Maker Pro
Maker Pro

MOSFET help?

E

Ed

Jan 1, 1970
0
Hi All possibly quite a simple question but I would like some advice.

I have been looking at ways to use an IRFZ44N as a solid state 'switch'
to switch on and of the positive supply of a 12 to 24 volt supply. As
far as current requirements go the IRFZ44 is ideal, but I'm sturggling
to find a good way to do this.

To fill in more info its for a high current serial bus of 3 wires,
12-30v Ground and Data.
I need to beable to control the Data and Supply lines from a 5 volt
logic level as simply and as ruggidly as possible, yet both must be
capable of high currents, and minimal on resistance. I looked at using a
P channel fet, such as the IRF5210 thinking this may be a solution, but
I think I am going totally the wrong way.

Any suggestions would be a MASSIVE help.

Thanks,

Ed
 
S

Stefan Heinzmann

Jan 1, 1970
0
Ed said:
Hi All possibly quite a simple question but I would like some advice.

I have been looking at ways to use an IRFZ44N as a solid state 'switch'
to switch on and of the positive supply of a 12 to 24 volt supply. As
far as current requirements go the IRFZ44 is ideal, but I'm sturggling
to find a good way to do this.

I guess your problem is to supply the gate drive voltage to switch the
MOSFET on. As it is an N-channel device this voltage must be above the
source voltage. You're switching the "high side" of the supply, so the
gate drive voltage must ride above the power supply. There are ways to
do this, but it is not the simplest approach.
To fill in more info its for a high current serial bus of 3 wires,
12-30v Ground and Data.
I need to beable to control the Data and Supply lines from a 5 volt
logic level as simply and as ruggidly as possible, yet both must be
capable of high currents, and minimal on resistance. I looked at using a
P channel fet, such as the IRF5210 thinking this may be a solution, but
I think I am going totally the wrong way.

Using a P-channel MOSFET does indeed simplify things considerably, as
your gate drive voltage needs to be lower than the source voltage. It
still can not be driven directly from a logic output as the logic levels
are grond referenced whereas the MOSFET doesn't see any ground voltage.

So my recommendation would be to use a P-channel MOSFET together with a
level-shifter between the logic output and the MOSFET gate. This level
shifter can be as simple as a transistor (NPN) and a couple of resistors.
Any suggestions would be a MASSIVE help.

Here's a sketch of the idea (to be read using a monospace font):

MOSFET

Power in ------o------+^+----- Switched power out
| |||
| ===
| R1 |
| ___ |
+-|___|--o
|
|
___ |/
Control signal-----|___|---| NPN
|>
R2 |
|
|
===
GND

created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

R1 ensures that the MOSFET turns off when the NPN turns off. Its value
should be low enough to make the MOSFET switch off quickly enough and
high enough not to exceed the NPN's ratings. R2 limits the base current
of the NPN.
 
E

Ed

Jan 1, 1970
0
Stefan said:
I guess your problem is to supply the gate drive voltage to switch the
MOSFET on. As it is an N-channel device this voltage must be above the
source voltage. You're switching the "high side" of the supply, so the
gate drive voltage must ride above the power supply. There are ways to
do this, but it is not the simplest approach.



Using a P-channel MOSFET does indeed simplify things considerably, as
your gate drive voltage needs to be lower than the source voltage. It
still can not be driven directly from a logic output as the logic levels
are grond referenced whereas the MOSFET doesn't see any ground voltage.

So my recommendation would be to use a P-channel MOSFET together with a
level-shifter between the logic output and the MOSFET gate. This level
shifter can be as simple as a transistor (NPN) and a couple of resistors.



Here's a sketch of the idea (to be read using a monospace font):

MOSFET

Power in ------o------+^+----- Switched power out
| |||
| ===
| R1 |
| ___ |
+-|___|--o
|
|
___ |/
Control signal-----|___|---| NPN
|>
R2 |
|
|
===
GND

created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

R1 ensures that the MOSFET turns off when the NPN turns off. Its value
should be low enough to make the MOSFET switch off quickly enough and
high enough not to exceed the NPN's ratings. R2 limits the base current
of the NPN.
Thanks!!!! This is EXACTLY what I wanted to see! I'm new to using
mosfets and their tricks, my head was spinning from looking through so
much data the other day, it seemed such a daft thing to be stuck on.
Thanks for this info, its greatly appreachiated - I'll order some
devices now :) thanks.

Ill let you know how I get on :)

Ed
 
W

Winfield Hill

Jan 1, 1970
0
Stefan Heinzmann wrote...
Hi All possibly quite a simple question but I would like some advice.

I have been looking at ways to use an IRFZ44N as a solid state 'switch'
to switch on and of the positive supply of a 12 to 24 volt supply. As
far as current requirements go the IRFZ44 is ideal, but I'm sturggling
to find a good way to do this. [ snip comment ] To fill in more info
its for a high current serial bus of 3 wires, 12-30v Ground and Data.

I need to beable to control the Data and Supply lines from a 5 volt
logic level as simply and as ruggidly as possible, yet both must be
capable of high currents, and minimal on resistance. I looked at using
a P channel fet, such as the IRF5210 thinking this may be a solution,
but I think I am going totally the wrong way.

Using a P-channel MOSFET does indeed simplify things considerably, as
your gate drive voltage needs to be lower than the source voltage. It
still can not be driven directly from a logic output as the logic levels
are grond referenced whereas the MOSFET doesn't see any ground voltage.

So my recommendation would be to use a P-channel MOSFET together with a
level-shifter between the logic output and the MOSFET gate. This level
shifter can be as simple as a transistor (NPN) and a couple of resistors.

Here's a sketch of the idea (to be read using a monospace font):

MOSFET

Power in ------o------+^+----- Switched power out
| |||
| ===
| R1 |
| ___ |
+-|___|--o
|
|
___ |/
Control signal-----|___|---| NPN
|>
R2 |
|
===
GND

R1 ensures that the MOSFET turns off when the NPN turns off. Its value
should be low enough to make the MOSFET switch off quickly enough and
high enough not to exceed the NPN's ratings. R2 limits the base current
of the NPN.

This configuration may be useful for switching fixed low-voltage
power supplies, but if the supply voltage is above 15 to 20V, the
FET's rated gate voltage will be exceeded. Furthermore, the gate
voltage is not predictable. A good general purpose configuration
is to develop fixed p-channel FET gate voltages by switching fixed
currents into the gate resistor, like this:

| p-type FET
| Power in ------o------, ,----- Switched power out
| 15 - 50V | S_|__|_
| (65V max) | ,---
| | |
| '- R1 -+ for 10V gate drive, use
| 1.0k | R1 = R2 (10/4.3) = 2.3 R2
| |
| |/
| 5V logic-control -----| 60V NPN BJT
| |\V 2n4401
| |
| R2 level-shift current-sink
| 430 high-side FET switching
| |
| GND


The fixed gate drive current of this configuration means that it's
easy to add a current-limit feature, unlike the simple CE switch.

| p-type FET
| Power in ---o-- Rs --+----, ,----- Switched power out
| 15 - 50V | | S_|__|_
| (65V max) | 2n4403_|_ ,--- current limit
| |______>/ \___| I_cl = 0.65 / Rs
| | |
| '---- R1 -----+
| 1.0k |
| |
| |/
| 5V logic-control -----| 60V NPN BJT
| |\V 2n4401
| |
| R2
| 430 ohms
| |
| GND

Actually, it's a wise idea to use foldback current limiting to
reduce the C.L. dissipation for the FET. Although this takes only
two more parts (as we show in AoE), two parts are time costly in
s.e.d. ASCII drawings, and one's intrinsic laziness prevails.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)
 
R

Robert Baer

Jan 1, 1970
0
Ed said:
Hi All possibly quite a simple question but I would like some advice.

I have been looking at ways to use an IRFZ44N as a solid state 'switch'
to switch on and of the positive supply of a 12 to 24 volt supply. As
far as current requirements go the IRFZ44 is ideal, but I'm sturggling
to find a good way to do this.

To fill in more info its for a high current serial bus of 3 wires,
12-30v Ground and Data.
I need to beable to control the Data and Supply lines from a 5 volt
logic level as simply and as ruggidly as possible, yet both must be
capable of high currents, and minimal on resistance. I looked at using a
P channel fet, such as the IRF5210 thinking this may be a solution, but
I think I am going totally the wrong way.

Any suggestions would be a MASSIVE help.

Thanks,

Ed

As far as using an N-channel FET as a series power switch, i would use
a "high-side" driver, like the IR2117 or IR2118 (one is inverting, the
other is not).
They will accept a logic level for input, and the isolated gate driver
is ideal for this useage.

For multiple channels in one IC, look at what Supertex has: HV20220 8
channel, HV209 12 channel, or the HV20822 analog switches.
They also have 16 and 32 channel switches, some with latches, etc.
Something they mkae might be useable for your needs.
 
W

Winfield Hill

Jan 1, 1970
0
Robert Baer wrote...
As far as using an N-channel FET as a series power switch, i would use
a "high-side" driver, like the IR2117 or IR2118 (one is inverting, the
other is not). They will accept a logic level for input, and the
isolated gate driver is ideal for this useage.

One caution, those are for periodic (e.g. PWM) switching, and don't
work for continuous switched FETs, unless the swinging gate-power
diode & capacitor is replaced with a separate isolated power supply.
For multiple channels in one IC, look at what Supertex has: HV20220
8 channel, HV209 12 channel, or the HV20822 analog switches.
They also have 16 and 32 channel switches, some with latches, etc.
Something they mkae might be useable for your needs.

Aren't those 200V signal switches, 22-ohm series resistance, etc.?

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)
 
E

Ed

Jan 1, 1970
0
Winfield said:
This configuration may be useful for switching fixed low-voltage
power supplies, but if the supply voltage is above 15 to 20V, the
FET's rated gate voltage will be exceeded. Furthermore, the gate
voltage is not predictable. A good general purpose configuration
is to develop fixed p-channel FET gate voltages by switching fixed
currents into the gate resistor, like this:

| p-type FET
| Power in ------o------, ,----- Switched power out
| 15 - 50V | S_|__|_
| (65V max) | ,---
| | |
| '- R1 -+ for 10V gate drive, use
| 1.0k | R1 = R2 (10/4.3) = 2.3 R2
| |
| |/
| 5V logic-control -----| 60V NPN BJT
| |\V 2n4401
| |
| R2 level-shift current-sink
| 430 high-side FET switching
| |
| GND


The fixed gate drive current of this configuration means that it's
easy to add a current-limit feature, unlike the simple CE switch.

| p-type FET
| Power in ---o-- Rs --+----, ,----- Switched power out
| 15 - 50V | | S_|__|_
| (65V max) | 2n4403_|_ ,--- current limit
| |______>/ \___| I_cl = 0.65 / Rs
| | |
| '---- R1 -----+
| 1.0k |
| |
| |/
| 5V logic-control -----| 60V NPN BJT
| |\V 2n4401
| |
| R2
| 430 ohms
| |
| GND

Actually, it's a wise idea to use foldback current limiting to
reduce the C.L. dissipation for the FET. Although this takes only
two more parts (as we show in AoE), two parts are time costly in
s.e.d. ASCII drawings, and one's intrinsic laziness prevails.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

Even more excellent information. So this looks like I can extend the
supply range further IF needed which is great. I have found a high
current P-Fet "SUP65P06-20". Which I have ordered a few samples.
Hopefully armed with this info I shall now make some good progress.

Thanks!

Ed
 
W

Winfield Hill

Jan 1, 1970
0
Ed wrote...
Even more excellent information. So this looks like I can extend the
supply range further IF needed which is great. I have found a high
current P-Fet "SUP65P06-20". Which I have ordered a few samples.
Hopefully armed with this info I shall now make some good progress.

If? You mentioned using a 24V supply. Since that's over 20V, which
is the FET gate-voltage limit, you will need this configuration or
an equivalent. If you're using a 12V battery, which is actually a
12 to 14.5V source, with sag to 8V, Stefan's configuration would be
more suitable. The circuit above provides 12-4.3 = 5.7V of FET gate
drive for 12V input, and only 3.7V for 8V in.

The SUB65P06-20 is a large-die-area FET, rated at 250W mounted on a
25C heat sink. Used as a switch it's capable of continuously passing
up to 40A. For load currents above 5 to 10A, fault protection for
the FET is certainly in order, such as a foldback current limit, and
perhaps a thermal cutout mounted on the heat sink as well. Ed, tell
us more about application, your load, and its current requirements.

Some may wonder at the low-value of the resistors, creating a 10mA
gate-switching current. But large FETs have high gate capacitance,
e.g. Ciss = 4500pF for the SUP65, where even a low 1k gate resistor
still yields a relatively slow 5us time constant.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)
 
J

Joerg

Jan 1, 1970
0
Hi Winfield,
If the power line out has any length this might need an additional cap
from gate to source or some other means of suppressing oscillation.

Regards, Joerg
 
K

Ken Smith

Jan 1, 1970
0
Hi Winfield,

If the power line out has any length this might need an additional cap
from gate to source or some other means of suppressing oscillation.

Oscillation smoscillation!

When the MOSFET is fully on, the VDS is near zero and hence so is the
gain. When the MOSFET is fully off, the Id is near zero and hence so is
the gain. While the MOSFET is turning off or on, the FCC isn't likely to
be looking.

It would be more important to limit the gate voltage swing.

BTW: Has anyone suggested one of the photovoltaic isolators for this job?
It would turn on an N channel FET.
 
J

Joerg

Jan 1, 1970
0
Hi Ken,
Oscillation smoscillation!

When the MOSFET is fully on, the VDS is near zero and hence so is the
gain. When the MOSFET is fully off, the Id is near zero and hence so is
the gain. While the MOSFET is turning off or on, the FCC isn't likely to
be looking.
That's right, Ken. The FCC might not see it. But when the TV glitches
out of sync everytime the FET switches that can cause some nasty
outbursts. Been there when I was a kid, me happily switching that brand
new hot rod FET and my father trying to watch a soccer game :) When
this happened at a friends house his father actually kicked us out for
that afternoon. Seriously.

Regards, Joerg
 
K

Ken Smith

Jan 1, 1970
0
Hi Ken,

That's right, Ken. The FCC might not see it. But when the TV glitches
out of sync everytime the FET switches that can cause some nasty
outbursts. Been there when I was a kid, me happily switching that brand
new hot rod FET and my father trying to watch a soccer game :) When
this happened at a friends house his father actually kicked us out for
that afternoon. Seriously.

There are two things your reply indicates:

(1)
You aren't as old as me: FET <-> kid unless it was a glass FET, this
wasn't too long ago.

(2)
You were dumb enough to get caught.

But that aside:

The best way to prevent oscillation in MOSFETs is to use a lossy element
like a resistor or RF bead. The source to gate capacitance idea can just
serve to change the frequency unless it is large enough to needlessly slow
the switching.
 
J

Joerg

Jan 1, 1970
0
Hi Ken,
You aren't as old as me: FET <-> kid unless it was a glass FET, this
wasn't too long ago.
I am 46. We didn't have really big FETs in those days, at least not ones
we could afford. But we made big ones by scrapping BF245 and other stuff
from reject boards and placing tons of those in parallel. That in itself
was an invitation to become an RF generator unless in the on or off
state. Tubes were much easier and had a lot more oomph per Dollar so
that's really what I grew up with.
You were dumb enough to get caught.
Yeah, the TV caught us :-(
The best way to prevent oscillation in MOSFETs is to use a lossy element
like a resistor or RF bead. The source to gate capacitance idea can just
serve to change the frequency unless it is large enough to needlessly slow
the switching.
Yes, beads are great. We used to scrap those out of TV sets when I was a
kid but of course back then I didn't have a clue that you could slide
them over transistor pins to quiet things down.

Regards, Joerg
 
K

Ken Smith

Jan 1, 1970
0
Joerg said:
state. Tubes were much easier and had a lot more oomph per Dollar so
that's really what I grew up with.

Yes for about $6, or a dumpster dive after an old radio, you could get a
6V6 or 6L6. They could make about 14W at 1MHz which isn't too bad when
you think about it.

I know some reader just said "so that was you!". Honest it wsn't, honest.

Tubes had the advantage that failure was either thermal or flashover. I
learned quickly to cleans the outsides of the tubes. An RF stage has a
peak voltage about twice the supply so the tube may have 2000V on it.
 
E

Ed

Jan 1, 1970
0
Winfield said:
Ed wrote...



If? You mentioned using a 24V supply. Since that's over 20V, which
is the FET gate-voltage limit, you will need this configuration or
an equivalent. If you're using a 12V battery, which is actually a
12 to 14.5V source, with sag to 8V, Stefan's configuration would be
more suitable. The circuit above provides 12-4.3 = 5.7V of FET gate
drive for 12V input, and only 3.7V for 8V in.

Yes the fact that a battery is as much as 14.5 when charged is why I
wanted to manage voltages as much as 30 since thats pretty close to what
two series fully charged would be. The components I ordered are now
here, and I shall put something together after I am back from lunch.
The SUB65P06-20 is a large-die-area FET, rated at 250W mounted on a
25C heat sink. Used as a switch it's capable of continuously passing
up to 40A. For load currents above 5 to 10A, fault protection for
the FET is certainly in order, such as a foldback current limit, and
perhaps a thermal cutout mounted on the heat sink as well. Ed, tell
us more about application, your load, and its current requirements.

Some may wonder at the low-value of the resistors, creating a 10mA
gate-switching current. But large FETs have high gate capacitance,
e.g. Ciss = 4500pF for the SUP65, where even a low 1k gate resistor
still yields a relatively slow 5us time constant.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

The application is simply this. I'm building a microcontroller based
pyrotechnic firing system. The fet is needed to switch on and off the
supply to isolate the modules from the pyro. Since typically these
systems will be running from battery, and it involves reasonably long
wire lengths I want to introduce as little voltage drop as possible.

I would imagine I could create a zener gate voltage limiter with a
suitable resistor? This should solve the problem I would imagine, and
surely allow me to operate the fet at a suitable gate voltage level
across a wide range of supply rails?

Ed
 
E

Ed

Jan 1, 1970
0
Joerg said:
Hi Ken,

That's right, Ken. The FCC might not see it. But when the TV glitches
out of sync everytime the FET switches that can cause some nasty
outbursts. Been there when I was a kid, me happily switching that brand
new hot rod FET and my father trying to watch a soccer game :) When
this happened at a friends house his father actually kicked us out for
that afternoon. Seriously.

Regards, Joerg

How amusing :) I was going to add this anyway. I have a 250meg spectrum
analyser which I shall check things on if I suspect anything strange is
happening :) If I'm still not sure I can get hold of a 8gig Advantest
one, but I'm sure I'll have no problems :)

Ed
 
J

Joerg

Jan 1, 1970
0
Hi Ed,
How amusing :) I was going to add this anyway. I have a 250meg
spectrum analyser which I shall check things on if I suspect anything
strange is happening :) If I'm still not sure I can get hold of a 8gig
Advantest one, but I'm sure I'll have no problems :)

If there are any oscillations it'll be very short bursts while switching
so a spectrum analyzer might not show much. But a fast digital (high
single-shot sample rate) scope would.

Regards, Joerg
 
W

Winfield Hill

Jan 1, 1970
0
Joerg wrote...
If there are any oscillations it'll be very short bursts while
switching so a spectrum analyzer might not show much. But a fast
digital (high single-shot sample rate) scope would.

I routinely examine the FET switching waveforms in my designs
with a 500MHz 2GS/s scope when debugging them, and have not
seen this problem. Perhaps that's because I use short wiring
leads along with low-inductance ceramic bypass caps. I *have*
seen severe RF oscillation in paralleled linear power MOSFETs.
This is a major issue, especially with high-voltage FETs.

Note, the circuit I suggested had rather low resistor values,
creating a 1 to 2us switching time. This is fairly fast and
provides a rapid slew rate through any vulnerable regions.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)
 
E

Ed

Jan 1, 1970
0
Joerg said:
Hi Ed,



If there are any oscillations it'll be very short bursts while switching
so a spectrum analyzer might not show much. But a fast digital (high
single-shot sample rate) scope would.

Regards, Joerg

Hmm I have access to a 100mhz digital scope. Ive no idea if this would
be suitable. I shall just do my best with stopping such things happening
in the first place. Thanks for the advice though, I totally forgot that
the time scale of these oscillations if they happen would be very short
indeed.

(Boy this newsgroup makes me feel dumb!)

Thanks,

Ed
 
Top