I need to switch on a couple power supplies with a 5V logic signal and want to do this with a MOSFET (for lower voltage loss than a BJT). The first is 3.3V at 5A max, and the other is 12V at 1A max. But I'm not sure how to design these switches and choose the MOSFETs. Can someone here help me with this design or know of a link that will teach me this. Thanks.

 

Is there a reason you can't switch a small relay for even lower loss?

__________________
\"You can\'t make a circuit fool-proof, cause fools are so ingenious!\"

 

Actually doing that currently, but wanted to change to the MOSFET since it would be much smaller. Space is tight. Also, from my high-level calculations ( P=I^2 x Rdson ), *average* power wasted should be quite low, since average current drawn is pretty low. I'm just not sure how to design it.

 

12V should'nt be a problem but the 3.3V level is near the g-s threshold of where MOSFET's only start to turn on.

__________________
\"You can\'t make a circuit fool-proof, cause fools are so ingenious!\"

 

Interesting statement. I always felt that it was based on the control voltage only (Vgs???) and not the *controlled* voltage. The good news here is that I have been able to find a substitiute for the 3.3V which can an on/off pin. So I'll make this a lower priority. So for the 12V line, can you direct me to how to go about designing this? Thanks.

 

Here is one way to do this with a PFET.

The way this circuit works, the source of the FET is tied to the supply. If you only have 0V available with a 3.3V line then the g-s voltage will not be enough to switch the device on properly.

Attached Images

load_switch.gif (20.8 KB, 885 views)

__________________
\"You can\'t make a circuit fool-proof, cause fools are so ingenious!\"

 

You just need to find a "Logic Level" MOSFET. These start to turn on at a much lower threshold and can be turned on quite strong by 3.3v. Many of the good ones with very low rdsOn are in an SO8 pkg. Not so friendly to use without a PC board though. Despite the size, the rdsOn is so low they can handle quite a few amps in switch mode.

http://www.digikey.com

Search MOSFET.
Select "Logic Level".
Select whether you want n-channel or p-channel. You probably want n.

 

A word of caution here. If you look at various logic level MOSFET specs, the gate threshold typically can be as high as 3.5V and as low as 1V. At this threshold only about 250uA of drain current is measured at about 10Vds at 25 deg C (gets worse at higher case temperatures). Unfortunately no manufacturer can guarantee what it will be. That is why they quote a min and max threshold. So if it is your unlucky day you may end up with some 3.5V devices. Definitely not the chance I will take to design such a device into a product for production with only 3.3V to switch it with.

__________________
\"You can\'t make a circuit fool-proof, cause fools are so ingenious!\"

 

This has sounded like a one-off project so the author can probably cherry pick a device if he has trouble with the threshold value. In my onw experience, I have found that most manufacturers make devices that have thresholds that are pretty darn close to spec values.. the wide margin in the spec I believe, is due to potential process variations which, on most days are controlled VERY well.

 

I don't understand your reasoning here. Looking at a spec, which spec is very well controlled, the 1,1.6 or the 3V?

If someone is a optimist he will like to work with the 1V spec. Me as a pessimist and designing for worst case would like to work with the 3V. At least I will feel confident that it will work very-very close to 100% of the time in my application (production).

Attached Images

mosfet.gif (4.6 KB, 741 views)

__________________
\"You can\'t make a circuit fool-proof, cause fools are so ingenious!\"

 

TheOne, thanks for the circuit... it's small/simple, and my switch-on signal is already active low, so I guess I can eliminate the NPN transistor. However, I'm interested in *how* this circuit was designed. Care to explain? So far any tutorials I've found on MOSFETS only discuss the design of these with holes and doping p-channel material, etc, etc. There is a big gap of practical design knowledge (that I'm missing) between that theory and a working schematic.

Oznog, I generally prefer TO-220 packages (easier to heatsink) so I started with that and found the FDP6030L. The datasheet says 13mOhm RdsOn at 10Vgs, which is close enough to the 12V that I should actually get. That translates to a nice low 13mV drop at a full 1A output. And also only 13mW of power consumption. Nice. Is that all there is to it?

I've dropped the requirement for the 3.3V switch. Instead, I need to switch 5V at up to 5A. If I use the same part, I'm looking at .5W used, which is considerable (in terms of heatsinking). Instead I (should) have an FDP6676 around here that has an RdsOn of .0075Ohms at 4.5Vgs. Calcs with this one give me .188 mW used at a full 5A, which is much nicer. Is that really all there is to it?

Yes, this is a one-off device. But I do also prefer to design for worst-case operation. I usually also add in an additional safety margin, depending on what is being designed and the consequences.

On a side note, if I wanted to make a MOSFET switch *really* turn on, can I just use a negative voltage that the gate to essentially raise the Vgs? Thanks.

 

I think you already explained it yourself :lol: . The device switches because of the Vgs difference. If you could pull the gate of a device on your 3.3V line below zero w.r.t the +3.3V it should switch just fine. If the device is used in fast switching applications with high currents, voltages, like in UPS's SMPS etc. then the gate drive calculations needs more careful attention.

__________________
\"You can\'t make a circuit fool-proof, cause fools are so ingenious!\"

 

If you are only going to make ONE of something, it is not a bad thing to choose a part that has a real threshold where you need it.

My _experience_ has been that most manufacturers produce parts that are actually close to nominal values even though the spec is wide. For your example, what I am saying is chances are GOOD that the real threshold will be very close to 1.6V so if he only buys a few and has to use only ONE, chances are good that he wont end up part that all have Vth of 3V.

The author has not stated he is going into mass production with this so he is not and doesnt necessarily need to be thinking about designing for such.