatsink with the appropriate thermal resistance to achieve this rise.

PRR said:

What is the current? What is the voltage drop?

Ass-uming 16V (at high wall voltage?) and 10mA for one Phantom, 2mA regulator current, it is shy of 0.2W. A TO-220 in free air is good for nearly a whole Watt. You seem to be fine with a naked reg.

Click to expand...

Thanks PRR. I guess it shouldn't be a problem with the bipolar rails, too. voltage drop shouldn't be more than about 7v or so.

you will be fine with one of this

Whoops said:

you will be fine with one of this

Click to expand...

That's what I was planing to use. Thanks

little come back here  8)

I'm calculating the heatsink required for 4 channels of api312.

Let's say one channel is about 70ma, and my voltage drop is about 6.5v across my LM317 and my LM337. Then, this gives me about 0.5W per channels, x 4 = 2W. So it's sure I need a heatsink of my regulators.

The junction to ambiant thermal resistance is about 23.5°C/W on the LM3617 data sheet, x 2W, I got about 47°C + ambiant temp, let's say 30°C, I'm near 80°C at the regulator. I would like to know if I'm still ok with the little heatsink shown a post or 2 above? Thanks

Deepdark said:

little come back here  8)

I'm calculating the heatsink required for 4 channels of api312.

Let's say one channel is about 70ma, and my voltage drop is about 6.5v across my LM317 and my LM337. Then, this gives me about 0.5W per channels, x 4 = 2W. So it's sure I need a heatsink of my regulators.

The junction to ambiant thermal resistance is about 23.5°C/W on the LM3617 data sheet, x 2W, I got about 47°C + ambiant temp, let's say 30°C, I'm near 80°C at the regulator. I would like to know if I'm still ok with the little heatsink shown a post or 2 above? Thanks

Click to expand...

Those little heatsinks have a thermal resistance of about 20degC/W. So for 2W per device you will get a rise of 40degC at the heatsink.
The junction to case thermal resistance is about 16degC/W.
So for an ambient of 30degC your junction temp would be about 72degC for 2W.
I'd suggest this is too high for long term reliability and would use a more substantial heatsink.
It may be more practical to use a single large heatsink and attach all the devices to it. Something less than 10degC/W.

Thanks a lot

A couple of errors in my last post.
The j-c value looks too large for a TO220 device. I don't have the datasheet for that particular device, but the j-c value for a TO220 LM317 device is 4degC/W. You should also add a thermail resistance from case to heatsink of about 0.25degC/W.
So for a 20degC/W heatsink with 2W power to dissipate, and a 30degC ambient you get (20+0.25+4)*2+30 = 78.5degC as the junction temperature. You really want to keep this less than 50degC.
So for a 50degC junction temp, the heatsink required for 2W dissipation and 30degC ambient will be 5.75degC/W

Thanks a lot. got it

I would not be unhappy with a steady 50 deg C *rise*. So 75 deg C in my cozy 70 deg F room.

LM317 is rated for 125 deg C junct temp.

PRR said:

I would not be unhappy with a steady 50 deg C *rise*. So 75 deg C in my cozy 70 deg F room.

LM317 is rated for 125 deg C junct temp.

Click to expand...

It was 45degC here a week or so ago and that is really toasty!

Still, with the 20degC/W heatsink from above, the heatsink temperature would still be around 70degC, which will result in PCB discolouration and solder joints becoming brittle.
You only have to check out some of the Behringer ADA8000's to see what too small a heatsink can do!

gswan said:

It was 45degC here a week or so ago and that is really toasty!

Still, with the 20degC/W heatsink from above, the heatsink temperature would still be around 70degC, which will result in PCB discolouration and solder joints becoming brittle.
You only have to check out some of the Behringer ADA8000's to see what too small a heatsink can do!

Click to expand...

I don't have my around but 220V seats nice in those 240V units  8)

For the other, leave long legs so the heat doesn't hit so hard the PCB, in extreme cases I've seen crimped wires on the device mounted on the heatsink directly... Mechanical stress could be a problem depending on application, studio gear isn't the most demanding application for that, live gear could suffer a bit more.

JS

So basically, i should be ok-ish with the small heatsink, as long as i keep a healthy distance between the pcb and the heatsink? 125C jonction temp is quiet huge  these things are rought

  Si can take much more, once I got a zener to desolder it self (PTP construction, fell right off) and after desoldering it it worked just fine.

  This small regulators have a temperature protection, 125ºC is where they give they specs, after the protection starts to protect, you don't want to get to that point as it will shut down and you will have no power. Cooler is better, but sometimes you don't need any better, sometimes you do. Get it to work to a confortable point, not to loose your fingerprints if you touch it while troubleshooting is a good thing.

JS

joaquins said:

  Si can take much more, once I got a zener to desolder it self (PTP construction, fell right off) and after desoldering it it worked just fine.

  This small regulators have a temperature protection, 125ºC is where they give they specs, after the protection starts to protect, you don't want to get to that point as it will shut down and you will have no power. Cooler is better, but sometimes you don't need any better, sometimes you do. Get it to work to a confortable point, not to loose your fingerprints if you touch it while troubleshooting is a good thing.

JS

Click to expand...

Thanks always great tips