Polyfuse selection for MCP1703 LDO

[rbm]

My question is whether or not to design in a PTC (polyfuse) to protect a LDO regulator on my board.

I've designed a Arduino-based circuit that drives an off-board 7-segment display.  The LDO regulator on the Arduino I've selected (Digispark) is too small to supply current for driving both the microcontroller and the associated circuitry, so I've designed in an on-board regulator just to supply current for the display.  I've selected an MCP1703-5003 series regulator to keep the board size extremely small.  This unit has sufficient capacity to supply the necessary current for a common-anode display.

Should I equip the board with a PTC in series with Vin on the regulator to protect it should the user connect the display incorrectly or the output shorts to ground?  Is there internal protection mechanisms in the MCP1703 series that will serve the same purpose?

[T3sl4co1l]

Well, let's have a look at the datasheet and see what we can find.  Anything interesting on the headline page?
http://ww1.microchip.com/downloads/en/DeviceDoc/22049e.pdf

"short circuit protection" and "overtemperature protection" sound good to me.  Are they specified?  Let's dig deeper.

Page 3, DC Characteristics, Output Short Circuit Current = 400mA typical.  Warning: no minimum or maximum specified.  One would hope the minimum is at least the rating (i.e., 200mA), but they don't say.  They do say the minimum maximum (yes..) output current is 200mA, given enough input voltage, so at least there's that.

On the next page, Temperature Specifications says it operates up to 125C, but it's not clear what the 125 and 150C limits mean.

Flipping pages, the Detailed Description provides some insight: both limits are hysteretic, not continuous, so it stutters on and off when pushed past ratings.

As for the polyfuse, it's highly unlikely that it would ever go open circuit, and that it would remain open during the fault -- the thermal time constant of those things is disgustingly massive (100s of milliseconds), much longer than the thermal time constant of the LDO itself (which will go into thermal shutdown if overloaded).  And an overload condition will only occur in that narrow sweet-spot between 200mA (the regulator's rating) and "about 400mA", beyond which, the current limit causes it to shut down momentarily, so the fuse will never build up enough heat to open.  In reality, that limit is probably more like 300-600mA -- LDOs are notoriously sloppy on their current limiting.  But even that is probably not enough.

If you needed to protect a much smaller amount of current, like 50-100mA, a polyfuse might be applicable.  The regulator will then be capable of delivering enough excess current to open the fuse, at least under some circumstances.  Or if you didn't need a regulator at all -- taking the raw input (which I presume comes from a DC adapter or something), if it has "unlimited" current capability (usually meaning... no active current limiting), a fuse would be desirable to prevent damage to the adapter or the wiring inbetween.

Which, by the way, is something to consider -- if your LED displays are supplied by resistors, you can run them from unregulated supply voltage, by simply using larger dropping resistors.  If the anodes are switched (as in a typical muxed LED display), you would need level shifting logic though.

Polyfuses are interesting devices, but most often, the interest is because of the challenge in applying them correctly.  As with all fuses, they should be used only to protect the wiring.  They open very much more slowly than proper single-use fuses, so it is a difficult challenge to absorb the excess energy while protecting a circuit, for example against incorrect supply voltages or surges.

Where circuit protection is required, you're very much better off using a source controller type IC -- the LDO would actually serve this purpose quite well, though it's probably not very reliable (temperature cycling around rated limits is not very healthy for anything).

A purpose-made load controller, like the power switches used for USB outputs, can do a fine job, both limiting (regulating) the current under fault conditions, and disabling the output under fault or overtemp conditions.  Of course, you still need a regulator to set the correct voltage; these devices don't also regulate voltage.

And, there's always the question of suitability: the choice of an LDO assumes you *need* low dropout -- Why do you have a 6.0V supply?  The Digispark board doesn't use an LDO, it uses a 78M05 (an old fashioned, "HDO", so to speak).  Which means it won't be useful below 7V input, assuming you're using external power of course.  If you're drawing from the same power source, then you can use a (cheaper, better behaving) HDO, like the 78M05 (which is good to 500mA or so, given enough heatsinking), and you won't have to worry about all this stuff.

You could very easily use the Digispark's onboard regulator itself, except that without any heatsinking whatsoever, it's probably not good for more than 100mA at modest supply voltages (say from a 9VDC adapter).  Heck... if you have some hunks of copper around, you could solder a heatsink onto the tab, and increase its dissipation rating that way.  Ugly but it'll work!

Tim