Power Gating Devices

[Luther Blackwood]

I've been toying around with the idea of using transistors to power gate devices, an SD card for example. Is there any proper practices when doing this and/or anything to be concerned about?

[rickman]

On 5/3/2013 3:29 PM, Luther Blackwood wrote:
> I've been toying around with the idea of using transistors to power gate devices, an SD card for example. Is there any proper practices when doing this and/or anything to be concerned about?

I'm no expert, but what I have observed is that there is a preference
for n-channel FETs for this purpose since they have a lower channel
resistance for a given chip size. The problem with an n-channel device
is that it needs a gate voltage higher than Vdd to turn on the FET. A
p-channel device does not have this issue, but you will need a larger
and more expensive device to get the same Ron and therefor the same low
voltage drop. This is just a matter of cost really. It will cost you a
bit more for the p-channel device which meets your spec.

However, do you really need this? I believe there are already power
control features in SD cards as well as most peripherals. Many devices
drop to single digit uA levels if not lower.

--

Rick

[Mark Borgerson]

In article <km1735$68b$1...@dont-email.me>, gnu...@gmail.com says...

Perhaps it's the code I've written, but I have found that there
are some cases where an SD card just refuses to go into low-power
idle mode or to accept commands in SPI mode. I've found it useful
to have the ability to switch off, then on the power to the
SD card to get the card back into a known state to start
communications. I usually do this with a small (SOT23-5) voltage
regulator with an enable input (MIC5233, for example.)

Mark Borgerson

[John Speth]

Make sure whatever design you choose that you have the flexibility to
deal with unexpected problems that might appear due to power sequencing
and even the circuit powered via I/O pins and not the intended power bus.

JJS

[lang...@fonz.dk]

On 6 Maj, 05:31, Mark Borgerson <mborger...@comcast.net> wrote:
> In article <km1735$68...@dont-email.me>, gnu...@gmail.com says...

using a regulator is simple and also gives you thermal and current
limit
but they cost in dropout voltage

There's lot of power switches for USB so for less than 5V there's
plenty to choose from


-Lasse

[Stef]

In comp.arch.embedded,

rickman <gnu...@gmail.com> wrote:
> On 5/3/2013 3:29 PM, Luther Blackwood wrote:
>> I've been toying around with the idea of using transistors to power gate devices, an SD card for example. Is there any proper practices when doing this and/or anything to be concerned about?
>
> I'm no expert, but what I have observed is that there is a preference
> for n-channel FETs for this purpose since they have a lower channel
> resistance for a given chip size. The problem with an n-channel device

I think this is still true, but p-chanel devices have become a lot
better in the last 10(?) years. But so have the n-channels and so there
is still a difference. But I think current p-channels can have a lower
resistance than 'old' n-channels.

> is that it needs a gate voltage higher than Vdd to turn on the FET. A
> p-channel device does not have this issue, but you will need a larger
> and more expensive device to get the same Ron and therefor the same low
> voltage drop. This is just a matter of cost really. It will cost you a
> bit more for the p-channel device which meets your spec.
>
> However, do you really need this? I believe there are already power
> control features in SD cards as well as most peripherals. Many devices
> drop to single digit uA levels if not lower.

Maybe we made the same mistakes as Mark, but we also experienced SD cards
using 0.5mA. We simply put a sot-23 p-channel in the power line and that
works fine. We used the FDC6324L, but there must be other suitable devices.


--
Stef (remove caps, dashes and .invalid from e-mail address to reply by mail)

Snakes. Why did it have to be snakes?
-- Indiana Jones, "Raiders of the Lost Ark"

[John Devereux]

Stef <ste...@yahooI-N-V-A-L-I-D.com.invalid> writes:

> In comp.arch.embedded,
> rickman <gnu...@gmail.com> wrote:
>> On 5/3/2013 3:29 PM, Luther Blackwood wrote:
>>> I've been toying around with the idea of using transistors to power
>>> gate devices, an SD card for example. Is there any proper practices
>>> when doing this and/or anything to be concerned about?
>>
>> I'm no expert, but what I have observed is that there is a preference
>> for n-channel FETs for this purpose since they have a lower channel
>> resistance for a given chip size. The problem with an n-channel device
>
> I think this is still true, but p-chanel devices have become a lot
> better in the last 10(?) years. But so have the n-channels and so there
> is still a difference. But I think current p-channels can have a lower
> resistance than 'old' n-channels.

There will always be a differential, but for low power switching it is
likely insignificant and outweighed by the simpler driving circuitry.

One issue I have not seen mentioned here is controlling inrush
current. A device may have bulk capacitance causing initial current to
be very high if not controlled. A simple feedback capacitor circuit can
be used with a P-fet to control this. See e.g. datasheet for FTDI parts
"power switching configuration" p25.

<http://www.ftdichip.com/Support/Documents/DataSheets/ICs/DS_FT232R.pdf>

>> is that it needs a gate voltage higher than Vdd to turn on the FET. A
>> p-channel device does not have this issue, but you will need a larger
>> and more expensive device to get the same Ron and therefor the same low
>> voltage drop. This is just a matter of cost really. It will cost you a
>> bit more for the p-channel device which meets your spec.
>>
>> However, do you really need this? I believe there are already power
>> control features in SD cards as well as most peripherals. Many devices
>> drop to single digit uA levels if not lower.
>
> Maybe we made the same mistakes as Mark, but we also experienced SD cards
> using 0.5mA. We simply put a sot-23 p-channel in the power line and that
> works fine. We used the FDC6324L, but there must be other suitable devices.

--

John Devereux

[Rob Gaddi]

That's what I was going to mention. Remember that any time you've got
voltage on an I/O to an unpowered device, you're going to backdrive
through the ESD diodes and charge the unpowered VCC.

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