5
\$\begingroup\$

I want a "power button" that is actually two buttons; one on, one off. This needs to switch on the order of 15 V and up to 10 A, for a LiFePO4 battery powered device. I don't want to use a hardwired switch where this is mechanical, because I later also want soft (MCU-based) control of the same on/off functions. I don't want to use a relay because 60 mA for a typical coil current is a lot (the typical load is perhaps 500 mA, but peak load on the device is 10 A.)

This is the best I could come up with. De-bouncing the buttons is not needed. When first connected, I think it enters "off" state because the gate is weakly pulled to ground through the load (output) -- and even if it isn't, the main power P-channel FET isn't going to have any gate charge.

I'm asking for two things here:

  1. If you want to spend a few minutes helping out a poor n00b on circuit design, please critique this design and suggest whatever you think needs improving in this circuit. (Three FETs? really?)
  2. If you think my diagram is hard to read (I got that last time :-) then please at least critique exactly what you think is poor about the diagram, so that I can better communicate in the future.

Thanks in advance for any help!

image

\$\endgroup\$
3
  • 2
    \$\begingroup\$ This is actually a nice little circuit. I like how you are providing feedback from the out put to latch state. R1 and R4 are not necessary unless someone does something silly, so that shows careful design for failure modes. I would simulate the off state of Q1 and how well you can hold it's gate to ensure that it does turn on slightly. Also what happens when the output load goes away? -> sub-threshold leakage will will slowly charge stray capacitance and the output will go hot eventually. \$\endgroup\$ Oct 10, 2012 at 4:49
  • \$\begingroup\$ Thanks for the comment! I think it's detailed enough it could be an answer so I could vote it up ;-) \$\endgroup\$
    – Jon Watte
    Oct 10, 2012 at 18:20
  • \$\begingroup\$ You can vote up comments, too -- hover over the comment and a small arrow will appear in the top left. \$\endgroup\$
    – Jim Paris
    Oct 10, 2012 at 20:04

1 Answer 1

7
\$\begingroup\$

Could it be that what you had in mind was this?

enter image description here

Purpose of components:

  • Q5 - P-channel MOSFET, high side power switch
  • SW1 - momentary normally open OFF switch, which drives the gate high. Since it's connected directly to the gate of Q5, it will overpower the ON switch, which is connected through 1kΩ.
  • R32 - pull up, makes sure that the circuit is off when it's not on.
  • SW2 - momentary normally open ON switch. When pressed, the gate of Q5 is driven low and Q5 starts to conduct. Output drives the gate of Q12 high.
  • Q12 - N-channel MOSFET, keeps the gate of Q5 low when SW2 is released and SW1 not pressed.
  • R30 - pull down, makes sure that the output is pulled to ground when the circuit is off and load is disconnected. Because of R30, the gate of Q12 can't float up and turn the circuit on spontaneously.
\$\endgroup\$
4
  • \$\begingroup\$ Thanks for the suggestion! I had actually considered using a weak pull-up instead of a P-channel MOSFET for the "off" state, but I'm worried that it will switch too slowly if that's something you rely on. I've burned more than one MOSFET by not providing enough switching current quickly enough to those pretty hefty gate capacitances :-/ Also, R30 does not match my requirements. This is a circuit that switches power for a battery. An idle current of 12 mA (12 V / 1 kOhm) is way too much. Maybe a 100 kOhm there would be sufficient, too. Thanks again for the answer! \$\endgroup\$
    – Jon Watte
    Oct 10, 2012 at 18:18
  • \$\begingroup\$ Btw: What software did you use to draw that diagram? \$\endgroup\$
    – Jon Watte
    Oct 10, 2012 at 18:24
  • 1
    \$\begingroup\$ @JonWatte This diagram was draw in OrCAD. \$\endgroup\$ Oct 10, 2012 at 18:41
  • \$\begingroup\$ Q12 could be a NPN BJT with base resisters .I know that your circuit will work because I did something similiar about 18 years ago that incorperated an UVLO function as well . + 1 \$\endgroup\$
    – Autistic
    Dec 31, 2015 at 20:53

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.