Power Gating Devices

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.

devices, an SD card for example. Is there any proper practices when doing this and/or anything to be concerned about?

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

gate devices, an SD card for example. Is there any proper practices when doing this and/or anything to be concerned about?

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

ate devices, an SD card for example. Is there any proper practices when doi ng this and/or anything to be concerned about?

ce

you a

es

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

devices, an SD card for example. Is there any proper practices when doing this and/or anything to be concerned about?

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.

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.

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.

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.