37 amps through a 55 amp Mosfet, should I use the TAB instead of pin 2?

[gary350]

I am pulling 0 to 37 amps through a 55 amp mosfet, P55NF06L. I have very good aluminum heat sinks they are not getting hot with heat sink compound. The mosfet is ok with 37 amps for a few minutes but it will not do 38 amps more than about 3 seconds. Current goes through pin 1 and pin 2, but the tab is the same at pin 2 so maybe the tab can handle more current than the pin? I have not tried running current through the tab yet. I am a little bit surprised the pins can actually handle 37 amps there physical size is about = to #18 copper wire. Pins are solidered to #10 solid copper wire the copper wire can be heat sinking the pins too. This is an educational question to learn about mosfets. I think this circuit has reached its limit, time to do something different. I need to learn and have a better understand why these mosfets go up in smoke at 38 amps?

 

[Ian Rogers]

I don't know too much about mosfets, but I do know that if you don't open them fully, they get VERY hot... Once the junction hits 100 degrees the current drops off very quickly..

N fets are hard to switch on fully as the gate has to be 10v+ more than the drain... so in a 12v circuit you have to generate 22v+ to switch it on.... Only then can you use it's full potential...

I use 4 Nfets in a bridge so I use a fet driver as I have no idea how to do it manually..... Also gate capacitance has to be at a certain level.... I have used 100 ohm gate resistors and it didn't work.... Swapped them for 33 ohm and they did work... Its a magic I don't understand!!!

 

[jpanhalt]

Here's a good description of some actual tests on the TO-220 package. Is that what you are using? Your device also comes in a D2Pak. The package for the TO-220 is what limits the current, not the die itself.

http://www.irf.com/technical-info/designtp/dt93-4.pdf

As for your question, yes, for a very high current application, I soldered the tab. That not only decreases the lead resistance, but solder makes an excellent conductor for the heat sink. Just FYI, at 75A, the internal bonding wires reached 220°C and were dissipating 1.73W. At 37A, you should be OK according to that note.

John

 

[gary350]

I notice on some manufacture PC boards they cut off pin 2, the tab is attached to the metal heat sink that is soldered to the PC board as pin 2.

If pin 2 is not connected directly to the tab but a separate internal connection to die then there could be an advantage to soldering pin 2 to the tab for high current.

The link says, the TO-220 package is the current limiting factor.

 

[jpanhalt]

Just do like you see. Solder the tab and ignore pin 2. My drives included 5 mosfets in parallel and passed more than 300 A.

John

 

[gary350]

I was considering soldering several mosfets in parallel. Can all the 1 pins be soldered in parallel? Can all the 3 pins be soldered in parallel too? Is it that easy?

 

[jpanhalt]

Solder the 3 pins in parallel. Use a separate gate resistor for each 1 pins (gate) (think of parallel led's) Use something around 10 to 22 Ω for the gate resistors.

John

 

[jpanhalt]

Here are some images of a prototype board that followed the principles abpve. It was built more than ten years ago:

Showing individual gate resistors:


John

 

[ronsimpson]

Solder the tab and ignore pin 2.

If pin 2 is not connected directly to the tab but a separate internal connection to die

The tab and pin2 are the same thing. Rip off the black and you will see there the same thing.

 

[AnalogKid]

N fets are hard to switch on fully as the gate has to be 10v+ more than the drain... so in a 12v circuit you have to generate 22v+ to switch it on.... Only then can you use it's full potential...

I have used 100 ohm gate resistors and it didn't work.... Swapped them for 33 ohm and they did work... Its a magic I don't understand!!!

The statement about gate drive voltage is incorrect. For a typical n-channel power MOSFET, the gate must be at least 10 V above the *source* to fully enhance the device ("saturate", minimum Rdson, etc.). When the FET is fully on, the drain potential will be very near that of the source, but the enhancement function is defined by Vgs (voltage from gate to source), not Vgd (voltage from gate to drain). In a 12 V circuit with the source tied to GND (typical for a power switch application), the transistor starts conducting at a gate voltage of around 2-4 volts, and is as on as it can be by around 8 V; hence the 10 V spec.

As for gate resistors, it's not magic. A power MOSFET is actually hundreds of small MOSFETs in parallel. The O in MOSFET stands for oxide, an insulator, basically, a sheet of glass a few atoms thick, the dielectric of a small capacitor formed by the gate electrode and the channel. Pile up a few hundred, and the total capacitance can be 2 or 5 or even 10 nF, enough capacitance to slow down the turn-on time of the device if being charged up through too large a resistance. To get a large power MOSFET to turn on quickly enough not to dissipate a lot of power, you need amperes of gate drive current for a microsecond. The smaller the series resistance the better, although zero ohms can cause other circuit problems.

ak

 

[Ian Rogers]

The statement about gate drive voltage is incorrect. For a typical n-channel power MOSFET, the gate must be at least 10 V above the *source* to fully enhance the device

Cheers for the understanding.... I thought it was a bit high!! I must be confused with the bridge where the source of one is connected to the drain of the other...... I knew it was something like that.... Any information I can get about fets is useful as I like the idea of next to zero RDon... As for gate resistors, I have seen 10 ohm, 20 ohm and 220 ohm... This is the bit that confuses me... A bipolar is simple.... base current -> collector current!!

Gate resistors seem to be.... well... No idea!!!

 

[spec]

I am pulling 0 to 37 amps through a 55 amp MOSFET P55NF06L. I have very good aluminium heat sinks they are not getting hot with heat sink compound.

The temperature of the heatsink is not the issue here.

The MOSFET is OK with 37 amps for a few minutes but it will not do 38 amps more than about 3 seconds. Current goes through pin 1 and pin 2, but the tab is the same at pin 2 so maybe the tab can handle more current than the pin? I have not tried running current through the tab yet.

It is not surprising the P55NF06L is blowing: is it going short circuit between drain/source, if so the MOSFET has fused due to excessive temperatue. Yes, the tab can handle much more current.

I am a little bit surprised the pins can actually handle 37 amps there physical size is about = to #18 copper wire.

They can't with the TO-220 case.

Pins are soldered to #10 solid copper wire, the copper wire can be heat sinking the pins too.

To an extent.

This is an educational question to learn about MOSFETs. I think this circuit has reached its limit, time to do something different.

Good project; do not move on. MOSFETS are easy to work with, and have many valuable characteristics that make them good for a wide range of circuits.

I need to learn and have and better understand why these MOSFETs go up in smoke at 38 amps?

You will.

I am putting together a post which hopefully will explain all

 

[spec]

Hi gary,

I forgot to ask:
When you tested the P55NF06L NMOSFET what circuit did you use? Can you post a schematic?
If not, can you say what voltage was on the drain. I assume the source was at zero volts.
Can you also say what voltage was on the gate and explain the circuit for driving the gate?

spec

 

[AnalogKid]

Gate resistors seem to be.... well... No idea!!!

In words of one acronym - EMI. High speed high power MOSFET circuits have all kinds of...personality. The gate resistor is there for a couple of reasons, but the main one is to dampen oscillations formed by an extremely fast edge banging on a series resonant tank and making it ring. The tank is formed by the gate capacitance and a combination of the pc board trace inductance and the MOSFET package lead inductance. Not only is the ringing implicitly bad because it causes conducted and radiated EMI in the gate circuit, it can be big enough to modulate the Rds of the FET, turning the FET into an RF power amplifier. Not good. The same thing can happen with high speed data/address bus transceivers, which is why some octal and hex bus drivers are available with and without internal 25 ohm buildout resistors. One ringy-dingy...

ak

 

[Ian Rogers]

So on slow moving on off switches the gate resistor doesn't have much influence.??

Driving a bridge with a 20khz PWM requires a low gate resistor to keep a bit of stability??

 

[ChrisP58]

So on slow moving on off switches the gate resistor doesn't have much influence.??

Driving a bridge with a 20khz PWM requires a low gate resistor to keep a bit of stability??

The ringing issue has more to do with the rise and fall times of the gate drive pulse than it does with the switching frequency.

That being said, lower switching frequencies can usually get by with lower gate drive currents, so the gate pulse rise and fall times are often slower.

 

[spec]

So on slow moving on off switches the gate resistor doesn't have much influence.??

Driving a bridge with a 20khz PWM requires a low gate resistor to keep a bit of stability??

No mention about frequency, just current balance. And I was talkind about tolerencing not absolute value. Besides which, it turns out that gary's circuit is DC just like he had implied from the start of the associated thread and this thread. Also there are many ways of charging and and discharging the vitual and real MOSFET gate capacitance, even with a high gate resistor value.

A well-shaped and fast gate drive has nothing to do with stability. It ensures a fast output edge to minimise power dissipation during transitions.

 

[Ian Rogers]

I tried to build a fet bridge.... I had the same troubles as everyone... Fets just burned up..

I bought a ready made bridge driver and 4 x 12A NFETs all on a little board.... It doesn't even get hot when stalling the motors!!! Same design... Same fets.... It's the same as welding.... I cannot weld, as much as I want to, it just isn't going to happen.... I leave it to someone else....

 

[spec]

I tried to build a fet bridge.... I had the same troubles as everyone... Fets just burned up..

I bought a ready made bridge driver and 4 x 12A NFETs all on a little board.... It doesn't even get hot when stalling the motors!!! Same design... Same fets.... It's the same as welding.... I cannot weld, as much as I want to, it just isn't going to happen.... I leave it to someone else....

Hi Ian,

Yes, know what you mean: there is quite a bit of design and development work involved in getting high-powered circuits working OK. If you just wan't to use the equipment, rather than having the fun of doing your own design and build, it's much better to buy, especially as you can usually shop around and get a good price, or even buy second-hand.

When you say you can't weld as much as you would like too, can you elaborate? What kind of welding are you doing? Did you design and make the welder yourself?