This time, I need something in a small SMT package, capable of handling 10mA or so, and maybe 15 Volts.
Special requirements, though, are: very low forward voltage at 10mA, very low reverse leakage (more so than usual), and fast switching. Low capacitance may also be a factor.
And, naturally, when I'm interested in VF@10mA, Digi-Key's filter is showing, e.g., VF@1.5A, which isn't that much of a clue. Though, a low value at 1.5A probably bodes well for 10mA.
Looks like Panasonic has something that may be suitable, though the capacitance is awfully high.
But then I think: What about germanium? Don't germanium diodes have very low VF?
Well, the venerable 1N34A is still available. For $6.76 each. And not SMT, nor tiny. And VF@5mA is a whopping Volt. Reverse leakage isn't all the impressive, either.
So much for that idea. Back to the regular Schottkies.
Update: A bit of oscilloscope-pokery suggests that the current issue involves both diode drop and not the best choice of coupled inductor. The board that should be turning up in tomorrow's mail uses a coupled inductor that's actually meant for this application (and is a fair bit smaller). The inductor itself should be showing up fairly soon. Then I start pondering synchronous rectification, which is not obviously absurd, though I do need to re-examine the topology and the waveforms involved.
Well, for the immediate project, making this work perfectly isn't all that important. The follow-on, though, will seriously need to pinch milliWatts, so a bit of development effort before crunch time is still in order.
Update 2: An old-school hot-carrier rectifier won't do the job. Not only is a 1B3GT quite bulky, its forward drop at 7mA is three orders of magnitude too high, and it needs (dang, amount not specified in this datasheet) a fair amount of power for heating the carriers.
Meanwhile, my super-easy scheme for synchronous rectification in this topology... um. The waveforms tell me it should work, more or less. Putting it into the schematic, my instincts say it's completely backward. But, my brain has been out of order for a couple of days now, so maybe I should just try tacking together a test, to see what happens. Assuming I have the right bits in my goodie box, of course.
Update 3: Kludged test of concept goes beyond "Fail" and well into "What is this I don't even" territory. Observed waveforms make negative sense. Oh, well: back to pondering. Best guess is that there's a fiddly timing issue (there certainly is one, but I was hoping I could get away with ignoring it) that's causing just enough interference to give the highly-integrated regulator chip heartburn.