MOSFET Driver cannot drive a MOSFET? or something wrong?

[[email protected]]

Here is my circuits, simple, yet, when I disconnect the MOSFET and
load from the circuits, the waveform, duty cycle, amplitude,
everything else is correct. After I connect the mosfet and load (here
I didn't use a laser diode, in stead I use a 1 ohm power resistor),
all 're in a messy! MOSFET is hot coz it's not fully on, square wave
form contains a lot of ringing, overshoot, duty cycle? no duty cycle
but some weired square waves with spikes.

It is a very common circuits but I wonder something wrong?


http://www.dianyuan.com/bbs/u/65/2567951210659430.jpg

 

[Bob]

Here is my circuits, simple, yet, when I disconnect the MOSFET and
load from the circuits, the waveform, duty cycle, amplitude,
everything else is correct. After I connect the mosfet and load (here
I didn't use a laser diode, in stead I use a 1 ohm power resistor),
all 're in a messy! MOSFET is hot coz it's not fully on, square wave
form contains a lot of ringing, overshoot, duty cycle? no duty cycle
but some weired square waves with spikes.

It is a very common circuits but I wonder something wrong?

http://www.dianyuan.com/bbs/u/65/2567951210659430.jpg

My guess is circuit layout or your 1ohm resistor is a wirewound
type with significant inductance.

Acording to the IXDD414 datasheet it has a risetime of under
30nanoseconds and can source or sink 14Amps.
With those numbers it dosn't take much inductance or
reflection to spoil a sharp edge.

If you have wired this up with six inch long hookup wires then
you are going to see ringing.

Bob

 

[James Arthur]

Here is my circuits, simple, yet, when I disconnect the MOSFET and
load from the circuits, the waveform, duty cycle, amplitude,
everything else is correct. After I connect the mosfet and load (here
I didn't use a laser diode, in stead I use a 1 ohm power resistor),
all 're in a messy! MOSFET is hot coz it's not fully on, square wave
form contains a lot of ringing, overshoot, duty cycle? no duty cycle
but some weired square waves with spikes.

It is a very common circuits but I wonder something wrong?

http://www.dianyuan.com/bbs/u/65/2567951210659430.jpg

Where's your gate resistor? You need one.

(Methinks your FET screams.)

Cheers,
James Arthur

 

[[email protected]]

1-2 ohm but didn't work.

Where's your gate resistor? You need one.

(Methinks your FET screams.)

Cheers,
James Arthur

 

[Mark]

1-2 ohm but didn't work.






- Show quoted text -

what rate are you switching at?

what is the gate capacitance of your FET?

can the driver drive that capacitance at that speed?

Mark

 

[Paul E. Schoen]

[bottom posted for clarity]

1-2 ohm but didn't work.

A gate resistor is not always needed, but it seems more likely there is not
enough bypass capacitance on the driver circuit. When the driver tries to
turn on, it needs to get its current from a voltage source that can handle
a high speed pulse. The driver should be bypassed with a low ESR capacitor
at least 10x the MOSFET gate capacitance. Also, the gate drive and return
tracks should be very short.

The source resistor shown in the schematic should be of low inductance, and
careful consideration must be given to the ground path for the high current
output. It is not even clear why the source resistor is present, unless it
is used for desaturation detection or current mode PWM.

Please provide the entire circuit, including details on what sort of power
supply is being used, and a PCB layout or sketch of wiring if it is on a
protoboard.

There is good information on high speed circuit design in a Linear
Technology app note AN47. It is more about amplifiers, but applies to this
as well. And AN25, "Switching regulators for poets". Also look at
International Rectifier app note AN-944 on gate charge and drivers for
MOSFETs.

Paul

 

[[email protected]]

now it's an open loop circuits. the 0.015 is for further close loop
circuits use. even I remove 0.015 ohm res the problem is still there.
I did it on bread board. I think it is caused by the inductance of
load, power supply wires. I don't know how to solve it. I added some
big capacitors between vcc and ground, I use separate power supply
for that mosfet, it didn't improve too much.

[bottom posted for clarity]

1-2 ohm but didn't work.

A gate resistor is not always needed, but it seems more likely there is not
enough bypass capacitance on the driver circuit. When the driver tries to
turn on, it needs to get its current from a voltage source that can handle
a high speed pulse. The driver should be bypassed with a low ESR capacitor
at least 10x the MOSFET gate capacitance. Also, the gate drive and return
tracks should be very short.

The source resistor shown in the schematic should be of low inductance, and
careful consideration must be given to the ground path for the high current
output. It is not even clear why the source resistor is present, unless it
is used for desaturation detection or current mode PWM.

Please provide the entire circuit, including details on what sort of power
supply is being used, and a PCB layout or sketch of wiring if it is on a
protoboard.

There is good information on high speed circuit design in a Linear
Technology app note AN47. It is more about amplifiers, but applies to this
as well. And AN25, "Switching regulators for poets". Also look at
International Rectifier app note AN-944 on gate charge and drivers for
MOSFETs.

Paul

 

[Paul E. Schoen]

You will get more help if you don't top post like this.

Paul

now it's an open loop circuits. the 0.015 is for further close loop
circuits use. even I remove 0.015 ohm res the problem is still there.
I did it on bread board. I think it is caused by the inductance of
load, power supply wires. I don't know how to solve it. I added some
big capacitors between vcc and ground, I use separate power supply
for that mosfet, it didn't improve too much.

On May 12, 11:31 pm, [email protected] wrote:

Here is my circuits, simple, yet, when I disconnect the MOSFET and
load from the circuits, the waveform, duty cycle, amplitude,
everything else is correct. After I connect the mosfet and load
(here
I didn't use a laser diode, in stead I use a 1 ohm power resistor),
all 're in a messy! MOSFET is hot coz it's not fully on, square
wave
form contains a lot of ringing, overshoot, duty cycle? no duty
cycle
but some weired square waves with spikes.

It is a very common circuits but I wonder something wrong?

Where's your gate resistor? You need one.

(Methinks your FET screams.)

[bottom posted for clarity]

1-2 ohm but didn't work.

A gate resistor is not always needed, but it seems more likely there is
not
enough bypass capacitance on the driver circuit. When the driver tries
to
turn on, it needs to get its current from a voltage source that can
handle
a high speed pulse. The driver should be bypassed with a low ESR
capacitor
at least 10x the MOSFET gate capacitance. Also, the gate drive and
return
tracks should be very short.

The source resistor shown in the schematic should be of low inductance,
and
careful consideration must be given to the ground path for the high
current
output. It is not even clear why the source resistor is present, unless
it
is used for desaturation detection or current mode PWM.

Please provide the entire circuit, including details on what sort of
power
supply is being used, and a PCB layout or sketch of wiring if it is on a
protoboard.

There is good information on high speed circuit design in a Linear
Technology app note AN47. It is more about amplifiers, but applies to
this
as well. And AN25, "Switching regulators for poets". Also look at
International Rectifier app note AN-944 on gate charge and drivers for
MOSFETs.

Paul

 

[James Arthur]

1-2 ohm but didn't work.

1-2 ohms? What are you trying to do? Pulse a laser diode, it seems.
How fast, how much current, etc.?

That's a massive FET. http://www.irf.com/product-info/datasheets/data/irl3803.pdf
(IRL3803: 140A, 0.006 ohms, Qg=140nC) Something smaller would be
easier. Are you sure you need such a monster?

To turn that beast on super quickly you'll need super layout, tight
decoupling, intelligent current paths, short traces, and so forth.

By temporarily using a 10 ohm gate resistor, much will be revealed.

Cheers,
James Arthur

 

[James Arthur]

I did it on bread board.

(I just saw this post, after posting my other reply...)

You can't do that. It won't work. You need a groundplane. And a
smaller FET with a larger gate resistor wouldn't hurt.

I think it is caused by the inductance of
load, power supply wires.
Yes.

I don't know how to solve it.

Build the thing on a solid piece of copper clad. Dead-bug style is
good. These problems will vanish.

Cheers,
James Arthur

 

[Paul E. Schoen]

1-2 ohms? What are you trying to do? Pulse a laser diode, it seems.
How fast, how much current, etc.?

That's a massive FET.
http://www.irf.com/product-info/datasheets/data/irl3803.pdf
(IRL3803: 140A, 0.006 ohms, Qg=140nC) Something smaller would be
easier. Are you sure you need such a monster?

To turn that beast on super quickly you'll need super layout, tight
decoupling, intelligent current paths, short traces, and so forth.

By temporarily using a 10 ohm gate resistor, much will be revealed.

Just a 0.1 uF or so bypass capacitor directly across the driver power
terminals should help a lot (like you said, decoupling). Also the
connection to the gate and back to the ground pin should be reasonably
short. I have a circuit that uses a Fairchild HUF75645 which is almost as
beastly as the IRL3803, and I have driven it up to at least 300 kHz with an
inductive load without problems, using a TI 9 amp driver. UCC27321. And my
first prototype on a perfboard was OK at 100 kHz.

The IRL2803 has 5000 pF gate capacitance, but it has 230 nSec rise time, so
a super fast gate driver is not really necessary. A 9 amp driver can
provide about 1.8 V/nSec, and realistically about 20 nSec to full on, but
the device itself will take ten times that to conduct fully.

There is an app note on the UCC35705 that is used to provide the PWM
signal: http://focus.ti.com/lit/ug/sluu141/sluu141.pdf

This is a 4 MHz part, which may be overkill for the application (or at
least the MOSFET). More information is really needed, such as supply
voltage and drive frequency.

Paul

 

[James Arthur]

[snip]

1-2 ohms? What are you trying to do? Pulse a laser diode, it seems.
How fast, how much current, etc.?

That's a massive FET.
http://www.irf.com/product-info/datasheets/data/irl3803.pdf
(IRL3803: 140A, 0.006 ohms, Qg=140nC) Something smaller would be
easier. Are you sure you need such a monster?

To turn that beast on super quickly you'll need super layout, tight
decoupling, intelligent current paths, short traces, and so forth.

By temporarily using a 10 ohm gate resistor, much will be revealed.

Just a 0.1 uF or so bypass capacitor directly across the driver power
terminals should help a lot (like you said, decoupling). Also the
connection to the gate and back to the ground pin should be reasonably
short. I have a circuit that uses a Fairchild HUF75645 which is almost as
beastly as the IRL3803, and I have driven it up to at least 300 kHz with an
inductive load without problems, using a TI 9 amp driver. UCC27321. And my
first prototype on a perfboard was OK at 100 kHz.

The IRL2803 has 5000 pF gate capacitance, but it has 230 nSec rise time, so
a super fast gate driver is not really necessary. A 9 amp driver can
provide about 1.8 V/nSec, and realistically about 20 nSec to full on, but
the device itself will take ten times that to conduct fully.

That's with just 4.5 volts of gate drive (Fig 10a. in the datasheet)
through a 1.3 ohm gate resistor, but 12v and a hot driver could
do the job much faster.

There is an app note on the UCC35705 that is used to provide the PWM
signal:http://focus.ti.com/lit/ug/sluu141/sluu141.pdf

This is a 4 MHz part, which may be overkill for the application (or at
least the MOSFET). More information is really needed, such as supply
voltage and drive frequency.

Paul

You can do a lot on a smart perfboard. The OP's "breadboard" though
clearly isn't cutting it, so recommending a groundplane makes up for
a lot.

It all comes down to what the OP is doing, which he hasn't told us.

If he's trying to get 15nS pulses from a laser diode with this
setup...

Grins,
James Arthur