5

I assembled my first PCB and started measuring everything to validate that it is correct.

When it was the turn of the a TO-220 linear regulator (LD1117V33C) I got surprised to see 3.3 V in the heat pad.

Is this normal? If yes... Why is that? When is this useful?

Enter image description here

This are my schematics:

Enter image description here

And this is how I placed it in the PCB:

Enter image description here

Datasheet: https://eu.mouser.com/datasheet/2/389/cd00000544-1795431.pdf

Full PCB schematics and circuit: https://github.com/crgarcia12/electronics-homeassistant-lightscontroll#version-2

Carlos Garcia
  • 1,116
  • 1
  • 7
  • 25
  • 3
    Where does the datasheet say the tab is connected? It's not unusual for it to be connected to the hottest running part of the device, for hopefully obvious reasons. –  Nov 22 '21 at 21:54
  • 1
    Is that first image right? It looks like it's highlighting pin 1, which would be GND. – ilkkachu Nov 23 '21 at 10:33
  • @ilkkachu exactly, I have 3.3v between the thermal pad and GND – Carlos Garcia Nov 23 '21 at 11:06
  • @CarlosGarcia, right, I read it as pointing to two places that are both at 3.3 V (against the ground by default). – ilkkachu Nov 23 '21 at 11:08

4 Answers4

13

See page 6 of the datasheet you linked in the question:

enter image description here

You'll find that in most cases, TO-220, DPAK, D²PAK, TO-263, and other packages with thermal tabs tend to have the tab connected to the center pin. The reason for this is simple: that's just how they make the leadframes, and if there's no good reason to connect the tab to something else, you wouldn't want to spend the extra money on a custom leadframe.

Hearth
  • 27,177
  • 3
  • 51
  • 115
  • Thank you for the explanation! I did not even know that "leadframes" existed :S ... learning electronics by myself make me miss a lot of interesting details in the way – Carlos Garcia Nov 22 '21 at 21:36
  • 5
    It's unfortunate the designers didn't use pin2=tab for ground (as on a 78M05 or any other sensibly pinned out part), as it is quite common to want to use a grounded chassis as a heat sink. As is, you'd have to add an insulator in order to heat sink to a chassis. – td127 Nov 22 '21 at 21:53
  • 1
    @CarlosGarcia The leadframe is the metal frame that contains all the metal parts of the package; the actual semiconductor die itself is attached to the leadframe and tiny wires used to connect each of the leads to it, before an epoxy is poured around it to form the black part of the package (at least in the case of plastic packages). The part of the leadframe that the die is attached to is generally connected through to the middle pin. (this is also why power MOSFETs almost always have the drain as the middle pin and tab; the back of the die is the drain contact in a VDMOS.) – Hearth Nov 22 '21 at 21:53
  • 7
    @td127 I imagine it's related to the fact that the drain of a VDMOS is on the backside of the wafer; the output of an LDO regulator is the drain of the pass transistor. This is not the case in a conventional, high-dropout regulator, which is why pin 2 of the 78xx series *is* ground. – Hearth Nov 22 '21 at 21:56
  • 2
    I'd always imagined that the tab was connected to the drain/collector of NPN devices because that's the part of the transistor that gets hot. In the case of a positive voltage regulator, the out pin is the drain/collector of the pass device. – Supa Nova Nov 22 '21 at 22:13
  • 2
    @SupaNova I can't speak for BJTs, but I know that for modern power MOSFETs (VDMOS and derivatives), it's because the drain contact is the backside of the wafer, and the tab is the same piece of metal that the die gets attached to. The part that gets hottest is actually the channel, on the top side of the wafer, and there's efforts in the power electronics world to build modules with what they call "double-sided cooling", where you also have thermal pads attached to the source contacts (which are physically near the channel). – Hearth Nov 23 '21 at 01:39
  • 1
    @SupaNova Thinking on this again, the reason that the tab is connected to the output of the 1117 may well be because that's the part that gets hottest--I expect the pass transistor gets far hotter than any other part of the device, and while it's not an option to bond it close to the channel (which has to be on the top of the wafer, which has to have multiple contacts that can't be shorted together, so you can't practically attach it to a big metal slab), putting it on the drain of the pass transistor makes sense for power dissipation. – Hearth Nov 28 '21 at 02:46
5

Yes, the tab is connected to middle pin which is Vout.

This is mentioned in the datasheet you linked as well.

This does apply to other TO-220 packages as well, so it is not unique to LD1117.

Justme
  • 127,425
  • 3
  • 97
  • 261
5

Page 6, note under the pinout:

The TAB is connected to the VOUT.

The DPak even omits a normal vout pin and uses the tab exclusively for voltage output.

No real rhyme or reason to why some devices have the tab at ground or not. It's all up to the initial engineers really. So don't assume it's always at one potential or another.

Passerby
  • 72,580
  • 7
  • 90
  • 202
2

Usually the leadframe and tab metalwork is directly connected to the bulk silicon of the die. This is

  • the easiest way to make the part and
  • offers the lowest thermal resistance from the power elements to ambient.

These two points together mean that you get the cheapest, best thermal performance, power device.

If you have to connect the tab to a heatsink then you can

  • use an isolated heatsink or
  • use electrically insulating thermally conductive hardware to attach it to a grounded heatsink

Some power device manufacturers take pity on designers who don't like dealing with live tabs, and provide a layer of electrical insulation inside the device. These parts sell at a premium however, and don't have quite the same power handling as their live tab counterparts.

Neil_UK
  • 158,152
  • 3
  • 173
  • 387