Using low current to switch higher current

[Jeremy D. Grotte]

I designed and built a circuit to control a bunch of LED's awhile back. It
uses a PIC16F628 to control a few n-channel MOSFETs, and those mosfets in
turn apply a ground to strings of LED's (which naturally have a power
sitting on them waiting to be grounded) which are brightness controlled by
PWM. The circuit works fine until I start loading it down with loads of
LED's, then everything starts to get hot, wiring, MOSFETs, all of it.
Normally this isn't a problem for my project, but in extreme cases (when I
start putting a zillion LED's on it), I'm going to want to design a sort of
'power distribution box'. The idea being that the MOSFETs control another
larger MOSFET, or something, somewhere else. The problem is that my circuit
(which is already built in mass quantity) applies a 'ground to a power', not
a 'power to ground' (if that makes any sense), so I guess you could say I'm
using low-side switching (if that's the correct term?).
I know the answer to this is simple, but I can't seem to wrap my brain
around it at the moment.

The only idea I can come up with is using the outputs I've already got to
switch an opto-isolator, and then use that output to turn on another MOSFET,
which can handle the current load. I know there's got to be a simpler way
to do it, again, I just can't think of what it is, and still keep the
project cheap. Oh, and I can't use a relay because the LED output is a PWM
signal, otherwise this would be a piece of cake... (sigh........)

Thanks for any ideas that come my way...
JDG
The 'D' stands for 'Damned if I can think of it'

 

[Dana Raymond, a minor God]

Divide all of the leds into seperate led strings with their own mosfet. The
PIC can drive many mosfet gates as they are very high impedance.

Why so many leds? If the leds are 'display' and not illumination, you may
want to consider multiplexing a large number of leds instead of what you are
doing. You will find tremendous power savings this way.

Hope this helps.
Dana Raymond

 

[Fritz Schlunder]

They're already divided into strings of 3 running from 12v, with suitable
current limiting resistors of course.

The point has been misunderstood (easily done considering my crappy
explanation).

What I need is a way to switch another set of mosfets (or whatever) using an
already existing set of mosfets (as already explained, have a large number
of controller units already built). BUT...the set that already exists only
applies a ground to a constant power supply (i.e. the output/led-strings
have 12v sitting on them just waiting for a ground supplied by the existing
mosfet).

So, I need something that'll switch on when I apply a ground to a power
that's already there. If it was just something I was switch on and off
occassionally, yes, I'd use a relay (with suitable clamps, etc) or an
opto-isolator running a relay, but these LED's are PWM controlled for
brightness at about 80Hz (8 bit 15 channel software PWM). Don't want to do
that mechanically. It just sounds bad...

JDG
The 'J' stands for 'Just can't wrap my brain around it'

Your explanation is still confusing to me, but it sounds to me like you need
a P-channel MOSFET, that way you can avoid using opto-isolators and high
side N-channel devices. 12V is just jim dandy for driving most P-channel
devices.

If heating in your N-channel MOSFETs is the problem then it just sounds like
you should use bigger devices. Surely a pin for pin compatible replacement
could be used with lower Rds(on). Just how much current and how many LEDs
are you switching here? Just what type of MOSFETs are you using (what part
number), and how much current is each one handling?

 

[Jeremy D. Grotte]

They're already divided into strings of 3 running from 12v, with suitable
current limiting resistors of course.

The point has been misunderstood (easily done considering my crappy
explanation).

What I need is a way to switch another set of mosfets (or whatever) using an
already existing set of mosfets (as already explained, have a large number
of controller units already built). BUT...the set that already exists only
applies a ground to a constant power supply (i.e. the output/led-strings
have 12v sitting on them just waiting for a ground supplied by the existing
mosfet).

So, I need something that'll switch on when I apply a ground to a power
that's already there. If it was just something I was switch on and off
occassionally, yes, I'd use a relay (with suitable clamps, etc) or an
opto-isolator running a relay, but these LED's are PWM controlled for
brightness at about 80Hz (8 bit 15 channel software PWM). Don't want to do
that mechanically. It just sounds bad...

JDG
The 'J' stands for 'Just can't wrap my brain around it'

 

[James Meyer]

I designed and built a circuit .....

The problem is that my circuit (which is already built in mass quantity)

You "designed" and built mass quantities and *now* you need to get it to
work? There's the beginning of a good Dilbert comic strip in there somewhere.

Jim

 

[A E]

You "designed" and built mass quantities and *now* you need to get it to
work? There's the beginning of a good Dilbert comic strip in there somewhere.

Jim

Oh well, if there's one thing I've learned about the workplace is that there's never
time and money to do it right, but there's always time and money to do it twice.

 

[Spehro Pefhany]

I designed and built a circuit to control a bunch of LED's awhile back. It
uses a PIC16F628 to control a few n-channel MOSFETs, and those mosfets in
turn apply a ground to strings of LED's (which naturally have a power
sitting on them waiting to be grounded) which are brightness controlled by
PWM. The circuit works fine until I start loading it down with loads of
LED's, then everything starts to get hot, wiring, MOSFETs, all of it.
Normally this isn't a problem for my project, but in extreme cases (when I
start putting a zillion LED's on it), I'm going to want to design a sort of
'power distribution box'. The idea being that the MOSFETs control another
larger MOSFET, or something, somewhere else. The problem is that my circuit
(which is already built in mass quantity) applies a 'ground to a power', not
a 'power to ground' (if that makes any sense), so I guess you could say I'm
using low-side switching (if that's the correct term?).
I know the answer to this is simple, but I can't seem to wrap my brain
around it at the moment.

The only idea I can come up with is using the outputs I've already got to
switch an opto-isolator, and then use that output to turn on another MOSFET,
which can handle the current load. I know there's got to be a simpler way
to do it, again, I just can't think of what it is, and still keep the
project cheap. Oh, and I can't use a relay because the LED output is a PWM
signal, otherwise this would be a piece of cake... (sigh........)

Can you quantify this "zillions" or more importantly the current and
voltage involved?

A single power logic-level (IMPORTANT) MOSFET can switch a lot of
current with a bit of heatsink. For example, a IRL2203 will switch 10A
with a very small, or possibly no, heatsink (depending on maximum
ambient temperature and package). Keep the PWM frequency low if you're
not using a MOSFET driver, a few hundred Hz is sufficient. At 20mA per
LED, that's 500 LEDs. If you really have a lot of high current LEDs,
then you could use something like a TC429 to drive the gates. At 6A
peak output, it would drive any reasonable number of gates fast enough
for LED PWM. You could also use a higher voltage supply and use
cheaper non-logic-level MOSFETs for the same performance.

Best regards,
Spehro Pefhany

 

[Jonathan Kirwan]

They're already divided into strings of 3 running from 12v, with suitable
current limiting resistors of course.

The point has been misunderstood (easily done considering my crappy
explanation).

What I need is a way to switch another set of mosfets (or whatever) using an
already existing set of mosfets (as already explained, have a large number
of controller units already built). BUT...the set that already exists only
applies a ground to a constant power supply (i.e. the output/led-strings
have 12v sitting on them just waiting for a ground supplied by the existing
mosfet).

So, I need something that'll switch on when I apply a ground to a power
that's already there. If it was just something I was switch on and off
occassionally, yes, I'd use a relay (with suitable clamps, etc) or an
opto-isolator running a relay, but these LED's are PWM controlled for
brightness at about 80Hz (8 bit 15 channel software PWM). Don't want to do
that mechanically. It just sounds bad...

So, if I've understood, you have:

(1) an existing system and you want to keep your existing
mosfets on that system,

(2) you already have *some* LEDs which work just great with the
existing system,

(3) the number of existing mosfets is the right number for both
the previous system *and* the new system, in that if you are say
controlling 10 mosfets in different PWM methods and patterns,
you'd like the new system to leave all that unchanged,

(4) you need to use *more* LEDs hanging off of each mosfet,
perhaps for more light or to spatially expand the region being
highlighted by that "gang" of LEDs, and that's all you need to
do, but,

(5) things get too hot, so you want a simple solution to your
existing system to allow you to string some more LEDs into each
gang and drive them just as you have been, using the existing
mosfets.

In other words, you don't want to "fix" the driving circuit you
currently have but want to append a colostomy bag to each mosfet
driver, where the existing LEDs would go now, so that it uses
the lighter drive capability as input to a reefed-up driver
which will drive the new, greater number of LEDs hung onto it?

And that your mosfets are n-channel, with the string of LEDs
hooked between the 12V rail and the drain, which is where you
want this new driver to get it's control input?

Jon

 

[Jeremy D. Grotte]

Zillions...ok, a bit overkill.
The system normally runs a hundred of so LED's. When I get extreme, I'm
running a couple hundred LED's+.
And again, my crappy explanation bit me.
It's not really the MOSFETs getting hot, it's the traces on the PCB, the
wiring, etc.
JDG

Can you quantify this "zillions" or more importantly the current and
voltage involved?

A single power logic-level (IMPORTANT) MOSFET can switch a lot of
current with a bit of heatsink. For example, a IRL2203 will switch 10A
with a very small, or possibly no, heatsink (depending on maximum
ambient temperature and package). Keep the PWM frequency low if you're
not using a MOSFET driver, a few hundred Hz is sufficient. At 20mA per
LED, that's 500 LEDs. If you really have a lot of high current LEDs,
then you could use something like a TC429 to drive the gates. At 6A
peak output, it would drive any reasonable number of gates fast enough
for LED PWM. You could also use a higher voltage supply and use
cheaper non-logic-level MOSFETs for the same performance.

Best regards,
Spehro Pefhany

http://www.speff.com

 

[Jeremy D. Grotte]

So, if I've understood, you have:

(1) an existing system and you want to keep your existing
mosfets on that system,

(2) you already have *some* LEDs which work just great with the
existing system,

(3) the number of existing mosfets is the right number for both
the previous system *and* the new system, in that if you are say
controlling 10 mosfets in different PWM methods and patterns,
you'd like the new system to leave all that unchanged,

(4) you need to use *more* LEDs hanging off of each mosfet,
perhaps for more light or to spatially expand the region being
highlighted by that "gang" of LEDs, and that's all you need to
do, but,

(5) things get too hot, so you want a simple solution to your
existing system to allow you to string some more LEDs into each
gang and drive them just as you have been, using the existing
mosfets.

In other words, you don't want to "fix" the driving circuit you
currently have but want to append a colostomy bag to each mosfet
driver, where the existing LEDs would go now, so that it uses
the lighter drive capability as input to a reefed-up driver
which will drive the new, greater number of LEDs hung onto it?

And that your mosfets are n-channel, with the string of LEDs
hooked between the 12V rail and the drain, which is where you
want this new driver to get it's control input?

Jon

Yep, so far so good! Never heard it referred to as a colostomy bag..of
tricks!
I'll try and post a schematic of the circuit later today, maybe that'll help
my case. If my drawing is as bad as my verbal explanation, it won't do much
good.
Points 1-5 ... right on the money...
If I was switching power, I wouldn't have a problem, but as stated, I'm
switching a ground in and out for each channel, with a constant power. The
MOSFET's are logic-level n-channel SOT-223. When switched on they've only
got an average of .2-.4v drop across them. They do get warm, but not warm
enough to worry about. It's the rest of the PCB, specifically the
protection diodes and the PCB's power rails, which get hot under a good
load. Yes, I used too small of diodes, but don't have the room to put
larger ones in (tight fit SMT board).
What else could I possibly forget?
JDG

 

[Jeremy D. Grotte]

Ok, I think I'm getting the jist of this...I think...probably not but here
goes...

I'm not on anybodys clock but my own (problem #1 probably!), I just didn't
think they'd work as well as they do, and I'm going a bit overboard with
them. I did calculate everything beforehand...for the lesser number of
LED's. Basically this is an 'expansion pack' or as I called it before a
'power distribution box', an add-on. The PCB can handle the current load on
them with the lesser load, but with the higher load, it starts getting too
hot for comfort, hence, the 'power distribution box'. And, yes, I am hoping
to line up a position eventually, get my foot in the door so to speak...(it
better work, otherwise it's bunch of $$$$ down the drain). And also, this
is worst case scenario, when the LED's are driven with 100% duty cycle for
an extended period. Anything less (roughly 50% duty cycle before you can
notice a dimming effect), and the boards don't get hot, but again, worst
case scenario.

So, I leave my LED strips the way they are, which is driven by 12-ish volts,
3 LEDs per 'string plus a current limiting resistor, 15 channels of these,
but I'll only talk about a single channel since they are all identical. The
controller board's MOSFET (n-channel logic level enhancement, STN2E10L) has
the gate switched by the PIC, the source goes direct to ground, and drain
goes out to the LED strings, which have Vdd sitting on them waiting for the
ground. Also, the protection diode is just a 1N4004 for polarity protection
and a 35v zener just in case I get some of those spikes that tend to happen
once in awhile.

Since this will be an add-on 'box', I take the output I've already got,
which is connected to the MOSFET's drain, feed that into a resistor, say 10k
for arguments sake. I pick off from the drain end of the resistor, feed
that into the gate of a P-channel enhancement MOSFET (don't need logic level
since it'll be a switched 12v, say a STP80PF55), connect the drain of the
P-channel to ground, connect the source back out to the LED strings. And
using much bigger traces of course, the P-channel MOSFET is a TO-220AB, so
no problems there.

I must be missing something about MOSFETs in general though, 'cause I'm not
seeing how this would turn the P-channel MOSFET on and off. I'm a rank
amateur when it comes to analog electronics, but I'm learning with my brand
new copy of 'The Art Of Electronics' (outstanding book).
Near as I can figure, an N-channel turns on when the gate is + with respect
to the source as long as the source is negative with respect to the drain.
The P-channel turns on when the gate is - with respect to the source as long
as the source is + with respect to the drain? Correct? Backwards?
Upside-down? If so, you basically take an n-channel, turn it upside down on
the schematic (and flip the little arrow inside around), reverse the gate
polarity and you're good to go?

N-channel:
On: Ids > 0 (and Rds < infinity) if Vgs > 0 and Vds > 0

P-channel:
On: Ids > 0 ( and Rds < infinity) if Vgs < 0 and Vds < 0

Therefore, with the n-channel off, the pickoff from the n-channel's drain is
sitting at Vcc, the p-channel's gate is at Vcc (p-ch Vgs = 0), the
p-channel's source is at Vcc, the p-channel's drain is at ground (p-ch Vds
= -12), there the p-channel is off (Ids = 0). Then when the n-channel gets
turned on, the pickoff from the n-channel's drain effectively becomes
grounded, the p-channel's gate goes to ground (p-ch Vgs= -12), the
p-channel's source is still at Vcc, the p-channel's drain is still at ground
(p-ch Vds roughly 0 since Rds still exists), therefore current flows thru
the p-channels source/drain. What did I miss? Something I'm sure...but I
can't put my finger on it...

So, basically, all I've created here is an inverter...but an inverter which
can handle a large load... how am I doing?

JDG
The 'G' stands for keeps 'Going and Going and Going...)

 

[Jeremy D. Grotte]

I just made up a quicky schematic of my project. Get to it at
www.ndak.net/~jdgrotte/powerdistbox1.pdf
I put in what I've already got (minus the PIC and associated components,
just the MOSFET's, LEDs, connectors, resistors, etc.), and what I think I
need to add based on what you wrote below.
JDG

 

[Jim Thompson]

I just made up a quicky schematic of my project. Get to it at
www.ndak.net/~jdgrotte/powerdistbox1.pdf
I put in what I've already got (minus the PIC and associated components,
just the MOSFET's, LEDs, connectors, resistors, etc.), and what I think I
need to add based on what you wrote below.
JDG

[snip]

P-channel device is not correctly connected.

Tie SOURCE to VCC2, DRAIN to Load, Loads to Ground.

...Jim Thompson

 

[Active8]

I just made up a quicky schematic of my project. Get to it at
www.ndak.net/~jdgrotte/powerdistbox1.pdf
I put in what I've already got (minus the PIC and associated components,
just the MOSFET's, LEDs, connectors, resistors, etc.), and what I think I
need to add based on what you wrote below.
JDG

[snip]

P-channel device is not correctly connected.

Tie SOURCE to VCC2, DRAIN to Load, Loads to Ground.

...Jim Thompson

why can't he use source-follower?

mike

 

[Jim Thompson]

I just made up a quicky schematic of my project. Get to it at
www.ndak.net/~jdgrotte/powerdistbox1.pdf
I put in what I've already got (minus the PIC and associated components,
just the MOSFET's, LEDs, connectors, resistors, etc.), and what I think I
need to add based on what you wrote below.
JDG

[snip]

P-channel device is not correctly connected.

Tie SOURCE to VCC2, DRAIN to Load, Loads to Ground.

...Jim Thompson

why can't he use source-follower?

mike

He has a follower fixation, but no gate drive to support it. Remember
his assertion (IIRC) he only got 2V out when he used an NMOS follower
;-)

...Jim Thompson

 

[Active8]

On Sat, 2 Aug 2003 03:55:26 -0600, "Jeremy D. Grotte"

I just made up a quicky schematic of my project. Get to it at
www.ndak.net/~jdgrotte/powerdistbox1.pdf
I put in what I've already got (minus the PIC and associated components,
just the MOSFET's, LEDs, connectors, resistors, etc.), and what I think I
need to add based on what you wrote below.
JDG

[snip]

P-channel device is not correctly connected.

Tie SOURCE to VCC2, DRAIN to Load, Loads to Ground.

...Jim Thompson

why can't he use source-follower?

mike

He has a follower fixation, but no gate drive to support it. Remember
his assertion (IIRC) he only got 2V out when he used an NMOS follower
;-)

...Jim Thompson

all i see is that he gets a .2-.4V Vds(on) and was talking about not
having a problem if he was "switching power" i.e., common drain (source-
follower) and he'd have close to 12V at the source had he done that and
could then drive another n-ch stage.

so with less than a volt at the drain of the n-ch device, he has enough
to drive a p-ch device in source-follower config...

ok, i see it, the drop across the load would lower the p-ch source-
follower's Vgs, duh!

solly

mike

 

[Active8]

wow! you did magnitudes better paraphrasing what i told you than you did
communicating the problem in the first place with a couple/few
exceptions. read on.

Ok, I think I'm getting the jist of this...I think...probably not but here
goes...

I'm not on anybodys clock but my own (problem #1 probably!), I just didn't
think they'd work as well as they do, and I'm going a bit overboard with
them. >

eh? they?

I did calculate everything beforehand...for the lesser number of
LED's. Basically this is an 'expansion pack' or as I called it before a
'power distribution box', an add-on. The PCB can handle the current load on
them with the lesser load, but with the higher load, it starts getting too
hot for comfort, hence, the 'power distribution box'. And, yes, I am hoping
to line up a position eventually, get my foot in the door so to speak...(it
better work, otherwise it's bunch of $$$$ down the drain).

leave a jimmie in it for me, would you? contractual work preferred. get
a mgmt position or the ear of one. it's not Fred Leuchter, is it? John
Maye? DART, Tesla, or other psuedoscience?

as for the $$... too bad polypacks isn't still around. you could've
gotten a big bag of leds for $1. can't remember what they called those
grab bags.

And also, this
is worst case scenario, when the LED's are driven with 100% duty cycle for
an extended period. Anything less (roughly 50% duty cycle before you can
notice a dimming effect), and the boards don't get hot, but again, worst
case scenario.

100% DC would be part of that, yes. artic/equatorial temps another.

So, I leave my LED strips the way they are,

no, i hate to tell you this, but i wanted them wrecked out, still do,
and i'll explain why shortly.

which is driven by 12-ish volts,
3 LEDs per 'string plus a current limiting resistor,

even more reason to wreck them out. i thought they might be paralleled
and you just miscommunicated. 3 leds in series will always drop 3xVf, so
if they're 1.5V leds, that's 4.5V dropped across them which will leave
the n-ch drain at 12 - 4.5 = 7.5V and that is the -Vgs that the p-ch
MOSFET will see. even -1.5V Vgs is a bad idea. you want -Vgs of the p-ch
device to be 0V when the device is supposed to be off.

15 channels of these,
but I'll only talk about a single channel since they are all identical. The
controller board's MOSFET (n-channel logic level enhancement, STN2E10L) has
the gate switched by the PIC, the source goes direct to ground, and drain
goes out to the LED strings, which have Vdd sitting on them waiting for the
ground. Also, the protection diode is just a 1N4004 for polarity protection
and a 35v zener just in case I get some of those spikes that tend to happen
once in awhile.

and the 1N4004 gets hot. i'll treat that later.

Since this will be an add-on 'box', I take the output I've already got,
which is connected to the MOSFET's drain, feed that into a resistor, say 10k
for arguments sake.

minus the existing LEDs

I pick off from the drain end of the resistor,

use courier font to view

vcc |
|

< R

<
|
|
|
|______ to p-ch gate
|
|
n-ch drain

that was an ugly font, last art i viewed looked better with it.

well, you managed to draw it out before i could finish my reply. so
anyway... good morning...

feed
that into the gate of a P-channel enhancement MOSFET (don't need logic level
since it'll be a switched 12v, say a STP80PF55), connect the drain of the
P-channel to ground, connect the source back out to the LED strings. And
using much bigger traces of course,

as Jim said, stick with common-source config. sorry. i'd have caught
that had i drawn it out.

the P-channel MOSFET is a TO-220AB, so
no problems there.

the other end of the leds comes back to a current limiting or ballast R,
the other end of which connects to Vcc.

i ususlly connect my LED resistors to the supply and the LEDs to the
driver, but who cares.

now with 4.5V across the LEDs or more for higher voltage LEDs, you only
have 7.5V left across the resistor and MOSFET, so 2.5V LEDs would be 12
- 7.5 = 4.5 ...

you'd have Id (100% DC or full on) Id X R = Vr, so Vr + Vled + Vds = 12V

i guess you worked that out for the existing design. we're just allowing
for more current.

I must be missing something about MOSFETs in general though, 'cause I'm not
seeing how this would turn the P-channel MOSFET on and off.

you sort of answer that youself, below

I'm a rank
amateur

and the job position is what?

when it comes to analog electronics, but I'm learning with my brand
new copy of 'The Art Of Electronics' (outstanding book).

that guy gets a lot of advertisement here. deservedly so, i guess. the
book has its uses. gets you pointed in the right direction, at least.

Near as I can figure, an N-channel turns on when the gate is + with respect
to the source as long as the source is negative with respect to the drain.
The P-channel turns on when the gate is - with respect to the source as long
as the source is + with respect to the drain? Correct? Backwards?
correct

Upside-down? If so, you basically take an n-channel, turn it upside down on
the schematic (and flip the little arrow inside around), reverse the gate
polarity and you're good to go?

just like with bipolars. i always thought of it as flipping the
transistor, load, and bias while keeping the supply fixed. easier to see
in a complementary pair amp with say 12V-0- -12V. you can see the
symetry that way.[snip verbosity]

good enough

JDG
The 'G' stands for keeps 'Going and Going and Going...)

and you'll explain the 'D' soon, i suppose. just remember the whole
current path of the load needs beefed up traces.

the 1N4004 will need to be replaced with whatever handles the current.

 

[Active8]

and you'll explain the 'D' soon, i suppose. just remember the whole
current path of the load needs beefed up traces.

the 1N4004 will need to be replaced with whatever handles the current.

nothing follows

BRs,
mike

 

[Jim Thompson]

I updated the schematic a bit...same address
www.ndak.net/~jdgrotte/powerdistbox1.pdf
Swapped around the LED connections to the p-channel mosfet.
P-channel MOSFET = IRF9Z24N...good choice?
JDG

[snip]

OK, Your schematic looks *much* better.

I can't comment on the IRF9Z24N though, I couldn't find a data sheet
for it... probably OK.

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice480)460-2350 | |
| Jim-T@analog_innovations.com Fax480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |

For proper E-mail replies SWAP "-" and "_"

I love to cook with wine. Sometimes I even put it in the food.

 

[Jeremy D. Grotte]

Comment to the reply to the answer to the post....etc....

wow! you did magnitudes better paraphrasing what i told you than you did
communicating the problem in the first place with a couple/few
exceptions. read on.


eh? they?

they = the project in general, the controller boxes, the led's, etc...

leave a jimmie in it for me, would you? contractual work preferred. get
a mgmt position or the ear of one. it's not Fred Leuchter, is it? John
Maye? DART, Tesla, or other psuedoscience?

as for the $$... too bad polypacks isn't still around. you could've
gotten a big bag of leds for $1. can't remember what they called those
grab bags.


100% DC would be part of that, yes. artic/equatorial temps another.

Well, duty cycle was one part of it. I also put them inside a box (no
airflow) and outside in the sun (high temps), and ran them at 15v (high
volts). This was about as close to worst case scenario as I could come up
with.

no, i hate to tell you this, but i wanted them wrecked out, still do,
and i'll explain why shortly.


even more reason to wreck them out. i thought they might be paralleled
and you just miscommunicated. 3 leds in series will always drop 3xVf, so
if they're 1.5V leds, that's 4.5V dropped across them which will leave
the n-ch drain at 12 - 4.5 = 7.5V and that is the -Vgs that the p-ch

But the n-channel (which is a logic level type, low Rds even at 2v) will
only be 'seeing' the p-channel gate and a resistor...???
This matters???

MOSFET will see. even -1.5V Vgs is a bad idea. you want -Vgs of the p-ch
device to be 0V when the device is supposed to be off.

I keep leaving out key info. I'm using superflux LED's, all different
colors. They've usually got 3-4v drop across them. So, I would assume that
makes this even worse.

15 channels of these,
but I'll only talk about a single channel since they are all identical. The
controller board's MOSFET (n-channel logic level enhancement, STN2E10L) has
the gate switched by the PIC, the source goes direct to ground, and drain
goes out to the LED strings, which have Vdd sitting on them waiting for the
ground. Also, the protection diode is just a 1N4004 for polarity protection
and a 35v zener just in case I get some of those spikes that tend to happen
once in awhile.

and the 1N4004 gets hot. i'll treat that later.

Understood...

Since this will be an add-on 'box', I take the output I've already got,
which is connected to the MOSFET's drain, feed that into a resistor, say 10k
for arguments sake.

minus the existing LEDs

I pick off from the drain end of the resistor,

use courier font to view

vcc |
|

< R

<
|
|
|
|______ to p-ch gate
|
|
n-ch drain

that was an ugly font, last art i viewed looked better with it.

well, you managed to draw it out before i could finish my reply. so
anyway... good morning...

feed
that into the gate of a P-channel enhancement MOSFET (don't need logic level
since it'll be a switched 12v, say a STP80PF55), connect the drain of the
P-channel to ground, connect the source back out to the LED strings. And
using much bigger traces of course,

as Jim said, stick with common-source config. sorry. i'd have caught
that had i drawn it out.

the P-channel MOSFET is a TO-220AB, so
no problems there.

the other end of the leds comes back to a current limiting or ballast R,
the other end of which connects to Vcc.

i ususlly connect my LED resistors to the supply and the LEDs to the
driver, but who cares.

now with 4.5V across the LEDs or more for higher voltage LEDs, you only
have 7.5V left across the resistor and MOSFET, so 2.5V LEDs would be 12
- 7.5 = 4.5 ...

you'd have Id (100% DC or full on) Id X R = Vr, so Vr + Vled + Vds = 12V

i guess you worked that out for the existing design. we're just allowing
for more current.

I must be missing something about MOSFETs in general though, 'cause I'm not
seeing how this would turn the P-channel MOSFET on and off.

you sort of answer that youself, below

I'm a rank
amateur

and the job position is what?

when it comes to analog electronics, but I'm learning with my brand
new copy of 'The Art Of Electronics' (outstanding book).

that guy gets a lot of advertisement here. deservedly so, i guess. the
book has its uses. gets you pointed in the right direction, at least.

Near as I can figure, an N-channel turns on when the gate is + with respect
to the source as long as the source is negative with respect to the drain.
The P-channel turns on when the gate is - with respect to the source as long
as the source is + with respect to the drain? Correct? Backwards?
correct

Upside-down? If so, you basically take an n-channel, turn it upside down on
the schematic (and flip the little arrow inside around), reverse the gate
polarity and you're good to go?

just like with bipolars. i always thought of it as flipping the
transistor, load, and bias while keeping the supply fixed. easier to see
in a complementary pair amp with say 12V-0- -12V. you can see the
symetry that way.[snip verbosity]

good enough

JDG
The 'G' stands for keeps 'Going and Going and Going...)

and you'll explain the 'D' soon, i suppose. just remember the whole
current path of the load needs beefed up traces.

the 1N4004 will need to be replaced with whatever handles the current.