Maker Pro
Maker Pro

FET driver recommendation needed

L

Lowly Engineer

Jan 1, 1970
0
I am "high-side" switching a P-FET at several hundred Khz. The source
is connected to +12V and I need to pull the gate to Vsource minus 10V
or so to fully turn it on. I'm using a ground-referenced, 5V swing PWM
circuit to perform the switching.

I need to really "slam" the gate of the FET to reduce the switching
time. In the past, I used a high-speed op-amp to provide the
necessary gate drive, but don't feel comfortable with this approach
because the output current required stresses the op-amp, and the 5V
swing probably doesn't make the input too happy either.

Does anyone know of a SMT FET driver that has a push-pull output,
100mA of drive and can tolerate a 16V supply? Any other ideas?


TIA
 
W

Winfield Hill

Jan 1, 1970
0
Lowly wrote...
I am "high-side" switching a P-FET at several hundred Khz. The source
is connected to +12V and I need to pull the gate to Vsource minus 10V
or so to fully turn it on. I'm using a ground-referenced, 5V swing PWM
circuit to perform the switching.

I need to really "slam" the gate of the FET to reduce the switching
time. In the past, I used a high-speed op-amp to provide the
necessary gate drive, but don't feel comfortable with this approach
because the output current required stresses the op-amp, and the 5V
swing probably doesn't make the input too happy either.

Does anyone know of a SMT FET driver that has a push-pull output,
100mA of drive and can tolerate a 16V supply? Any other ideas?

I don't see how you come up with the 16V number, but standard FET
driver ICs can operate with supplies as high as 20V. And most FETs
are rated for 20V maximum gate voltage, although some types are
rated at 30V. The obvious and simple solution for you is to use a
common driver IC powered from the same 12V as your P-type FET's
source. The driver will switch to ground, creating 12V gate pulses,
which are fine.

The driver will happily operate directly off your ground-referenced
5V or TTL-logic PWM input signal.

You can use one of the parts from the Microchip (formerly Telcom)
TC4426 to 4428 series. These are among the lowest-current parts
offered, and are available in soic packages. They can deliver up
to 1.5A to the gate, but it's customary to add a series resistor.
If the resistor is large enough, it'll reduce the peak gate-drive
current and thereby slow the FET's switching speed, if you like.

Now, if your 12V is really a car battery, that's another matter.
An automobile's 12V can experience surges up to 60V, which must
be dealt with and which are not suited for high-side p-type FETs.
Instead the industry-standard solution is to use high-side n-type
FETs with driver ICs that create a gate voltage referenced to the
FET's source pin.

Thanks,
- Win
 
W

Winfield Hill

Jan 1, 1970
0
Tilmann Reh wrote...
... and if you need one: look at the IR2117.

(This also applies to other circuits with voltages larger than 12V.)

Right.

That's a nice chip, capable of operation to 600V. But chips of
this class tend to have rather high supply current when operated
at high frequencies (even with no load), so other lower-voltage
parts are often better. Intersil's HIP4080 series, for example,
is rated for use up to 80V. These parts can be used at higher
frequencies and generally have lower supply current drain.

Thanks,
- Win
 
P

Paul Burridge

Jan 1, 1970
0
You can use one of the parts from the Microchip (formerly Telcom)
TC4426 to 4428 series. These are among the lowest-current parts
offered, and are available in soic packages. They can deliver up
to 1.5A to the gate, but it's customary to add a series resistor.
If the resistor is large enough, it'll reduce the peak gate-drive
current and thereby slow the FET's switching speed, if you like.

Sorry, Win, but I'm a bit confused here. Why would anyone want to slow
the FET's switching speed? I thought the idea was to keep it out of
the linear region as much as possible; not dawdle through it wasting
power and overheating the device.
Can you clarify, please?

Thanks,

p.
 
J

John Popelish

Jan 1, 1970
0
Lowly said:
I am "high-side" switching a P-FET at several hundred Khz. The source
is connected to +12V and I need to pull the gate to Vsource minus 10V
or so to fully turn it on. I'm using a ground-referenced, 5V swing PWM
circuit to perform the switching.

I need to really "slam" the gate of the FET to reduce the switching
time. In the past, I used a high-speed op-amp to provide the
necessary gate drive, but don't feel comfortable with this approach
because the output current required stresses the op-amp, and the 5V
swing probably doesn't make the input too happy either.

Does anyone know of a SMT FET driver that has a push-pull output,
100mA of drive and can tolerate a 16V supply? Any other ideas?

TIA

Have you looked at the mosfet driver chips from Microchip?
http://www.microchip.com/1000/pline/analog/anicateg/power/pwrmang/mosfet/index.htm
 
C

Chris Carlen

Jan 1, 1970
0
Winfield said:
Hey, do you _really_ want 10ns switching speed? :>)

Seriously, a resistor in series with the driver's output is common
both to act as an L-C resonance damper and to limit currents from
any beyond-the-rail transients caused by all the rapidly switched
high currents. But such a resistor need not slow down our FET's
switching speed. Normally during its drain-voltage transition our
FET driver's output FETs (the other FETs) are operating in saturated
mode, delivering their maximum current (1.5A for TC4426 series parts)
to our FET's gate, which would be in its active region near 4 to 6
volts. So if we add a resistor that drops less than say 3V at 1.5A,
or 2 ohms, we'll not reduce our FET's gate drive at all. However,
resistors larger than say 3 ohms will start to slow down our FET's
switching speed.

A 1.5A or higher gate current will switch most FETs very quickly,
even as fast as 5 to 10ns for some FETs. Such a fast switching
speed is useful for some applications, but in most cases slowing
the FET down a bit will not greatly increase its power dissipation.
But slowing the FET to 20 or even 50ns switching, instead of 10ns,
will reduce the chance of voltage overshoot-ringing troubles, and
will likely reduce the generated RFI.

The O.P. mentioned 100mA or lower gate drive, that's rather low for
a PWM-modulated power supply.


Keep in mind TC4426-8 devices can have up to 40ns rise/fall times into
1000pF. TC4421-2 are even stiffer at up to 9A peak.

Good day!
 
W

Winfield Hill

Jan 1, 1970
0
Chris Carlen wrote...
Keep in mind TC4426-8 devices can have up to 40ns rise/fall times
into 1000pF. TC4421-2 are even stiffer at up to 9A peak.

Chris, kindly keep in mind who you are talking to here - it's not
wise for a neophyte to lecture an expert! :>)

And avoid the mistake of assuming that the rise and fall time of
a driven power MOSFET is constrained to the speed of the driver,
for this is not at all the case. Consider, the driver's rating is
based on the time required to ramp its voltage between zero and say
12V. By contrast, a FET's entire drain-switching excursion occurs
in just a portion of that time, during which time its gate voltage
is in its linear region (usually someplace around 4 to 6V), barely
changing at all.

So in fact a 40ns gate driver can easily create a 5ns FET switching
speed. The switching speed is determined from the FET's data sheet
parameter typically called Qgd, the Gate-to-Drain ("Miller") Charge,
and is given by t = Qgd / Ig, where Ig is the gate-driving current.
The switching speed of the driver does not appear in that formula.

There are several things that do constrain FET switching speed, e.g.
by limiting our ability to rapidly assert a high current Ig. These
include V = L dI/dt voltage drops across the lead inductances, and
the FET's internal gate spreading resistance. With respect to the
latter issue, it's useful to keep in mind that a portion of the FET
may be immediately available to a rapidly-changed gate voltage, well
before the rest of its structure. For this reason, when working in
the ns time scale, a FET can be turned on somewhat faster than off.

Of course there are a number of other aspects involved, which if
properly understood and dealt with can allow one to take a power
MOSFET from the 5ns region down to the 1 to 2ns region! But for
you Chris, the issues I've mentioned should help give you a glimmer
and help to gradually raise you from neophyte status. Keep plugging
Chris, but in the meantime be more respectful of your seniors. :>)

Thanks,
- Win
 
F

Frank Bemelman

Jan 1, 1970
0
Chris Carlen said:
But it sure is fun!

Win, what I meant to say is: Keeping in mind that ... what is the
relationship between the risetime and the FET switching speed?

Practising on the witty stuff? Hahaha....

[snip]
 
Top