Negative Voltage Load Switch

[Al Clark]

I have an application where I want to turn on a bipolar +/-15
volt supply from 3.3V. The supply already exists, I am
disconnecting a part of a circuit to save power during an idle
state.

Integrated load switches comprised of P channel and N channel
enhancement mode FETs are readily available for positive
supplies.

I also know how to achieve a negative voltage version with FETs
using discrete components.

The parts are not expensive, but there are quite a few of them.

Does anyone know of an inexpensive integrated solution? Ideally
this would be an 8 pin part with a logic level enable and both
positive and negative supply switching. In my case, I don't need
to switch more than 100mA. Alternatively, I might consider just
a negative voltage version (combined with a standard load
switch).

Thanks

Al

[edward....@gmail.com]

A 8 pins relay, can switch 2A on + and - power rails. Only need a transistor to buffer and drive the coil.

http://www.digikey.com/product-detail/en/EE2-5NU-L/399-11017-6-ND/4506558

[Rob Gaddi]

A pair of 4 pin SSRs? I know we're paying about $0.70 a pop for
CPC1008Ns, which I think would cover your problem.

Alternatively, if you're generating the +/-15 rather than taking
it, could you shut down the supply generation itself, either with a
shutdown pin or by cutting the power to it?

--
Rob Gaddi, Highland Technology -- www.highlandtechnology.com
Email address domain is currently out of order. See above to fix.

[Dave Nadler]

On Wednesday, October 1, 2014 6:59:28 PM UTC-4, Rob Gaddi wrote:
> A pair of 4 pin SSRs? I know we're paying about $0.70 a pop for
> CPC1008Ns, which I think would cover your problem.

Another vote but I've used CPC1018 (slightly more expensive
but lower turn-on current, not that CPC1008 take so much).

Hope that helps,
Best Regards, Dave

[Tim Wescott]

Two transistors and two resistors is "a lot" for the negative voltage part
of it?

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

[Dave Nadler]

On Wednesday, October 1, 2014 6:19:36 PM UTC-4, edward....@gmail.com wrote:
> A 8 pins relay, can switch 2A on + and - power rails.
> Only need a transistor to buffer and drive the coil.

And if you use a relay, don't forget the diode!!

Seen the results of that a few times ;-)

[Al Clark]

Tim Wescott <seemyw...@myfooter.really> wrote in
news:qI2dnQPhAuXwLbHJ...@giganews.com:

My circuit took three fets for the negative supply switching
(some can be BJTs. This was so that on was active high.

Otherwise my positive supply is the wrong polarity.

So my overall circuit uses 3 cheap FETs and a Dual N/P channel
lower Rds type (IRF7209 for example)

What do you use for 3.3V control with two transistors?

Al

[Ian Malcolm]

Al Clark <acl...@danvillesignal.com> wrote in
news:XnsA3BAABC56D77Bac...@69.16.179.21:

Can you stand to drive the -ve rail switch off the +ve rail coming up?

Hang an 18V Zener off the +ve rail to the gate of a N channel MOSFET
switching the -ve rail. Add a gate-source pulldown resistor to keep it
hard off when the +ve rail is off.


--
Ian Malcolm. London, ENGLAND. (NEWSGROUP REPLY PREFERRED)
ianm[at]the[dash]malcolms[dot]freeserve[dot]co[dot]uk
[at]=@, [dash]=- & [dot]=. *Warning* HTML & >32K emails --> NUL

[Tim Wescott]

The upper resistor limits the current going into the emitter of the PNP,
and thus the PNP collector current. The lower resistor helps to turn off
the NPN -- you could leave it off if you wanted to be really sleazy.

You do need to size the upper resistor to flow enough current to really
turn the lower transistor on: I suppose that if you don't have a pin that
can source several this may not be the way to go.

LATE BREAKING NEWS: I got the schematic all done, then realized that if
the bottom transistor is an N-channel FET, the only problem with lightly
driving the upper transistor is the turn-off speed of the supply. If you
can stand it working at a mosey, you can size the upper resistor for 1mA
or so of current (or whatever your pin will drive), then size the lower
resistor to drop 10-15V at 1mA. Turn-off won't be snappy, but that may
not matter to you.

____
o--|____|-.
|
>|
|----.
/| |
| ///
| .------o switched -15V
| |/
o-----------|
| |>.
- |
| | |
| | |
- |
| |
'--------------o
|
|
o
-15V

[Tim Wescott]

Aw, now that's too easy.

[Al Clark]

Tim Wescott <seemyw...@myfooter.really> wrote in

news:67KdnTauUsazgrPJ...@giganews.com:

Yeah, I thought Ian's idea was clever.

Al

[glen herrmannsfeldt]

Al Clark <acl...@danvillesignal.com> wrote:
> I have an application where I want to turn on a bipolar +/-15
> volt supply from 3.3V. The supply already exists, I am
> disconnecting a part of a circuit to save power during an idle
> state.

> Integrated load switches comprised of P channel and N channel
> enhancement mode FETs are readily available for positive
> supplies.

If it has P and N channel FETs, isn't it supposed to work both ways?

The only one I know about, the AO9926B, is described as "uni-directional
or bi-directional load switch".

I believe the AO9926B is designed for 3.3V or 5V supplies, and
can switch up to 7A or so. It is N channel, so needs a positive
voltage (Vgs) to turn on. Two fit in a tiny SOIC-8 package.

There might be some others in the family that would work for you.

(I found out about this one when working on a circuit that used it.
I looked up the data sheet to see what it was.)

-- glen

[glen herrmannsfeldt]

Al Clark <acl...@danvillesignal.com> wrote:
> I have an application where I want to turn on a bipolar +/-15
> volt supply from 3.3V. The supply already exists, I am
> disconnecting a part of a circuit to save power during an idle
> state.

> Integrated load switches comprised of P channel and N channel
> enhancement mode FETs are readily available for positive
> supplies.

The AO6602 has a P and N channel FET in one package, Vds of 30V.

-- glen

[Ian Malcolm]

Al Clark <acl...@danvillesignal.com> wrote in

news:XnsA3BB5F9981AE5ac...@69.16.179.21:

It will have problems if there is too much capacitance on the +ve rail as
the rail must come up from 3V to approx 10V fast enough to get the MOSFET
through its linear region before it gets too hot.

An alternative to directly switch both rails is P and N channel discrete
MOSFETS, each source to the corrisponding supply with a gate-source
resistor to turn them off, and an optocoupler shorting the gates togger
to turn them on. It will need a limiting resisistor for the opto's LED
so thats three active parts and three resistors total. The MOSFETs each
get 15V of gate drive - if you want to reduce that, add an extra resistor
in series with the opto's collector.

[Al Clark]

Ian Malcolm <See.My.Sig...@totally.invalid> wrote in
news:XnsA3BBDED20E...@213.239.209.88:

I have a working circuit now with a pair of BJTs and a dual
N/P channel FET.

Al






>

[Ian Malcolm]

glen herrmannsfeldt <g...@ugcs.caltech.edu> wrote in news:m0mkhm$vjk$1
@speranza.aioe.org:

> AO6602

Add a BC847BVN (NPN/PNP complimentry pair) using the PNP transistor in
common base as a level shifter for the -ve switch (ground PNP base, strap
emitters togater, control signal to NPN base), and we have a two package +
3 resistor solution.

[Al Clark]

Ian Malcolm <See.My.Sig...@totally.invalid> wrote in

news:XnsA3BCCCBEC...@213.239.209.88:

> glen herrmannsfeldt <g...@ugcs.caltech.edu> wrote in
> news:m0mkhm$vjk$1 @speranza.aioe.org:
>
>> AO6602
>
> Add a BC847BVN (NPN/PNP complimentry pair) using the PNP
> transistor in common base as a level shifter for the -ve
> switch (ground PNP base, strap emitters togater, control
> signal to NPN base), and we have a two package + 3 resistor
> solution.
>

And now you have the same solution that I ended up with (with
help from Tim). Place a resistor (R1) between the emitters, and
bias each FET (R >> R1) from gate to source. I added .1u caps
from gate to drain to control inrush currents during switching.
Turn on threshold is 2 x Vbe or about 1.4V which will work well
with 3.3V logic. I added a pulldown on the input to insure that
the circuit stays off in tri-state or weak pullup conditions. It
doesn't take very much current to turn on the FETs.

I also used the SSOT-6 FET pair (AO6602) that was suggested
earlier. These are very cheap. Fairchild & Diodes both make pin
compatible substitutes. The BJTs can be just about any 2 cent
transistors (3904/06 for example)or a dual pair like Ian
suggests (5 cents).

If you need more current the IRF7509PbF seems reasonable.

Thanks everyone for ideas and contributing.

Al

[Ian Malcolm]

Al Clark <acl...@danvillesignal.com> wrote in

news:XnsA3BC68736AAFBac...@69.16.179.20:

> Ian Malcolm <See.My.Sig...@totally.invalid> wrote in
> news:XnsA3BCCCBEC...@213.239.209.88:
>
>> glen herrmannsfeldt <g...@ugcs.caltech.edu> wrote in
>> news:m0mkhm$vjk$1 @speranza.aioe.org:
>>
>>> AO6602
>>
>> Add a BC847BVN (NPN/PNP complimentry pair) using the PNP
>> transistor in common base as a level shifter for the -ve

>> switch (ground PNP base, strap emitters together, control

>> signal to NPN base), and we have a two package + 3 resistor
>> solution.
>>
>
> And now you have the same solution that I ended up with (with
> help from Tim). Place a resistor (R1) between the emitters, and
> bias each FET (R >> R1) from gate to source. I added .1u caps
> from gate to drain to control inrush currents during switching.
> Turn on threshold is 2 x Vbe or about 1.4V which will work well
> with 3.3V logic. I added a pulldown on the input to insure that
> the circuit stays off in tri-state or weak pullup conditions. It
> doesn't take very much current to turn on the FETs.
>
> I also used the SSOT-6 FET pair (AO6602) that was suggested
> earlier. These are very cheap. Fairchild & Diodes both make pin
> compatible substitutes. The BJTs can be just about any 2 cent
> transistors (3904/06 for example)or a dual pair like Ian
> suggests (5 cents).
>
> If you need more current the IRF7509PbF seems reasonable.
>
> Thanks everyone for ideas and contributing.
>
> Al

The resistor between the emitters controls both collector currents which
I suppose you need if you are slugging the switching time by adding
Miller caps on both MOSFETs.

I *ASSUME* you've verified the MOSFETs remain within their S.O.A while
switching.

[Al Clark]

Ian Malcolm <See.My.Sig...@totally.invalid> wrote in

news:XnsA3BCAA233F...@78.46.70.116:

SOA, not a problem in my case, but clearly could be an issue
in other implementations.

Thanks Ian

Al