We make a gadget that's powered by a 12 volt wart, and sometimes a
user applies 24 and blows one up.
We have a polyfuse and a 12 volt TVS and they sometimes fry the TVS.
It's posssible that most any useful polyfuse+TVS combo can be teased
I think Phil H mentioned some gadget, a polyzorb or something, that
would be better. I can't find it.
We might fine-tune the polyfuse+TVS, or maybe go polyfuse and SCR
crowbar, or something.
This is all entangled with parts availabity. Ideally the box would
just work from 12 or 24, but that has separate complications.
We once used a TI electronic fuse IC, but it liked to blow up.
Tyco PolyZens were amazing, but unfortunately they were available for
only a few years. I think they had to have too many part numbers to
fill all the (current, voltage) space.
It was a pretty slick idea, laminating the TVS and polyswitch together
so that you couldn't fry the one without switching the other.
That's an idea, a pfet current limiter that entirely shuts off if the
input goes above maybe 14 volts. At maybe a 500 mA, fet dissipation
would be OK.
Someone here suggested making our own polyzen sort of thing, but I
suspect we couldn't successfully transfer the heat from the TVS to the
polyfuse fast enough.
An SMB12A TVS is rated for peak power, 600 watts, but not rated for DC
power. 5 watts maybe with big pads?
On a sunny day (Fri, 30 Sep 2022 13:14:03 -0700) it happened John Larkin
<jlarkin@highland_atwork_technology.com> wrote in
Better switcher regulator that accepts up to 24 V
I've occasionally looked at the specialized OVP/power limiter chips, but
they tend to cost $$$, and for my purposes they don't look as good as
doing the limiting locally. (If I were controlling big motors or
something, that might be a different matter.)
At various times I've used LP2951s to make small amounts of +3.3V
directly from the wall wart to run startup circuitry and so forth. An
LM393 is a 358 with different metal--the input devices are lateral PNPs,
and so will take a lot of positive voltage regardless of the supplies.
A few microseconds of delay is no problem for that job.
A few years back the late Jim Thompson of this very bazaar posted a
circuit using two TL431 and two pmos fets to provide series undervolt,
overvolt and reverse polarity protection. All jelly bean very low cost
parts in unlimited supply?
We tried the TPS26600 "hot swap efuse" and it liked to blow up. It's a
horrible complex expensive power-pad thing.
I might just start with an LM2576 and switch down to maybe +7, and
work from there. That will work from 10 to 40 volts. 2576 is an old,
slow, klunky, non-synch switcher that's available. Wonderful part.
We could assume the dpak but we might lead-form the TO220 version in a
pinch. TO220s seem to be available. Maybe nobody wants them.
The guys wanted to use the LT8603 quad switcher but we tested it and
it makes nasty 400 MHz ringy things at the switching edges. In a
"silent switcher" !
We used the LTM8078 dual BGA module and had a similar problem; it
sprayed huge RF bursts all over the board. Probably the same
technology. The nasties are on both edges, so it's just switching too
fast, not an SRD effect.
A polyfuse isn't a bad thing, to prevent blowing traces off a board
worst-case. I'm emotionally opposed to using semiconductors to protect
semiconductors from power blunders.
Poly + diode handles the reversed supply situation OK. I've just got
to handle the overvoltage case somehow.
We only use leaded polyfuses, the ones that look like disc caps. The
surface-mount ones are generally bad news.
Polyfuse + TVS can fail, but the TVS shorts, sort of like the skinny
guy jumping on top a hand grenade. Why didn't everybody just run?
Yes, but in a tangle of wart cords, people sometimes make mistakes. It
is prudent for instruments to be rugged. All our new stuff uses +24.
It's unfortunate that barrel connectors weren't keyed for voltages. Or
That would work, with a polyfuse and a big diode for reverse voltage.
I can get a dpak SCR that will handle 50 amps for a while.
Just a zener to the gate of an SCR is probably OK for modest
short-circuit currents. Big crowbars need a fast gate driver, diac or