Transistor Recommendation Please

[Anthony Fremont]

I've been playing around with LTSpice and I'm ready to start soldering.
[I'm a hobbyist, not an EE student, so I can actually solder ;-)] The
circuitry is simple, but I need a logic level MOSFET (N channel) that
can take a Vds of at least 400V and 20-30A, though I won't be putting
that much current thru it for any amount of time (only around 5-6A avg).

The circuit is a fairly simple configuration with the MOSFET switching a
330uH inductor (wound it myself) to ground via a PWM signal. At the
junction of the inductor and the MOSFET I'll have a fast recovery diode
(the NTE substitute for MUR460) passing the high voltage peaks to a 400V
820uF cap. The plan here is to build up as much as 350V on the cap and
then dump that thru an electromagnet coil in order to provide some
impetus to a ball bearing. The initial circuit will have only one coil,
but I plan to add more stages later (if I survive) and control the whole
thing from a PIC. A simple voltage divider at the capacitor will
provide the feedback to the PIC so that it will know the cap is charged
to the wanted value. After looking at the LTSpice plots, I've decided
to add another fast diode between the inductor and the MOSFET to stop
those nasty negative currents that apparently flow thru the body diode
of the MOSFET after it turns off.

Fairchild makes a nice part (FDH44N50), but I'd like something more
locally obtainable like maybe an NTE part. Any recommendations on a
suitable SCR for dumping the 50J of current in about 10 milliseconds
would also be appreciated.

Thanks

 

[Winfield Hill]

Anthony Fremont wrote...

Fairchild makes a nice part (FDH44N50), but I'd like something more
locally obtainable like maybe an NTE part. Any recommendations on a
suitable SCR for dumping the 50J of current in about 10 milliseconds
would also be appreciated.

Hah, forget NTE, go with the big boys and save money. You can get
samples in many cases to experiment with. As for the FDH44N50, yes
that's a BIG part, with a BIG die, as evidenced by its massive 750W
power rating. This also means it has high capacitance (i.e. Ciss =
5335pF), which creates special problems, not to be taken lightly!
It requires a high 108nC of gate charge. Anyway, if you're still
interested, Mouser has them in stock, $16.62 each.

Now, the FDH44N50 is spec'd as a 44A 500V part. I'd guess that your
application does not require faster than 50 - 100ns switching speeds.
For voltages above 200V, you'd be better off with IGBTs. I've found
that the IGBT-manufacturer crowd prefers to make 600V rather than 500V
parts. You can get *really* big ones, like the IXYS IXGN200N60, a
200-amp part. Anyway, consider Fairchild's FGH50N6S2. Even though
it's rated higher at 50A, it only requires 85nC of gate charge, and
it's cheaper, $12.36 at Mouser. (The bad news is Mouser doesn't have
the "D" version in stock. When switching high currents it's useful
to have a stealth diode onboard.) I suggest you learn about IGBTs.

 

[[email protected]]

Is this a coil gun application as it sounds like one? If it is (or
someting similar) round metal objects make extremely poor projectiles,
basically due to minimum amount of flux lines thru their short length.
Much better to go with long skinny metal objects (dry wall screws, etc)
that maximise the flux).

Have you considered an SCR. The drive circuitry is a bit more involved
but they are more readily available in the voltage and current ratings
you require.

Craig

 

[Anthony Fremont]

Anthony Fremont wrote...

Hah, forget NTE, go with the big boys and save money. You can get
samples in many cases to experiment with. As for the FDH44N50, yes
that's a BIG part, with a BIG die, as evidenced by its massive 750W
power rating. This also means it has high capacitance (i.e. Ciss =
5335pF), which creates special problems, not to be taken lightly!
It requires a high 108nC of gate charge. Anyway, if you're still
interested, Mouser has them in stock, $16.62 each.
Ouch.

Now, the FDH44N50 is spec'd as a 44A 500V part. I'd guess that your
application does not require faster than 50 - 100ns switching speeds.
For voltages above 200V, you'd be better off with IGBTs. I've found
that the IGBT-manufacturer crowd prefers to make 600V rather than 500V
parts. You can get *really* big ones, like the IXYS IXGN200N60, a
200-amp part. Anyway, consider Fairchild's FGH50N6S2. Even though
it's rated higher at 50A, it only requires 85nC of gate charge, and
it's cheaper, $12.36 at Mouser. (The bad news is Mouser doesn't have
the "D" version in stock. When switching high currents it's useful
to have a stealth diode onboard.) I suggest you learn about IGBTs.

Thank you very much. I've been looking thru the datasheet for the
Fairchild FGH50N6S2, and it appears to be designed for this very type of
application. But, being a self-taught hobbyist, I'm pretty weak when it
comes to some (allot/all) of this stuff, so I have some stupid questions
for you.

1) The VGE(TH) has a max spec of 5V and a typical of 4.3V. Will 5V from
a PIC be enough drive to the gate or will I need to make arrangements
for a higher driving voltage?

2) As for a protection diode should I put one between the inductor and
the collector of the IGBT so that it conducts only when the IGBT is
turned on, or should it be turned around and placed across the collector
and emitter to allow for "negative" current flow into the inductor from
ground when the IGBT is switched off?

3) Will using a fast recovery diode (as described in question 2) be
enough, or should I be holding out for the "D" version of the
transistor.

Thanks again for all your help.

 

[Anthony Fremont]

Is this a coil gun application as it sounds like one? If it is (or

Pretty much. I actually have an application in mind for a form of
kinetic art where ball bearings will be required since they will need to
be rolling. That is if I don't kill myself first. ;-) I won't need
enough propulsion to kill anyone, just enough to kick the ball into the
air a few feet.

someting similar) round metal objects make extremely poor projectiles,
basically due to minimum amount of flux lines thru their short length.
Much better to go with long skinny metal objects (dry wall screws, etc)
that maximise the flux).

OK, that pretty much makes sense. Thanks for the tip.

Have you considered an SCR. The drive circuitry is a bit more involved
but they are more readily available in the voltage and current ratings
you require.

I guess the problem with that would be in getting it turned back off
after switching it on. My circuit works allot like a boosting SMPS in
that an inductor is rapidly switched to ground as the voltage across a
capacitor is built up. The only form of regulation though is by
feedback from a voltage divider to tell the PIC to stop the PWM signal.
I will need an SCR to dump the charge in the capacitor to the coil
though.

 

[Winfield Hill]

Anthony Fremont wrote...

Winfield Hill wrote ...

Ouch.

What, that's a good price for something with so much silicon.
The question is if it's overkill for your needs?

Thank you very much. I've been looking thru the datasheet for the
Fairchild FGH50N6S2, and it appears to be designed for this very type of
application. But, being a self-taught hobbyist, I'm pretty weak when it
comes to some (allot/all) of this stuff, so I have some stupid questions
for you.

1) The VGE(TH) has a max spec of 5V and a typical of 4.3V. Will
5V from a PIC be enough drive to the gate or will I need to make
arrangements for a higher driving voltage?

Whoa!! The threshold spec is where the current just starts to flow,
did you note the IC = 250µA parameter? You'll need a serious gate
voltage, did you see that VGE = 15V is used in most of the specs.

Furthermore, in general, you should rapidly switch the gate voltage,
so the IGBT is either on, or off, spending a minimum time in between.
Note the IGBT's 85nC gate charge, which comes from its nearly 3nF of
gate capacitance. It takes I = Q/t = 1.7A of gate current to switch
the gate in say 50ns. You can't get 15V from a PIC, nor 1.7A either,
so you'll need a gate driver, like Microchip's TC4423 dual 3A part.
http://ww1.microchip.com/downloads/en/DeviceDoc/21937a.pdf

2) As for a protection diode should I put one between the inductor
and the collector of the IGBT so that it conducts only when the IGBT
is turned on, or should it be turned around and placed across the
collector and emitter to allow for "negative" current flow into the
inductor from ground when the IGBT is switched off?

The latter is what you get with the "D" IGBT versions.

3) Will using a fast recovery diode (as described in question 2)
be enough, or should I be holding out for the "D" version of the
transistor.

The goal is to protect the IGBT from short transient spikes. You
certainly can use an external diode if you wish for convenience.

 

[Rich Grise]

Pretty much. I actually have an application in mind for a form of kinetic
art where ball bearings will be required since they will need to be
rolling. That is if I don't kill myself first. ;-) I won't need enough
propulsion to kill anyone, just enough to kick the ball into the air a few
feet.

Find a local pinball machine shop - some tech there will know about
slinging steel balls around with electromagnets.

Good Luck!
Rich

 

[[email protected]]

In that case, here are a few links (plenty more on the internet) I have
found that give some good theory and practical application covering
small to large devices, as well as experimental data, circuits,
components, etc.

http://www.oz.net/~coilgun/home.htm
http://www.powerlabs.org/emguns.htm
http://members.tripod.com/Lissandro/

May give you some ideas or point you in a slightly different direction
(and save you some time), or at the very least a good learning
experience.

Exercise extreme caution when working with your 820uF cap charged up to
350V (and the coil, and well the rest of the circuit itself if it isn't
isolated) as you can kill yourself with the circuit that you are using!
Given your limited electrical experience (I'm assuming mind you),
perhaps you should consider using a safer circuit with a larger
capacitance and lower voltage capacitor (or array of capacitors) to do
the same thing.

Also when you talk about wanting a more locally obtainable MOSFET, what
country are you in (as it is not clear)?

 

[Anthony Fremont]

Anthony Fremont wrote...

What, that's a good price for something with so much silicon.
The question is if it's overkill for your needs?

It's way overkill I do believe. ;-)

Whoa!! The threshold spec is where the current just starts to flow,
did you note the IC = 250µA parameter? You'll need a serious gate
voltage, did you see that VGE = 15V is used in most of the specs.

That's what I was afraid of.

Furthermore, in general, you should rapidly switch the gate voltage,
so the IGBT is either on, or off, spending a minimum time in between.
Note the IGBT's 85nC gate charge, which comes from its nearly 3nF of
gate capacitance. It takes I = Q/t = 1.7A of gate current to switch
the gate in say 50ns. You can't get 15V from a PIC, nor 1.7A either,
so you'll need a gate driver, like Microchip's TC4423 dual 3A part.
http://ww1.microchip.com/downloads/en/DeviceDoc/21937a.pdf

OK, thanks for the pointer to the gate driver. I'm considering going
this route since they're cheap. As you can clearly tell, I'm not much
of an analog designer especially where high power/voltage are involved,
but I am learning (slowly ;-). Micros and particulary software are much
more my forte.

After spending some time digging around Fairchild's site, I've come
accross a couple of "logic level" IGBTs. The ISL9V3040P3 (who makes
these crazy part numbers up) is rated for 17A continuous and 430V in a
TO-220 package. With a 5V gate drive, it only needs 17nC of charge.
But I guess that means that I still need a gate driver since the PIC can
only supply about 25mA, and I still need around 340mA or so. Is that
right? That 430V voltage rating should be fine and I'm only planning on
having <10A flowing when the transistor is on. FWIW, I'll be using a
50% duty cycle when I'm charging the caps. Oh yeah, only $4.20 single
qty at Mouser.

I came accross this article when poking around the net:
http://powerelectronics.com/mag/power_why_opt_igbts/
I can make some sense out of about half of it. I had already noticed
that IGBTs seem to like to state their on "resistance" in terms of
voltage drop as opposed to the way that MOSFETs are categorized. This
seems to really pay off at high currents. Heat doesn't seem to be much
of an issue with these parts too, very odd. I also noticed the author
making a blanket statement about 15V gate drive requirements for IGBTs.
So does that mean that these "logic level" IGBTs represent something
relatively new or ? I'm trying to learn.

The latter is what you get with the "D" IGBT versions.


The goal is to protect the IGBT from short transient spikes. You
certainly can use an external diode if you wish for convenience.

Ok thanks again, but which is the better way to do it? I'm thinking
that paralleling with the transistor probably takes a less hardy/speedy
diode. Is that correct?

Before you say it too, I know that I could "potentially" kill myself
with this so I will be very careful. I intend to start out with much
smaller voltages and work my way up until the smoke comes out of
something. Hopefully that won't be me. ;-)

Now here's a really stupid question. Will I have problems soldering
lead-free parts using "regular" solder?

 

[Anthony Fremont]

In that case, here are a few links (plenty more on the internet) I have
found that give some good theory and practical application covering
small to large devices, as well as experimental data, circuits,
components, etc.

http://www.oz.net/~coilgun/home.htm
http://www.powerlabs.org/emguns.htm
http://members.tripod.com/Lissandro/

Cool, I'll check those out.

May give you some ideas or point you in a slightly different direction
(and save you some time), or at the very least a good learning
experience.

Exercise extreme caution when working with your 820uF cap charged up to
350V (and the coil, and well the rest of the circuit itself if it isn't
isolated) as you can kill yourself with the circuit that you are

using!

I know, but we only live once anyway. ;-)

Given your limited electrical experience (I'm assuming mind you),

I've been tinkering for a long time, but not with high voltage AND big
caps. But I do know the danger.

perhaps you should consider using a safer circuit with a larger
capacitance and lower voltage capacitor (or array of capacitors) to do
the same thing.

It wouldn't be exactly the same thing, though it would be the same
amount of energy AIUI. However, my experiments are going to be with the
innards of a standard Ford fender mounted starter solenoid. They're
strong and cheap, but they're not real long so I'm thinking that some
voltage will be required to build the biggest magnet field in the
shortest amount of time.

Also when you talk about wanting a more locally obtainable MOSFET, what
country are you in (as it is not clear)?

Texas ;-)