This is a documented feature of the DS18B20 digital
temperature sensor. They call it "parasite power."

Allen

> -----Original Message-----
> From: piclist-bounces@mit.edu [mailto:piclist-
> bounces@mit.edu] On Behalf Of David Meiklejohn
> Sent: Monday, February 2, 2015 7:59 PM
> To: 'Microcontroller discussion list - Public.'
> Subject: [EE] ICs working with Vdd/Vcc disconnected?
>=20
> I've learned something while creating a test procedure for
a
> product I've
> been developing - and written it up (with pictures and
> schematics) here:
> http://www.gooligum.com.au/blog-section/blogart13
>=20
>=20
>=20
> Short version:
>=20
>=20
>=20
> I assumed that, if an ADC chip (MCP3008 in this case)
passes its
> tests
> (samples an input and returns a result in a defined
acceptance
> range, a few
> times in succession), it's safe to conclude that, not only
in that
> chip ok,
> but it's also being powered and that therefore the 3.3V
rail that
> it's
> sitting on and therefore the LDO regulator and upstream 5V
> power are also
> ok.
>=20
>=20
>=20
> "Ok" in this case meaning "there is one".  Sure, it's not
an
> exhaustive
> power supply test by any means.  Sure, maybe the supply is
> marginal.  Maybe
> it would fall over if pushed.  Maybe it's supplying the
wrong
> voltage.  But
> this was supposed to be a final acceptance test on a known
> (more fully
> tested) design - a "go / no go" before each board leaves
the
> factory.  If
> there's a power supply there - enough to power the ADC -
then
> presumably the
> regulator wasn't let off the board, it's not DOA, the
micro-USB
> connector
> (from which 5V comes) wasn't incorrectly fitted, etc.  The
idea
> is to check
> for things that might have gone wrong in PCB assembly, or
> major component
> failures.
>=20
>=20
>=20
> In other words, I assumed, if the ADC works, there must be
> nothing majorly
> wrong with the power supply.
>=20
>=20
>=20
> But in fact - I found, to my initial amazement, that the
ADC
> worked just
> fine, with no power supply at all!  I could remove the 5V
> supply, with no
> apparent effect.  No power coming in at all, and it still
worked.
>=20
>=20
>=20
> To prove this, I tried making up a basic circuit with a
MCP3008
> connected to
> the SPI bus, a couple of resistors to provide an analog
voltage
> to measure,
> and no connection to Vdd at all.  Didn't work (as it
shouldn't),
> but adding
> a 1uF cap across Vdd/GND, and still no power supply in
sight,
> made it work
> just fine.
>=20
>=20
>=20
> Eventually I figured it out - for more detail read the
blog, but
> anyway -
> enough power was supplied via a digital input (normally
held
> high, but goes
> low for each conversion) to power the rest of the chip via
(I
> assume) the
> internal protection diode on that input.  3.3V into an
input led
> to 2.5V
> being supplied out of (yes, out of!) the Vdd pin, charging
the
> cap, which
> was then able to hold the chip up for the short period
that that
> input
> dropped.
>=20
>=20
>=20
> And this seems to be common behaviour - a FTDI FT232RL on
> the same board did
> the same thing, powered quite happily from its RXD input,
and
> supplying
> power out of (!) it's Vcc pin.
>=20
>=20
>=20
> Obvious now that I see it, and maybe obvious to most folk
> here, so sorry if
> that's the case.  But I'd never observed this before -
> presumably because
> I've always applied power to any ICs I've used.  Going
back to
> university
> days, one of our mantras when starting EE pracs was "power
> to the chips!"
> Ah, but it seems that they don't always need it...
>=20
>=20
>=20
> Cheers,
>=20
> David
>=20
>=20
>=20
> --
> http://www.piclist.com/techref/piclist PIC/SX FAQ & list
> archive
> View/change your membership options at
> http://mailman.mit.edu/mailman/listinfo/piclist

--=20
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.