how much heat can a 78L05 disipate ?

[bountyhunter]

Do you test for thermal cycling? Is there a general limit before the IC is killed?

Yes, it's always tested on a newly released part and occasionally checked on released parts. There is no time limit, the part should never die from either over current or overtemp. The problem is, once in production all kinds of changes get made and things get screwed up.

 

[Hero999]

Is there a reason for the TO-92 package to be different and what's with the wierd pin out for the negative versions, i.e. LM7905, LM337 etc?

Sorry for all the questions, I've always wanted to ask the manufacturers this kind of thing but these aren't the sort of questions they normally answer.

 

[bountyhunter]

Is there a reason for the TO-92 package to be different

Relating to the thermal resistance, the TO-92 is not really "different", it's just that it is plastic. I'll bet a million bucks you will understand what I couldn't get the idiots at National semiconductor to understand:

The term Theta is thermal resistance form one point to another. Why have such a term? to calculate the total temp rise of the die.

Example: use a TO-3 transistor whose theta (J-C) is 2C/W. That means 2C/W from die to the metal flat surface on the bottom of the case.

Now bolt it to a heatsink: theta (C-S) is maybe 0.5C/W from the transitor to the heatsink.

Now the heatsink might be 5C/W theta (S-A) from the sink to the air.

Add them all together and you get 7.5C/W from die to air, which is Theta (J-A)

Theta J-A is all you care about, it's how you calculate die temp given power dissipation.

NOW: how about a little plastic T0-92? How do you attach it to a heatsink? What do you define as the "case" mounting surface when there is none? You can't attach it to a heatsink, you can't solder it down to a heatsink copper layer. Therfore, Theta(J-C) is undefined unless you want to designate a point on the plastic to glue it to..... and the heatsink does almost nothing because the heat flows so poorly through plastic that more heat gets out down the metal leads.

You see why I used to pull my hair out when marketing morons insisted on publishing theta(J-C) numbers on packages where it is undefined. You can only reliably measure theta (J-A) on those. But, as you see given the datasheet, marketing gets it's way when the lunatics are running the asylum. Common sense be damned.

what's with the wierd pin out for the negative versions, i.e. LM7905, LM337 etc?

That's mandatory because the center pin is always connected to substrate. The neg regs are wired differently inside and use NPN power transistors instead of PNP and it flips things around. I recall the substrate on pos regs is always ground, but on neg regs it is the output. The substrate is usually the center pin.

 

[Thunderchild]

I thought the pinout on the positive regs was always I-G-O from left to right looking at the flat face with writing on it and the legs pointing down (Input - Ground - Output).

I might be in error, that's just what I remember. The ground is nearly always center because it is the lead frame internal connection.

well these are exact reverse, see attached data sheet

 

[audioguru]

well these are exact reverse, see attached data sheet

The Taiwan 78L05 has exactly the same pins as a Texas Instruments one.
The pins are reversed on the larger TO-220 package.

 

[bountyhunter]

Looks the same to me: TO-92 pinout is O-G-I from left to right looking at the flat face with legs hanging down on the 78L12 for both TI and the Taiwan data sheet.. Also same on national semi datasheet for the 78L12 in TO-92 package.

 

[Mr RB]

That's what I said, the 78L05 is the funky pinout, but all 78L05 are the same. I did a commercial product using 78L05 a few years back.

No idea why it's reversed compared to 7805, but MAYBE its because the closest thing to a heatsink surface is the flat face, ie maybe thta is officially the REAR of a TO-92?? If you clamp TO-92 to a heatsink plate (which I have seen many times in commercial devices) then the flat face goes to the sink.

Also bountyhunter I love reading your posts about the inside working of NS but I have to disagree that you cant heatsink a TO-92. Like I said above they are often clamped to vertical flat sinks, and for years many TVs used TO-92 (and the taller same shaped package TO-??) on the neck board with circular heatsinks pressed on. They make a huge difference to the amount of heat the package can safely dissipate even with the package's high thermal resistance.

 

[bountyhunter]

No idea why it's reversed compared to 7805

The 78L and 78 use two different die paddles which is what the die mounts on. It's usually because in the higher current ones they align the die to minimize the lead lengths of the bond wires from the input and output leads to keep the wire resistance down. On a low current regulator, the die is smaller, the paddle is smaller, and they turn it whatever way it has to be to go in.

MAYBE its because the closest thing to a heatsink surface is the flat face, ie maybe thta is officially the REAR of a TO-92?? If you clamp TO-92 to a heatsink plate (which I have seen many times in commercial devices) then the flat face goes to the sink.

My point is, it makes no sense to heatsink a TO-92 because the thermal resistance through the plastic to get to the flat surface is huge. That thermal impedance is in series with the heatsink which greatly reduces it's effect. So much so, it makes no sense to believe there is a "designated" heatsink surface on the package because the flow through it is dismal.

Also bountyhunter I love reading your posts about the inside working of NS

It's free therapy....

I have to disagree that you cant heatsink a TO-92.

I have glued them to aliminum fins too when I wanted them to track thermally.... in other words, I wanted them to be the same temp. But, I have no illusion the heatsink does much. As an educated guess: the TO-92 in free air is about 200C/w. With its flat glued to an aluminum block, I would be amazed if it was better than MAYBE 125C/W...... sounds like a big improvement until you realize how bad that is. A standard TO-220 standing in air is about 65 - 70C/W just from the heat radiation it gets from the exposed metal tabs and leads. I never said you could not heatsink a TO-92 or any other plastic package, but it would give very little advantage.

Like I said above they are often clamped to vertical flat sinks, and for years many TVs used TO-92 (and the taller same shaped package TO-??) on the neck board with circular heatsinks pressed on. They make a huge difference to the amount of heat the package can safely dissipate even with the package's high thermal resistance.

They do make a difference if the part is so hot that it's burning up and the heatsink brings it down enough to survive. I don't really put it in the "huge difference" category since a TO-92 in any form will probably never be able to dissipate more than maybe a Watt.

This is simple reality: any package where the primary heat path from die to the air must pass through plastic is screwed. You can heat sink the hell out of it, and it still sucks. This is why Siliconix pioneered "littlefoot" and "lightfoot" packages a long time back which are tiny SMD packages with a package where the metal die paddle is exposed underneath the die and it solders straight down to the PCB. Those suckers can really get the heat out, and in some cases can get as low as maybe 20 - 25C/W. NS and others also have "power tab" SMD packages with a metal tab that solders down. The point is, you have to have short distance from the die to the heat "spreading" area and that distance had better be passing through copper or aluminum and NOT plastic... or the heatsink after the plastic path can't do jack because there is too much thermal impedance getting there.

 

[Thunderchild]

am I going to do any harm by drilling holes before etching ?

 

[mneary]

If you etch first, then the ring around the hole is delicate. Drilling the hole first reduces accidents. But drilling can crack the toner if your drill is dull.

I should comment on the TO-92 heat sink issue: By the time you put any form of heat sink on a TO-92, it's bigger than a TO-220 and it's still less effective. The only reason to do so would be if your labor costs are low, you're really watching pennies, and building large volumes.

 

[Thunderchild]

ok actually got the wrong post but never mind, plan is to drill the holes before putting pen to copper so that I don't mis draw the traces for the IC

 

[3v0]

am I going to do any harm by drilling holes before etching ?

I drill all my holes with a CNC mill prior to etching.

When etching a two sided PCB mask the side you are not etching with box tape or etch both sides at the same time.

3v0

 

[Ghosty_Ghoul]

I've always drilled after etching, I've never thought about doing it before, maybe I'll give it a go.

 

[Thunderchild]

you may want to under drill apparantly though as the etchant will get into the coper and make holes a bit bigger I've been told

 

[Ghosty_Ghoul]

If you drill after etching that dosen't matter, especially if you make the holes on the print smaller than the drill bit.