AMS1117 voltage regulator heat sink connected to OUTPUT, not GND

I made the mistake of thinking all heat sinks are connected to GND, such as the LM7805. Apparently AMS1117 has heat sink connected to OUTPUT. I made a dozen coasters. I have to blame their spec sheet. It has no mention of where the heat sink is connected to in SOT-223 package. TI and a few other companies have clearly marked tab is connected to output.

www.advanced-monolithic.com/pdf/ds1117.pdf

Hi, I agree, missed it by that much, in fact all the packages have their tab at output.
BUGGER...(We can use that term in public in Australia, TOYOTA told us so.)

Tom.... :slight_smile:

That is why there are insulated heat sink mounting kits. They use Mica washers and pads along with plastic washers that present the screw and the side of the hole comming in contact. Heat sink compound is usually used on both sides of the washers.

liudr:
I have to blame their spec sheet. It has no mention of where the heat sink is connected to in SOT-223 package.

OK, but to be truthful, it does specify that for the TO-252 package and it would be rather foolish to presume that the SOT-223 package would differ in such a respect. Since it is all but universal for these various packages (as with the TO-220) that the centre pin is the tab/ substrate, it should be pretty clear. :smiley:

Paul__B:

liudr:
I have to blame their spec sheet. It has no mention of where the heat sink is connected to in SOT-223 package.

OK, but to be truthful, it does specify that for the TO-252 package and it would be rather foolish to presume that the SOT-223 package would differ in such a respect. Since it is all but universal for these various packages (as with the TO-220) that the centre pin is the tab/ substrate, it should be pretty clear. :smiley:

This is where I say I hate people throwing around acronyms without defining them. If there is such a culture of connecting tab to center pin, is it defined somewhere? Do you have to graduate an EE to know? I think not. TI's description is clear and has no redundancy. Another spec sheet defines the tab as pin 4 and states it connects to pin 2 next to the SOT-223 package.

If there is such a culture of connecting tab to center pin, is it defined somewhere?

No it is not defined and no there is not a "culture of connecting tab to center pin".

Thanks Mike. I'll look for where the tab is connected from now on.

I have found that negitave voltage regulators nearly always have a non ground pin for the heat sink tab. I think it has something to do with the way the silicon has to be doped. The heat sink is always connected to the “base” silicon with the extra doping layers on top of that.

Grumpy_Mike:
No it is not defined and no there is not a "culture of connecting tab to center pin".

Check the standard.

First paragraph.

There's a productive amount of paranoia that it pays to employ when using a new
component, "I bet the pin-out is different from the norm" and "lets check the datasheet
from the actual manufacturer, not just any datasheet for a clone/copy".

There's a similar issue with LEDs where one lead is shorter than the other - some
manufacturers make the anode shorter, some the cathode. Easy to get fooled by
that (however there is a rule for all diodes - the cathode is always marked, which
for LEDs means a notch or flat on the LED body on the cathode lead side).

I always check the pinout of SMT MOSFETs in common packages (SIOC8, D2PaK etc)
despite them always being the same to my knowledge, because it takes so little effort
compared to redesigning and re-ordering a PCB.

SOT-223:

TO-220:

TO-252 & others.

Thanks for the reference and pdf book! I'll read it in a few days.

@Paul__B
So you have found some fancy pictures. I suspect you think this proves your point of a standard. It does not. Pins configuration can be and is anything the manufacturer wants it to be.
As MarkT says experience shows it is ALWAYS a good idea to check the data sheet of the actual device you are using. If you have never found any anatomies then you have not had enough experience.

There are two reasons why this picture is not useful.

  1. It does not tell you what the contact on the back of the die is. It could be ground, V+ or V-.
  2. Some MFGs have actually put little BeO insulators between the die and the heatsink to electrically isolate the back of the die, while still providing a good thermal path. I usually see this in a TO-8 can, but it can be done with any package.

Well, Mike will be Mike.

KeithRB:
It does not tell you what the contact on the back of the die is. It could be ground, V+ or V-.

It is not supposed to tell you which connection has which function.

The point is simply - this is how the package is manufactured. It is the standard way of constructing packages which require heatsinking of the chip. It is the most sensible and obvious way to do it. It is possible that a manufacturer might choose to set up a production line to do it differently, but this would represent a minuscule proportion of manufacture.

In fact, the "Iso-tab" product line of TO-220 Triacs is an example of your second point - the tab is connected to none of the electrically active pins. And this product line appears to have almost died if searching on Google is any indication!

Grumpy_Mike:
So you have found some fancy pictures. I suspect you think this proves your point of a standard. It does not.

No, the "fancy pictures" do not automatically imply a standard but do tend to make it clear why this is a standard. I was going more on the quote from the Amkor website that I had pointed out:

This industry standard package utilizes a large tab (lead 2 and tab 4 are common)

A "standard" is what a group - in this case the industry - decides it to be. Individuals (companies) can choose whether to follow the standard - or not. That is not necessarily a problem, but such an exception does not by any means negate the existence of a standard. It is well not to become confused on this essential point.

Grumpy_Mike:
Pins configuration can be and is anything the manufacturer wants it to be.

Yes, either follow the standard, or do something else. This is appropriately referred to in documentation as a "variant".

Grumpy_Mike:
As MarkT says experience shows it is ALWAYS a good idea to check the data sheet of the actual device you are using. If you have never found any anatomies then you have not had enough experience.

Anomalies? (I find a lot of anatomies, thank you very much. No lack of experience here! XD)

Never disputed the need to check a data sheet to see whether a pinout is standard, or variant.