# drive mosfet with 3.3V, minimal component count

#### Irving

Joined Jan 30, 2016
3,190
Okay.
How about this circuit (LTspice simulation below) which uses only 2 resistors and one NPN transistor besides the MOSFET:
It biases the NPN ON when the input is Hi-Z (S1, which simulates the micro Hi-Z output, is open, green trace) so that the MOSFET is off (yellow trace) for that condition.
A low on the input (S1 closed) then turns off Q1 which turns on the MOSFET with a full 5V on the gate.

Will that work for you?

View attachment 243674
The potential issue with this approach is what happens if the 3.3v rail comes up after the 5v rail, which may be true if 3.3v is derived from 5v. Then you'll have a 1.4A spike on 5v rail on start up...

#### crutschow

Joined Mar 14, 2008
31,128
drive the gate by toggling the direction bit, so it will go from Output Low to Input.
You can also just toggle the output high and low to switch the MOSFET.
Enough to turn a MOSFET on, albeit slowly if its got a large gate charge.
That's likely to slow for the OP's purpose as the device is switching PWM.
the PWM is only about 1200Hz.

#### crutschow

Joined Mar 14, 2008
31,128
The potential issue with this approach is what happens if the 3.3v rail comes up after the 5v rail, which may be true if 3.3v is derived from 5v. Then you'll have a 1.4A spike on 5v rail on start up...
Good point.
So the proper operation of the circuit I proposed will depend on the 5V supply not coming up before the 3.3V supply.

#### mikewax

Joined Apr 11, 2016
184
Is the MCU pin in question 5V tolerant?
No the tolerance is Vcc + .5 :-(

#### mikewax

Joined Apr 11, 2016
184
They actually generate a voltage at the output, typically 15 - 20v at a few uA.
WOW thanx i never knew about those. Good alternative to a bulky relay

#### mikewax

Joined Apr 11, 2016
184
Whats the reason for iPak/TO251 rather than an SMD part?
smaller footprint on the board. I first looked at some TO92-type fets, but they had to much Rds. The TO251 type were the smallest ones with Rds < 75mohm.

#### mikewax

Joined Apr 11, 2016
184
Good point.
So the proper operation of the circuit I proposed will depend on the 5V supply not coming up before the 3.3V supply.
yes it is. The uC is an atmega16u2 and it has it's own built-in 3.3 regulator. The circuit schematic i posted above is oversimplified for purpose of illustration. It's actually built-in, and it has a "3.3 out" pin that i'll use to pull up the NPN. So at POR the "3.3v out" appears first, before the uC even resets. So yeah a little spike for < 1uS doesn't matter.
The device i'm designing is powered from the USB port on my phone, and what i've found is that if i plug it in and it draws > 1.5A for longer than about 600mS, the voltage drops out and the device keeps PORing indefinitely.
And as for the PWM frequency, it's really arbitrary and could be as low as 20Hz or so because the load is just a heating element.

#### Irving

Joined Jan 30, 2016
3,190
smaller footprint on the board. I first looked at some TO92-type fets, but they had to much Rds. The TO251 type were the smallest ones with Rds < 75mohm.
Hmmm, you'll struggle for $$I_D$$ >=1.4A in a TO92 case.... Is footprint that critical? A SOT-23 footprint isn't so different from a TO251, its max 3 x 2.5mm v 6.7 x 2.4 viewed from above.

Or this FDMC8588 3.4 x 3.4mm Power33 package, Rds < 10mOhm @ Vgs = 3 and Id < 5A on a minimal footprint for a temp rise < 32degC

Last edited:

#### Irving

Joined Jan 30, 2016
3,190
The device i'm designing is powered from the USB port on my phone, and what i've found is that if i plug it in and it draws > 1.5A for longer than about 600mS, the voltage drops out and the device keeps PORing indefinitely.
Hmmm. OTG should be 2A, but I always felt that was too much...

What is it that you're heating?

Last edited:

#### Phil-S

Joined Dec 4, 2015
231
I too have heard that the 2N7000 isn't wonderful.
I gather the DigiKey MOSFET selector is the place to look, but if it's an N channel, logic level device you want, the choice rapidly goes from several thousand to tens or less.
Bald Engineer website has something to say on this and the Arduino Forum should have some suggestions.
If the 2N7000 works, all well and good.

#### mikewax

Joined Apr 11, 2016
184
Hmmm. OTG should be 2A
USB 3.1 means no more than 7.5W, and the phone makers are real serious about following the exact letter of the usb standard, and not 1mW more.
so that FDMC8588 would almost fit on the board, but it's complicated. There's actually a mechanical element involved. The fet is inserted and soldered into the circuit board, then gate and source leads are snipped off underneath, but the drain lead is not snipped. It's bent over and actually curls around the edge of the PCB and forms a contact surface that has a tiny bit of elasticity. The 3.5ohm load is a vape cartridge that screws into a fitting and touches the ring-shaped drain lead which has about 5 or 6 mil of spring action which appears to be well within the elastic limit of the metal.
The TO251 fet has just the right electrical and mechanical properties that we need. My manager was quite impressed when i came up with that idea, a spring-loaded contact without the spring.

#### mikewax

Joined Apr 11, 2016
184
the choice rapidly goes from several thousand to tens or less.
It sure does. I needed a fet with an Rds of between 50 - 80mohm. What a PITA that was.

#### Irving

Joined Jan 30, 2016
3,190
The TO251 fet has just the right electrical and mechanical properties that we need. My manager was quite impressed when i came up with that idea, a spring-loaded contact without the spring.
I think your ongoing problem is that many manufacturers have deprecated TO251. It was a space saving design with poorer thermal characteristics than TO220 which itself seems to be losing out to the significantly better TO247 package. And though SMD has many thermal challenges for serious power, in the sub-2W logicFET space it's becoming ubiquitous to the point I actually have a TO220 size carrier to solder SOT23/SOT89/etc devices on to so I can use in breadboards or in perf-board for prototyping.

#### mikewax

Joined Apr 11, 2016
184
many manufacturers have deprecated TO251.
Well that explains why they're so damn hard to find. It means i gotta find another way before they completely run out. But whatthehell there should be a way to fit a TO220 if i edit the footprint.

#### Irving

Joined Jan 30, 2016
3,190
Well that explains why they're so damn hard to find. It means i gotta find another way before they completely run out. But whatthehell there should be a way to fit a TO220 if i edit the footprint.
Until they go out of fashion
Of course you could always just solder a 'pin' onto the board independently of the MOSFET...