Is pcb mounting the only option for a D2pak? Like this one
Is it possible to clamp the package to a bus bar?
Any thoughts?
Cheers
Is pcb mounting the only option for a D2pak? Like this one
Is it possible to clamp the package to a bus bar?
Any thoughts?
Cheers
Sure, why not? We did something like that...
ftp://jjlarkin.lmi.net/Amp.jpg
John
"John Larkin" wrote in message news: snipped-for-privacy@4ax.com...
Yea, that the concept I was thinking, problem is it?s a 7 pin device.
1-gate 5-source and pad as Drain. I want to mount the pad on a buss bar( er clamp) and use screws into a second bussbar from the source pins. The question is whether the pad can conduct to the bussbar when its clamped or do I need to flow solder, then it becomes a 2 piece Buss bar, which I'd rather not do.Cheers
Wow, impressive. What is that thing doing?
Note that John's DPAK devices seem to have the drain additionally piped out at pin 2. Most FETs have that pin snipped so then there is only the bus bar contact. Meaning you'll have to deal with dissimilar metal issues, corrosion, etc.
Can't use TO220?
-- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
It's the output stage of a 120 amp, +-180 volt MRI gradient coil driver.
I'm using TO-247 types. The drains of all 32 fets (half Ns, half Ps) are clamped to the nickel-plated copper heat spreaders. The heat sink is "hot" to the output.
John
can't you mount it on a pcb and clamp the bar on top of the pcb?
extend the pad for the tap, clamp bar on top of that, extend a big pad from the sources clamp a bar on that. gate goes through via to backside.
-Lasse
"John Larkin" wrote in message news: snipped-for-privacy@4ax.com...
But are they heat spreaders or Bussbars? Where is the bulk of the current flowing, thru the pins or the heat spreaders?
Cheers
The drain connections are actually made through the center pin to the pc board. I don't know that I'd like to depend on the drains to electrically connect to the heat spreaders alone, at least without some serious consideration; we use filled silicone grease and the transistors and the spreaders are very flat.
The big pcb copper pour that all 32 drains connect to is also conductive to the heat sink through all the mounting screws, which probably helps some.
The big problem here was thermal, not so much electrical.
John
That?s the look of it, 4 oz copper, maybe 8. I have to shunt 1-2KA for
100us or so.
Yea and Stake the hell out of D and S with Vias. Too bad I don?t see anything good in a TO-277 package, That I could work with. The D packs are cheaper than the 277's. Takeup less space. And the Rdson will be lower for a half dozen or so devices.
.
or
I am facing same problem (using Infineon IPB030N08N3 G, two in parallel). I use custom made Cu block to which these MOSFETs are soldered. The tinned side of the block is bolted to the huge pad on the PCB. It is OK for our potential volumes (~1k/year). FYI. I tried to talk to Infineon FAI to find out if/how they solved this problem (they've got to test these MOSFETs, don't they?). The guy seemed to be the most incompetent engineer I have ever encountered (the concept of thermal resistance seemed to be very confusing to him). Keeping expectations high does not always help, what else is new?. ;o)
Thank you for the link to IRF - somehow I managed to miss IRFS3107-7PPBF, it looks similar to the Infineon IPB030N08N3 G, which seems to be of "unobtainium" variety.
.
or
I need to have heaviest possible copper on my 4-layer board. I have MSOP packages. I know that I cannot have deep copper "canyons". How deep is "deep". Can anybody suggest a document describing how to design PCB with SMT parts and thick Cu? Does it cost a lot to have different Cu thickness on different layers?
How does that affect reliability of the devices? Soldering them directly to the big massive copper bus must involve slow heat-up and cool-down time, not to mention pretty high device temp in between.
Fried any fets that way?
Mark
- snipped-for-privacy@o13g2000vbl.googlegroups.com...
probably not with in spec, guess you could use some low temp solder something that melts below the max 175C storage temp.
-Lasse
I envisioned soldering the devices to copper foil, then clamping them to the heat sink. This may work, I have to go thru the numbers on electrical resistance.
Cheers
heat sink.
Sounds nasty and not elegant.
M- snipped-for-privacy@o13g2000vbl.googlegroups.com...
I am in the prototyping stage, can't say much about reliability. The Cu block is not THAT big (1"x0.7"x0.15" or so).
snipped-for-privacy@news.eternal-september.org...
snipped-for-privacy@news.siol.net...
snipped-for-privacy@o13g2000vbl.googlegroups.com...
ly
o the heat sink.
Nasty - am not sure. I have seen many things that are so much nastier. Not elegant - OK, I'll take it... Can you suggest better solution? I love how manufacturers list 100W+ MOSFET power rating. All I have to do is to provide 0.0degC/W heat sink. And this is for SMT part with tab being sole drain connection.... I just got an answer from a Infineon app manager. My best standard option (aside from Cu block) is Al substrate board (doesn't it mean single layer SMT-only?)
messagenews:he23nl$ej5$ snipped-for-privacy@news.eternal-september.org...
messagenews:NR_Mm.15115$ snipped-for-privacy@news.siol.net...
messagenews: snipped-for-privacy@o13g2000vbl.googlegroups.com...
the heat sink.
No.
messagenews:he23nl$ej5$ snipped-for-privacy@news.eternal-september.org...
messagenews:NR_Mm.15115$ snipped-for-privacy@news.siol.net...
messagenews: snipped-for-privacy@o13g2000vbl.googlegroups.com...
heat sink.
Well, I think copper foil that's sufficiently thin to avoid problems with overheating is going to warp, keeping it flat will be a problem.
Yes, a better solution is what I'd like to see, too. These parts are not easy to cool.
MElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.