[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Power distribution across an S100 Board and voltage issues with TO-3 Regulators..

On Tuesday, June 3, 2014 2:01:57 AM UTC-5, monahanz wrote:
Just yesterday I got 2 "LM323K's (5V, 3A) from Unicorn. One put 6.2 volts on a test board!!!!   I had others from Jameco that put out 4V's.

Yeah I've had DOA 323K regs too.  Annoying isn't it?  Now I test my regs before soldering them in.
Mouser stopped carrying the LM323's. If anybody has a good source to name brand 78H05's or LM323's please let us know.

STMicro and TI and all those guys stopped making LM323K ranging from a very long time ago to recently.  Its been NOS since then.  TO3 regs are dead and never coming back AFAIK.  On the other hand there's multiple sources selling LM323T which are pretty much the same regulator die in a TO220 package.  Same 3 amp rating and same electrical characteristics and everything.  Also unlike the (counterfeit?) TO3 parts I've never seen a DOA 323T.  At least not yet.  In fact other than turning LDOs into oscillators I've never even managed to kill a TO220 reg even with "abuse" like soldering the HS tab to the PCB.

Yes, TO-3 LM323K are essentially unavailable for me so I now always wire in a TO-220 LM323T.  Looks ugly, works great.  Same 3 amp rating, in fact its basically the same reg in a different package.
I'm reluctant to go with two L7805ACV's TO-220 (5V 1.5A) because they use up so much board space with their heat sinks.

Yeah don't use a 1.5 amp part when a 3 amp part costs like 50 cents more.

If you pull the datasheets the "only real difference" between a LM323T (TO-220) and LM323K (TO-3) is the junction thermal resistance.  2 C/W for the TO3 and 3 C/W for the TO220.  I don' t think it can matter much because as a linear it can't be dumping more than 3 watts per amp, so even if you drew the rated 3 amps, at 8 V - 5 V equals delta 3 V times 3 A equals 9 watts, that means the junction resistance of the TO3 will be 18 degrees hotter than the package and the TO220 will be 27 degrees hotter than the package, so as long as you can literally keep the package from boiling water, in both cases, you're all good nothing to worry about.  TO-220 heatsinks are slightly higher availability and cheaper and about the same C/W ratings too.

Heatsinks being how they are, you might be surprised to discover that sometimes for some topologies and applications its cheaper to use "several" small regs with practically zero size heatsinks, than to run everything off one monstrous beast of a heatsink.  Heatsink size and price do not in any way scale linearly with C/W so as a rule of thumb if you're specing a heatsink that costs more than twice the reg costs its cheaper to just install two regs with tiny heatsinks.  Of course that results in arguments about the cost of PCB per inch being even more expensive than the cost of heatsink per sq inch of PCB.

The TLDR is just hot wire in LM323T and call it good, works for me.