Why are there multiple MOSFETs in parallel in power inverters H bridge driver?

[abhishekkumar1902]

Although I understand the fact that adding MOSFETs in parallel will increase the current handling capacity of that node/path in the circuit but in a lot of designs of FULL H bridge driver circuits where it is designed to drive the low side of a power transformer (thus requiring higher currents), I see FULL H bridge driving circuits uses multiple MOSFETs in parallel even though that particular MOSFET model is capable of driving the current in that node/path with just one unit only(as per the datasheet).

For example in a 12v, 1000W inverter design, the secondary driving current would be around 84Amps. But they used three IRF3205 in parallel in each node of the full h bridge provided as per the datasheet, IRF3205 is capable of sourcing continuous drain current of 110 Amps (at 25'C rating) and 80 Amps (at 100'C rating).

Is it just to bring the thermal dissipation down and improve stability or some other reasons as well?

[Seekonk]

Just calculate the wattage of heat generated at that 80A.  First with one FET and then with three in parallel.  Not even looking at the data sheet I'd say just using three is a bad design.

[Zero999]

The IRF3205 isn't even rated to 84A. It can handle 75A tops. Refer to footnote  #5 on page 2 of the data sheet.
http://www.irf.com/product-info/datasheets/data/irf3205.pdf

Operating it as 75A would give a power dissipation of 45W and the temperature rise would be nearly 34^oC, even on a perfect heat sink.

[abhishekkumar1902]

Just calculate the wattage of heat generated at that 80A.  First with one FET and then with three in parallel.  Not even looking at the data sheet I'd say just using three is a bad design.

So you mean they should use even more mosfets in parallel? 3 is not sufficient?

[T3sl4co1l]

1. It isn't practical to sink more than about 50A through a TO-220 device, no matter what the datasheet might lie to you about it (tip: read the specs carefully and skeptically).

2. It isn't practical to design a switching circuit that switches more than 50A, through a single device, at low voltages.

Switching is more practical, the closer the switching impedance (Zsw = Vpp / Ipk) is to the impedance of the layout (which is proportional to the impedance of free space, Zo ~= 377 ohms, and usually a few times lower because of layout).  Typical PCB layout impedances, for power switching circuits, are in the 10-50 ohm range.

This makes it most optimal to design, say, an inverter at 50A and 500V, and indeed, industrial inverters have quite good performance in that area.

At low voltages (<50V and >50A), the switching impedance is under 1 ohm, and layout cannot be small enough.  (One TO-220 device has about 10nH of stray inductance.
 The inductor equation is: V = L * dI/dt  Therefore, under those conditions, it cannot switch faster than (50V) = (10nH) * (50A) / dt ==> dt = 10ns.  And most likely, in a practical circuit, the real figure will be 5 or 10 times longer than that, or ~100ns.  This results in higher switching losses, and the switching frequency must be reduced to prevent meltdown.

By using devices in parallel, the inductance is reduced through many parallel branches, and snubbers can be applied per transistor, to control what's left.

Tim

[Cerebus]

1. It isn't practical to sink more than about 50A through a TO-220 device, no matter what the datasheet might lie to you about it (tip: read the specs carefully and skeptically).

Corollary: The smaller the SOA graph is printed in the datasheet, the more likely it is that it tells you something you don't want to hear.

[Zero999]

Although I understand the fact that adding MOSFETs in parallel will increase the current handling capacity of that node/path in the circuit but in a lot of designs of FULL H bridge driver circuits where it is designed to drive the low side of a power transformer (thus requiring higher currents)

When you say power transformer, are you talking about a large mains frequency unit?

If so, I'm surprised they still make them like that. Normally a DC-DC converter is used to create the peak mains voltage 170V or 340V, depending on whether it's 120V or 240V and an h-bridge on the output converts it to AC mains frequency. The cost of the additional electronics is outweighed by a smaller, more efficient transformer.

Decent inverters will use PWM and filter the output, to make a good sine wave. Cheap inverters will have no filter and use a modified sine wave.

[Seekonk]

Well, "they" don't but there are plenty people building them worth LF transformers. Can't believe what a hunk of iron costs these days.

http://www.fieldlines.com/index.php/topic,148827.0.html

[Muxr]

I would also imagine it improves the overall Rds(on), and the efficiency. Operating the MOSFET at lower temps/sharing the load also improves this aspect.