1

I am in the process of designing a circuit which will allow me to safely OR two power supplies.

Constraints:

  • 5V
  • up to 5A from Vin_hi_prio
  • up to 2A from Vin_low_prio
  • Both come from the same PSU but through different routes
  • Vin_hi_pro should be preferred if present
  • Voltage drop should be under 200mV
  • When the power is turned on, it stays on - this is not a hot swap circuit. Both voltages will be switched off at roughly the same time.

What I have considered:

  • Diodes - voltage drop was too high
  • Dedicated IC - I had trouble finding one which would look good to me and I don't want to pay 3-4$ when it seems like a BC817 will do

My questions:

  • Does this circuit look good? Or did I miss something that should be relatively obvious?
  • Will this circuit, designed in a simulation, hold up in real life with real components and their manufacturing variations? I'm particularly worried about Q1 Vbe - it seems like the circuit will be very sensitive to variations here and might allow short connections between the two voltage sources.
  • Since both both power inputs come from the same power supply but through different paths, do I even have to care about switching/ORing those two voltages? They might come from different rails.
  • Do I need to care about any extra current running through my ground?

jaskij
  • 1,223
  • 8
  • 13

1 Answers1

1

Your schematic looks okay, but ...

  1. An IRF7404 has a parasitic diode from Drain-to-Source, and I think you have it backwards on the schematic. The diode will allow current to flow irrespective of the gate.

  2. I'd probably use a BSS138P instead of that BC817. Then you don't have to worry about bias currents or Vbe(on).

Andrew Lentvorski
  • 1,091
  • 4
  • 12
  • 1) I'll double check the IRF 2) The point of the divider on the base is to delay M1 turn on until M2 turns off, that's what gives me the low margins. Would replacing it with an NMOS there help? – jaskij Dec 09 '20 at 09:32
  • I'm not sure that divider actually does what you think with a bipolar transistor (the transistor has current gain). However, if you're *really* worried about M1 vs M2, you can replace Q1 with that nFET and replace R5 with a capacitor. Now you have an RC delay on turn on/turn off. Personally, I suspect that the threshold voltage of the IRF7404 (0.7V max) vs the BSS138P (0.9V min) is more than sufficient to guarantee the result you want. – Andrew Lentvorski Dec 20 '20 at 10:47
  • delay as in lower Vbe so that Q1 turns on at the right time during Vin_hi_prio ramp up. I did think about an RC circuit but that shifts the problem to M1 Vth. I was hoping to eliminate lack of power on output. But that doesn't seem like a big issue in my use case as it's only during power on and I'd rather expect Vin_hi_prio to ramp up first as it's directly from PSU. I haven't gotten back to the project since I asked the question due to real life getting in the way – jaskij Dec 20 '20 at 14:42