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

Re: [N8VEM-S100:4642] 8080 CPU - Heat is ON - Wet Finger Test



On Sunday, July 20, 2014 9:36:30 PM UTC-5, Crustyomo wrote:
Since mine is putting out 8.9V, I wonder how much difference it would make at 8V.

Linear regs are all the same thermally no matter package, or traditional/LDO, so assuming 1.5 amps constant (unlikely) and a small cheap 24 C/W heatsink (the tiny U shaped ones, can't buy any smaller)

24 * 1.5 * (8.9 - 5) = 140 C above ambient.  You'll be lucky to get 5 C/W off an old style TO-220 reg for tab to junction temp although the 3 amp rated LM323T get down to 3 C/W if I recall.  Anyway its hot enough the reg won't live long if you're drawing 1.5 amps so you're probably not drawing 1.5 amps.

If you turn the voltage down to "standard"
24 * 1.5 * (8 - 5) = 110 C above ambient.  Still smokin'

If you crank it down into barely regulated 7.5 depending on device drop
24 * 1.5 * (7.5 - 5) = 90C borderline but it'll probably live.

I use the 10 C/W heatsinks that look like flipped over centipedes but the "legs" toward the edge of the card need some trimming, especially when you mount a TO-220 on a board laid out for a TO-3.  Thats border line for a 3 amp TO-220 reg at the full 3 amps if you run the numbers.  But I don't run constant 3 amps anyway.

The real benefit of the modern TO-220 regs like the LM323T is the junction/tab thermal resistance is way lower.  So hook up an old fashioned 7805 at 1 amp to the same heatsink as a modern LM323T on the same temp heatsink at 1 amp, and the junction is going to be quite a bit cooler / more reliable.

Most of the rep that TO-3 regs have for running cool is just larger heatsinks.  Note that TO-3 heatsinks at 1.5 C/W (thats one point five not fifteen) do exist but they're the size of a lunchbox.  TO-3 junctions do run cooler than TO-220 junctions but you can't test that with a finger that is a bit more complicated to measure.

Reading between the lines from some surface mount stuff I've done soldering a TO-220 to the PCB is probably around 15 or so C/W.  Its a lot bigger than the DPAKs I've soldered and international rectifier claims they'll do better than 20 C/W when they're soldered to the board, and dpaks are kind of the SMD spiritual successor to TO-220 (err, sorta) so I think that reasonable.  So soldering to the board works better than the cheapest of all heatsinks, but even the cheapest heatsink is easier to desolder and replace if necessary.