Attention, highly non-political post.
Here is a puzzler. This ST datasheet says on page 5 that the junction to ambient temp rise is 55C/W for SO8 and 115C/W for SOT89:
This TI datasheet for the positive regulator states pretty much the opposite on page 4:
One is a negative and one a positive regulator which shouldn't make much of a difference as I assume they calculated with similar copper. So it seems one of them may be wrong. Question is, which one?
Looks like they heat-sunk the SO8 but maybe didn't the SOT89.
Unless the SO8 has a power pad, the SOT89, soldered to some copper, should win.
I've dissipated 3 watts from a SOT89 soldered to a copper pour.
Thermal vias could make this better.
There's probably a die-size difference between positive and negative regulators... since good PNP's aren't native to a P-substrate process.
Don't ask me to calculate... I've reached my energy-expenditure limit for the day... time to continue with our viewing of the five-seasons of "Boardwalk" ;-) ...Jim Thompson
That would be a major flaw in a datasheet. It's just puzzling that the data is almost exactly opposite between the two.
Datasheets these days ain't what they use to be. I just had to fire off another support request because the AD8337 datasheet merely says to connect the pad "to the PCB". Great. To GND? The negative rails? They forgot ...
All they have on the SO8 is that the four center pins go to the respective plane (GND on the positive regulator and VIN on the negative one).
Good words of wisdom. I'll stay with SOT89 then. That is what I used before and while they do become toasty it seems SO8 sans pad won't be better.
/L79L.pdf
both the negative and positive use NPN output
Doesn't make much sense to have a junction to ambient thermal resistance _lower_ than the junction to case, does it?
I think they vertically transposed the top right two cells in the grid on page 4 of the ua78l datasheet
Someone could maybe make a business out of proofing data sheets.
Why pay for it when one's customers apparently so willingly do it for free? :D :D :D
Sziklai-style. ...Jim Thompson
But the errors are rarely fixed, after years and decades.
Maybe they let some intern do that.
That's why people don't even report it anymore.
Company X claims their datasheets are perfect even when they're not, Company Y accepts corrections from users and makes regular fixes.
Customer looks thru the list of all the bug-fixes on Company Y's revision page and thinks "Ugh, these guys don't know what they're doing." Company X smiles and say "We stand by the accuracy of our documentation."
User thinks "Great! I'll go with Company X." User encounters a serious problem in Company X's documentation. Already being invested in the choice, user works hard to rationalize the mistake to actually being their own fault. Company X stands by the accuracy of their documentation, after all.
Alternative scenario at yours truly: Company X gets blacklisted and remains on that blacklist for a few decades. They won't ever know how much revenue went elsewhere because the user never communicates with them anymore.
L79L.pdf
Explained by two footnotes--
"(1) Our SO-8 package used for Voltage Regulators is modified internally to have pins 2, 3, 6 and 7 electrically communed to the die attach flag. This particular frame decreases the total thermal resistance of the packa ge and increases its ability to dissipate power when an appropriate area of copper on the printed circuit board is available for heat-sinking. The external dimensions are the same as for the standard SO-8."
"RthJA Thermal resistance junction-ambient (Max) 55 (1) 200 115 ? ?C/W
==> Notes: (1) ***Considering 6 cm^2 of copper Board heat-sink.***" < ===
Cheers, James Arthur
o
S/L79L.pdf
h
it might be possible, if the junction to plastic case is seen in isolation and the junction to ambient is solder to some area of copper
That's fine but an apples to apples comparison will also provide 6cm^2 for the SOT89. Nobody in their right mind would mount them without serious copper area if heavily loaded.
[...]
to
FS/L79L.pdf
lly
lag.
package
ea of
he
**" >From ST's numbers, the SOT89 with the same heatsink would have
It's hard to beat a solid copper tab.
Cheers, James Arthur
I only see 115C/W for the SOT89 in table 2.
How much power do you need to dump?
Not sure yet but probably 500mW or a little more per device. Things shouldn't get too hot to touch though. It is also a sort of hi-rel application where the product lifetime should be several decades.
Have something to add? Share your thoughts — no account required.
Ask the community — no account required