Maker Pro
Maker Pro

Logic Level MOSFET

J

James Fraser

I'm working on some electronics for part of my personal winter project
and I could use some advice.

I would like to control an incandescent bulb (yes, can't be an LED.)
with the output from a 3.3 volt logic device (<= 5mA current). The
bulbs are going to be pretty lightweight loads, maybe 150 mA at 5
volts.

Looks like a MOSFET would be a good controller on the low side, and my
biggest issue will be making sure that it handles the bulb-on
transition well. (Occasionally the bulb might be set to blink every
second.)

Two questions:
Is a single MOSFET with a pull down resistor on the gate a good
solution for this? I'm concerned that the 3.3 volts isn't enough for
this.
If so, how do I begin to spec the MOSFET? Digikey's interface doesn't
let me filter on Vgs(th). I think I need a Vgs(th) of <=2.0 volts, and
Ids of 1A to handle the bulb-on transition current. Reasonable? Is
there some standard or popular part I should use.

In case it matters, I'll have eight of these in the same device.


Uncertain hobbiest newb...
James Fraser
 
R

Robert Wilson

James said:
I'm working on some electronics for part of my personal winter project
and I could use some advice.

I would like to control an incandescent bulb (yes, can't be an LED.)
with the output from a 3.3 volt logic device (<= 5mA current). The
bulbs are going to be pretty lightweight loads, maybe 150 mA at 5
volts.

Looks like a MOSFET would be a good controller on the low side, and my
biggest issue will be making sure that it handles the bulb-on
transition well. (Occasionally the bulb might be set to blink every
second.)

Two questions:
Is a single MOSFET with a pull down resistor on the gate a good
solution for this? I'm concerned that the 3.3 volts isn't enough for
this.
If so, how do I begin to spec the MOSFET? Digikey's interface doesn't
let me filter on Vgs(th). I think I need a Vgs(th) of <=2.0 volts, and
Ids of 1A to handle the bulb-on transition current. Reasonable? Is
there some standard or popular part I should use.

In case it matters, I'll have eight of these in the same device.


Uncertain hobbiest newb...
James Fraser
Well it depends what kind of MOSFET you are using, whether it is hard on
at 3.3v. Also the pull up resistor and the impdeance of your drive
output from your uC(?).

Look at the spec for your device. Is it hard on at 3.3v.

Regards,
Rob.
 
J

James Fraser

Well it depends what kind of MOSFET you are using, whether it is hard on
at 3.3v. Also the pull up resistor and the impdeance of your drive
output from your uC(?).

Look at the spec for your device. Is it hard on at 3.3v.

Regards,
Rob.

I haven't chosen a MOSFET. That's my question. How do I find the right
piece? Browse through lots of data sheets? Or does anyone have a
recommendation?


Hobbyist with less than perfect spelling...
James Fraser
 
J

Joerg

James said:
I'm working on some electronics for part of my personal winter project
and I could use some advice.

I would like to control an incandescent bulb (yes, can't be an LED.)
with the output from a 3.3 volt logic device (<= 5mA current). The
bulbs are going to be pretty lightweight loads, maybe 150 mA at 5
volts.

Looks like a MOSFET would be a good controller on the low side, and my
biggest issue will be making sure that it handles the bulb-on
transition well. (Occasionally the bulb might be set to blink every
second.)

Two questions:
Is a single MOSFET with a pull down resistor on the gate a good
solution for this? I'm concerned that the 3.3 volts isn't enough for
this.
If so, how do I begin to spec the MOSFET? Digikey's interface doesn't
let me filter on Vgs(th). I think I need a Vgs(th) of <=2.0 volts, and
Ids of 1A to handle the bulb-on transition current. Reasonable? Is
there some standard or popular part I should use.

In case it matters, I'll have eight of these in the same device.

You can use a FET with a low threshold but that threshold should be
guaranteed. For example:
http://www.vishay.com/docs/73235/73235.pdf

Digikey P/N: SI2312BDS-T1-E3CT-ND

Around 50 cents a pop if you buy 25 and it's in stock. It is SMT, pretty
much like everything else these days. If needed it can switch a whole
lot more than 150mA so you might want to get some spares for other
projects. Happy soldering ;-)
 
J

Jan Panteltje

I'm working on some electronics for part of my personal winter project
and I could use some advice.

I would like to control an incandescent bulb (yes, can't be an LED.)
with the output from a 3.3 volt logic device (<= 5mA current). The
bulbs are going to be pretty lightweight loads, maybe 150 mA at 5
volts.

A lightbulb, when cold, has about 1/10 the resistance from a hot one.
So your MOSFET will have to be able to switch at least 1.5A with minimal losses.

A IRLZ34N MOSFET may do, it has a maximum gate threshold of 2V (minimum is 1),
a TO220 case, and does 30A at 55V.
It also has a build in zener to protect against avalance breakdown.
 
J

Joerg

James said:
I haven't chosen a MOSFET. That's my question. How do I find the right
piece? Browse through lots of data sheets? Or does anyone have a
recommendation?

See my other post for a part number.

Other than that, Digikey helps: Key in FET, click "in stock", then scoot
to the right and pick some that have the right current and Rdson @ Vgs.
You can hold the CTRL key and left click to select all that could fit.

One more hint: Place a 1K resistor or so between gate and uC, in case
your uC is expensive or hard to unsolder. It protects the uC in case you
shorted a lamp an the FET "nukes out".
Hobbyist with less than perfect spelling...


None of us is perfect. Some might think they are but they aren't :)
 
J

John Larkin

I'm working on some electronics for part of my personal winter project
and I could use some advice.

I would like to control an incandescent bulb (yes, can't be an LED.)
with the output from a 3.3 volt logic device (<= 5mA current). The
bulbs are going to be pretty lightweight loads, maybe 150 mA at 5
volts.

Looks like a MOSFET would be a good controller on the low side, and my
biggest issue will be making sure that it handles the bulb-on
transition well. (Occasionally the bulb might be set to blink every
second.)

Two questions:
Is a single MOSFET with a pull down resistor on the gate a good
solution for this? I'm concerned that the 3.3 volts isn't enough for
this.
If so, how do I begin to spec the MOSFET? Digikey's interface doesn't
let me filter on Vgs(th). I think I need a Vgs(th) of <=2.0 volts, and
Ids of 1A to handle the bulb-on transition current. Reasonable? Is
there some standard or popular part I should use.

In case it matters, I'll have eight of these in the same device.


Uncertain hobbiest newb...
James Fraser

Vgs(th) doesn't tell you much; it tells you how much gate voltage you
need to get a tiny drain current.

You need a part with seriously low Rds(on) with 3.3 volts on the
gate... a couple of ohms max. There are lots of "logic level fets"
around, so it shouldn't be a problem. Take a look at the Fairchild or
IR or Zetex web sites.

The blinking/turnon transient may warm up the fet a bit, so don't use
a really tiny (like SOT-23) package.

One trick is to add a resistor in parallel with the fet, to keep the
filament warm all the time, to reduce the turnon surge. Cute, but you
really don't need that here.

John
 
J

Jamie

James said:
I'm working on some electronics for part of my personal winter project
and I could use some advice.

I would like to control an incandescent bulb (yes, can't be an LED.)
with the output from a 3.3 volt logic device (<= 5mA current). The
bulbs are going to be pretty lightweight loads, maybe 150 mA at 5
volts.

Looks like a MOSFET would be a good controller on the low side, and my
biggest issue will be making sure that it handles the bulb-on
transition well. (Occasionally the bulb might be set to blink every
second.)

Two questions:
Is a single MOSFET with a pull down resistor on the gate a good
solution for this? I'm concerned that the 3.3 volts isn't enough for
this.
If so, how do I begin to spec the MOSFET? Digikey's interface doesn't
let me filter on Vgs(th). I think I need a Vgs(th) of <=2.0 volts, and
Ids of 1A to handle the bulb-on transition current. Reasonable? Is
there some standard or popular part I should use.

In case it matters, I'll have eight of these in the same device.


Uncertain hobbiest newb...
James Fraser
Why not a NPN common emitter with open collector to the lamp as the load?
 
P

Phil Hobbs

Joerg said:
See my other post for a part number.

Other than that, Digikey helps: Key in FET, click "in stock", then scoot
to the right and pick some that have the right current and Rdson @ Vgs.
You can hold the CTRL key and left click to select all that could fit.

One more hint: Place a 1K resistor or so between gate and uC, in case
your uC is expensive or hard to unsolder. It protects the uC in case you
shorted a lamp an the FET "nukes out".



None of us is perfect. Some might think they are but they aren't :)

Oh, quit being hobbier-than-thou.

Cheers,

Phil Hobbs
 
J

John Larkin

Mmmmmmmmmmnah...not done that way:
http://www.onsemi.com/pub/Collateral/AN1542-D.PDF (fig 6) or for a very
simplified approach on light loads (pun intended) US6518713....

If I correctly interpret your "not done that way" as "nobody does
this", you're just flat wrong. It's been done for decades, in
thousands of applications.

If one must use incandescants, and inrush is a problem, PWM would be a
simpler solution than all that stuff in the ancient Motorola appnote.

The patent is absurd. Why not just taper the PWM drive and be done
with it?

John
 
S

Spehro Pefhany

Oh, quit being hobbier-than-thou.

Cheers,

Phil Hobbs

The hardest and hobbiest hobs are rumoured to be located in
H-E-double-hockeysticks.

Best regards,
Spehro Pefhany
 
J

Jim Thompson

If I correctly interpret your "not done that way" as "nobody does
this", you're just flat wrong. It's been done for decades, in
thousands of applications.

If one must use incandescants, and inrush is a problem, PWM would be a
simpler solution than all that stuff in the ancient Motorola appnote.

The patent is absurd. Why not just taper the PWM drive and be done
with it?

John

The first non-mechanical-rotary-switch turn signals for the T-bird
used honking Ge transistors as the switches and had a "pre-warm"
current active as soon as the ignition switch came on.

Later I used SCR's...

http://analog-innovations.com/SED/Pat-3544962.pdf

...Jim Thompson
 
J

Jan Panteltje

The first non-mechanical-rotary-switch turn signals for the T-bird
used honking Ge transistors as the switches and had a "pre-warm"
current active as soon as the ignition switch came on.

Later I used SCR's...

http://analog-innovations.com/SED/Pat-3544962.pdf

That they allow that patent... seems trivial to me.
You did not need the thermal strip. you can switch of a SCR in several ways,
one to short it for say a few ms with a transistor (makes hold current zero),
the other to reverse-voltage it by firing a second SCR connected via a capacitor.
me 1968.
 
J

James Fraser

One more hint: Place a 1K resistor or so between gate and uC, in case
your uC is expensive or hard to unsolder. It protects the uC in case you
shorted a lamp an the FET "nukes out".

Thanks all for the tips and advice. Right now I'm leaning towards
IRF3706 series that specs max Rds at 22 mOhm at Vgs 2.8V.

But your comment got me thinking more. Yes, my uC is expensive and
hard to change. I hadn't really thought about failure modes much. If I
put in a resistor between uC and gate, then I have to up my pull-down
resistors (I suppose I don't need those and the FET gate can discharge
through the uC. I need to look into that.)

Should I look at more isolation between the FET and controller? Not
hard to do, I think, since I'm not worried about frequency response.


Again, thanks to all for the help and tips...
James Fraser
 
J

Jim Thompson

Middle name "Elbert"? Is there some German blood flowing in your veins?

Yep. My grandmother Thompson's maiden name was Rhoda Warner and
Riverton, WV, is in an area of the Appalachians called Germany Valley.

...Jim Thompson
 
J

Joerg

James said:
Thanks all for the tips and advice. Right now I'm leaning towards
IRF3706 series that specs max Rds at 22 mOhm at Vgs 2.8V.

Wow, those are huge.

But your comment got me thinking more. Yes, my uC is expensive and
hard to change. I hadn't really thought about failure modes much. If I
put in a resistor between uC and gate, then I have to up my pull-down
resistors (I suppose I don't need those and the FET gate can discharge
through the uC. I need to look into that.)

Nah. Just make sure the FET turns on and off fast enough so it spends
the least amount of time in the linear region. Your FETs are a bit
largish, about 2500pF gate capacitance. 1K in series from uC pin to gate
would mean an RC constant of 2.5usec, still pretty fast. This roughly
will be your turn on and turn off time. In the event that the gate
isolation melts down or pops for some reason it might worst case get
connected to +5V, meaning above the VCC of your uC. However, since there
is that 1K series resistor the substrate diode will only be hit with a
few mA. Look in the absolute maximum table. Most can tolerate 10mA,
except for example the MSP430 which is spec'd at 2mA although it should
certainly be able to take more.

What is more important is that the 3.3V rail has enough load that a
broken gate isolation layer doesn't drive this 3.3V rail up through the
1K resistor. This can happen regardless of whether you have a buffer
driver or not.

Should I look at more isolation between the FET and controller? Not
hard to do, I think, since I'm not worried about frequency response.

I only do that if I need to preserve very fast transition times. Then I
use 74HC buffers and the like. Since you are just turning lamps on and
off at slow rates I don't see the need for that here.
 
J

John Larkin

Yep. My grandmother Thompson's maiden name was Rhoda Warner and
Riverton, WV, is in an area of the Appalachians called Germany Valley.

...Jim Thompson

Me too: Wussner, Munsch, and Wycks.

John
 
Top