1
\$\begingroup\$

The microprocessor on my board can switch off all the other components via an NPN transistor.

This means that I have the 3V3 line going from the processor to everything, a ground going from all the components to the transistor, then the transistor's emitter going to the processor's ground.

Which of the following should I have:

  • A ground plane on either side of the transistor
  • A power plane and a ground plane for the components' grounds to the transistor
  • A power plane and a ground plane for transistor's ground back to the processor

Edit The schematic in question. The far left is an Adafruit GPS breakout. The center is a Feather M0 AdaLogger, Bottom left is an ms5611 altitude/temperature sensor, bottom right is an LSM9DS1 9-axis accelerometer breakout, top right is an NRF24L01 radio. The resistors, bottom-center, are a voltage divider that allow me to determine if the system is running on USB or battery, and the transistor/resistor at the top are, of course, the offending switching circuitry. enter image description here

\$\endgroup\$
7
  • 5
    \$\begingroup\$ Please show the schematics. Switching the ground of connected components usually results into side-effects you are not prepared for, such as devices powering themselves via data buses and IO pins of components that are grounded. \$\endgroup\$
    – Justme
    Commented Nov 27, 2021 at 20:43
  • 4
    \$\begingroup\$ Use a high-side PMOS switch. There are so many problems with switching on the low side, including the fragmentation of the GND plane, that it is just not worth it. Posting as a comment because it doesn't really answer your question. \$\endgroup\$
    – user57037
    Commented Nov 27, 2021 at 21:18
  • 1
    \$\begingroup\$ Echoing @mkeith 's excellent point, use a high-side switch. The is more on GND than just the returning supply current. You want as low an impedance as you can get from a wide ground plane, not funnelling it all through the impedance of a thin transistor leg. Decouple well on the load side of the transistor. Consider inrush current at power-up. But before any of that: edit your question and put the schematic in it. \$\endgroup\$
    – TonyM
    Commented Nov 27, 2021 at 21:25
  • \$\begingroup\$ @Justme I power down the components just before powering down the microprocessor, so there's nothing making it to the subcomponents. In full powerdown, the system doesn't burn any more power than just the microprocessor does by itself when in powerdown. (Unfortunately, the *#(&$ battery management hardware still pulls 250uA from the battery in this mode, but I'm stuck with it.) Good suggestion, though. Thanks. \$\endgroup\$
    – user30997
    Commented Nov 27, 2021 at 21:55
  • \$\begingroup\$ @mkeith It's been a while since I've done it, but I recall that when I set up a circuit w/ the PNP on the high side, nothing powered up. The voltage difference between ground and the emitter wasn't enough to run the circuit. I couldn't explain why it happened - is there something I don't know about transistors that would explain this? Should I be using something other than a 547B for the task? All told, the peripherals draw about 60mA and I'm driving the base through a 470Ohm resistor, which should be good to 70mA. \$\endgroup\$
    – user30997
    Commented Nov 27, 2021 at 21:59

1 Answer 1

Reset to default
1
\$\begingroup\$

A GND plane on the bottom layer is the minimum you can do for low EMI emissions.

If you can, but that depends on the number of parts and on the topology of the circuit, use the top layer for 3.3 V as well.

The 3.3 V layer along with the GND bottom layer allows you to have a distributed capacitance all over the PCB.

That distributed capacitance is the dream of all high speed PCB designers because it filters out and doesn't allow high frequencies radiations to leave the board.

It's a capacitance distributed all over the board. It's value increases by decreasing the thickness of the PCB.

Yet, that distributed capacitance is much more effective than the classical 10 nF or 100 nF capacitors because it has a very wide bandwidth, up to 10 GHz, that capacitors don't have.

Discrete capacitors after a certain frequency works as inductors.

Murata's X7R 0402 10 nF capacitors are the widest bandwidth capacitors for VDD filtering and have a bandwidth of 3 GHz, well below the corresponding value of the distributed capacitance.

Now you know why smartphones' PCB thickness is below 1 mm.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.