Design help for switching a 5V-supply: FET or BJT, relay or not?

Background: I've used BJTs before, but never FETs. Been reading the short intro at https://oscarliang.com/how-to-use-mosfet-beginner-tutorial/ and that's about everything I know about FETs (i.e, not a thing really).

My scenario now is that I have a voltage monitor for a 5V-rail. The monitor, well, monitors the rail and if/when it hits 5.5V I get a signal for that, and I will react (cut the supply) upon it.

The "cut the supply" part is where I am now. The easy way - i.e. the way I know how to do - is a BJT controlling a relay that cuts the rail. This works, but my concern is that 1. The BJT is slow on reacting, and 2. The relay is slow on reacting.

Finding a real fast BJT will solve 1, but finding a relay with reaction time not within milliseconds is harder, so 2 is unsolved. Here's where I thought of a FET. A real world scenario is a better incentive for me than just playing with numbers.

I could use a rapid FET + relay, but that will still keep the bottleneck of the responsetime for the relay.

So, here's the idea that just popped my head: what if I crap the "transistor + relay" idea and simply let a power-FET act as a rail-switch? No more than 2A will flow through the rail.

Questions:

1. Will a single FET be able to make or break the power rail? I'm thinking that a BJT has the Uce_sat drop of about ~0.4V, would a FET have this drop as well? (0.4V-drop from 5V is very bad in this scenario, that's why kept trying with the relay, to avoid the drop)

2. Similiar to the above question, how would an AC-rail work through a FET? Let's say I want to switch 12VAC, will there be a voltage drop / loss, or will the full 12VAC come out clean?

• the BJT won't matter. Relay acting times are orders of magnitude larger. So your approach makes a lot of sense. Feb 10, 2017 at 9:26
• so, do you have a second voltage rail to control the relay or the gate of the mosfet? Feb 10, 2017 at 9:33
• Would replacing a fuse or reseting a breaker be acceptable? This seems to be an exception/fault detection/protection system. If it is a crowbar circuit would probably work. It would give you the speed of a semiconductor device acting as an extreme shunt regulator until the fuse/breaker trips and the on characteristics of a relay.
– RoyC
Feb 10, 2017 at 10:35
• Root cause: why is your supply potentially rising above the limiting value - fix it at source and don't use a band-aid. Feb 10, 2017 at 10:54
• @RoyC: I had this approach at first, using a TRIAC as crowbar. It worked, but my monitor was too sensitive and would trigger the crowbar on short voltage spikes / transients, so after switching a lot of fuses I finally gave up on that idea. I could probably have built a current sensor / limiter, but I don't have a lot of space on the PCB so I abandoned it.
– bos
Feb 10, 2017 at 11:38

The tutorial you've linked is way too basic. I'd recommend you to take a look at this one, which has deeper detail without becoming excessively complex: MOSFET as a switch - Tutorial

Q1:

If my understanding is correct, your application requires cutting off a rail that supplies 5V to a load (or more than one) connected to ground.

If you want to do that, you'll need a high side switch. A P-channel enhancement mode power MOSFET may be your best option, as there are plenty of devices available with logic-level compatible gate threshold voltage (a very convenient feature).

Check Vishay Siliconix, International Rectifier or any other manufacturer MOSFET Selection Guide for a suitable device.

This one Si4477DY could be good your application due to its very low RdsON (< 6 mOhm i.e., 12 mV drop at 2 Amps) and because it will be deep driven when grounding its gate (you'll have Vgs = -5V). You could also trade off some of its capabilities for a lower priced or more conveniently packaged device, it's up to you.

If you need to an extremely low RdsON, you could opt for a N-channel MOSFET, but then you'll need additional circuitry to drive its gate (for example, LTC1154) and complexity can escalate pretty quickly. I'll recommend against it.

Q2:

You can switch an AC load through two MOSFETs, but be aware that their body diodes will be conducting alternatively thus dropping some voltage. For a 12V AC load you may be better off just using a relay.

1. Yes, FETs are very well suited for switching DC power. Actually FETs are better than BJTs in this respect. There is no problem to have FET that can handle 2A and much more than that.
2. If you want to switch AC power, I would recommend you to look at triacs and optotriacs.

I think the device you are looking for is e.g. the LTC4360 in case of the 5V protection. It's a easy solution for over voltage protection and turns your output off within $1\mu s$ of (according to the FET). Attached the matching schematic from LT.