SO14 package at 154C/W? Really?

If Tj max is 150C, and if Theta(ja) is really 154C/W and you're dissipating

150 - 0.500*154 = 73C

It'll be effing hot, but it should work.

Also, some of the thermal power will go through the leadframe and into your pcb, so the die temperature won't really be as high as the Theta(ja) calculation suggests, because (iirc) Theta(ja) doesn't include this path.

Bob

Yeah, and I don't like things getting effing hot :-)

Are you sure about that? It would not be too useful to publish data for a situation that is impossible in practice. I mean, an opamp needs to be soldered down before it can do anything.

It *is* useful, and nothing about it is impossible in practice. It's difficult to characterize the heat conduction through the lead frame, and not including it in Theta(ja) makes the formula more conservative. If you do the design based on Theta(ja), you'll have a slight bit of additional margin.

Anyhow, I'll go with single amps in SO8 then. More conservative, I don't want anything to get hot.

Actually, one of the few situations where it is possible, in practice, to characterize the thermal impedance of an IC (in a specific package), is when you are the manufacturer of that IC. All they need is a standard panel template for the mounting arrangement in question. That is where the published data should originate.

Normally, to ensure a reasonable MTBF, you'd want a pre-established margin between predicted operating extremes and specified junction limits.

Everyone has their own opinion when it comes to characterization and employment of thermal data - so it's worth while tracking down any available info on the specific part in question, should the issue become obviously relevant. Larger packages, or packages with many pins (~ more than six) become increasingly predictable, for a fixed construction.

It is my understanding that the thermal impedance test methods outlined by jedec (JESD051) include mounting hardware for surface-captive components, whether surface-mount or through-hole. TI app notes below reflect this.

Check out:

Nat Semi AN336 Nat Semi AN1028 SOT223 TI SZZA017A - board trace effects TI SCAA022A - board trace effects TI SLMA002 - enhanced packaging FSC AN1029 SO8 FSC AN1025 SOT23

There are on-line tools at some mfr's web sites. eg:

RL

Thanks! Pretty cool tool. Kinda sez to me not to use SO14 ;-)

Epoxy a heat sink on top. I have a board right here on my desk that has a heatsink on an FPGA, and two more on some dual dacs.

ftp://66.117.156.8/DSC01786.JPG

Or put a power pad below on the pcb layout, some thermal vias to a plane, and use a dab of epoxy between the part and the board.

John

A heatsink glued onto a BGA? Wow, that takes guts.

Nah, I'll go to singles but right now I am trying to knock down the power via some inductive kicker tricks. It's a HV switch bank where these opamps have to hold programmable biases. If I can shave off another 500uA/channel I might just be back to a quad package. Unless there is a really compelling reason like I am sure there is in many of your cases I try to keep everything cool, to the point where you can operate without heatsinks yet touch everything with screaming. Well, ok, one should not touch this high voltage stuff ...

Guts? Why? It's soldered to the board in 456 places, which should be pretty stiff.

I've got used to touching everything to see how hot it is, or to finagle suspected RF oscillations. And get occasionally blistered/shocked for the trouble.

John

Yeah but if that number is reduced to 455 places some grief could set in :-)

So do I. But if I get a blister I want to know why and what can be done about the circuit to chill it.

Reminds me of a sad case. A very experienced power engineer guided a tour for some bigshots. He was a neat freak. The whole place looked spiffy, not a speck of dust. Then, a fly landed on a distribution rail. Instinctively he wanted to flick it away. That was the last millisecond of his life :-(

Are those commercial heatsinks? They don't look as nice as the rest of the board. I wonder why they didn't machine square or rectangular posts rather than round.

Is that LPI 'silk screen' ?

Nice.

Is 'MS' your layout guy/gal?

Best regards, Spehro Pefhany

Die cast, I think. They are from Cool Innovations, who are up near you. What I'd really like is anodized colors.

Dunno, but the ref desigs are 60 mils high, and are pretty readable. That doesn't always happen.

Mike. He does our pcb stuff and Autocad/Solidworks stuff. He's good, so I won't give his last name.

Incidentally, this is rev A of the board, and it works: no kluges.

Lower-left is where the Xport ethernet brick goes. We press-released it and got a lot of hits, so you may see it in a mag or something.

Hey, google T346

This is amazing. Google has changed the whole game.

John

Sure, but they don't know what you will have the lead frame soldered to, how well it will be soldered, etc.

Not particularly. Once the BGA is soldered down, there's nothing special about attaching a heatsink to the top of it.

Joerg wrote:

You'd have to be doing something extremely wrong in order for epoxying a small heat sink to a properly soldered BGA to result in any of the bonds failing. Are you applying the heat sink with a hammer or crowbar or something?

They don't have to know what you will use. They just have to tell you want they used.

I'm pretty sure I've seen info like that is some app notes.

No, but I have seen too many BGA failures. Not in my designs because (with one exception) I never used BGA.

We just use regular epoxy, gently squished down.

Opposite experience. We've used hundreds of BGAs, mostly FG456's, nearly all placed and soldered in-house, with exactly 100% success.

The only one we've ever removed and replaced was a couple of weeks ago, and that was a mistake. There was a power plane short, nobody could find it, so somebody decided to remove the BGA. NEEP! That wasn't it. I dumped a 6-amp power supply into the plane and hit it with the Flir imager... sure enough, a ceramic cap glowed in the thermal IR... a simple solder bridge that nobody spotted.

We have a lot more trouble with TSSOPs and other fine-pitch leaded parts. US8's are the pits. One nice thing about BGA solder joints is that you can't inspect them, which saves a lot of production time.

It's liberating to have 200,000 gates and 300 i/o pins, in about a square inch, at your disposal.

John

:-)

Same here, MLFs and QFNs are just as difficult to deal with in production lines. We are skipping packagings all together. Chip On Board (COG) die bonding is just as easily to order as anything else, but usually 1K minimum.

It's even better to be able to define your own footprints. We are using a 68 pads die with 32 pins in one project and 44 pins in another. They are just die bonded differently.