Why MOSFETs in parallel in this circuit?

Thread Starter

Xavier Pacheco Paulino

Joined Oct 21, 2015
728
I attach a circuit which is a motor controller that I saw while searching on the web. I know that it controls high power DC motors, but why do they use 3 MOSFETs in parallel? It's like increasing the power? Can they be replaced with one MOSFET with higher ratings?
 

Attachments

danadak

Joined Mar 10, 2018
4,057
They can share but unevenly as they have no ohmic ballast Rs
in the source legs. You would have to do a worst case design to
see what greatest imbalance would be to make sure what the
worst case would be for the specific MOSFET that takes more
current than the others. If you want a reliable design.

Would definitely be better if one MOSFET with the appropriate ratings.
But gate drive circuit may need tweaking because of C gate load differ-
ences.

Regards, Dana.
 

ebp

Joined Feb 8, 2018
2,332
It is hard to say why they would opt for three FETs. It may be that the the same FET is used in some other product. It may be that the design is simply a few years old and that was one of the FET used was a good choice at the time. As Bob said, sometimes heatsinking is easier with more devices.

Provided they are switched rapidly and layout is reasonable, there is no need to ballast power MOSFETs when they are paralleled. It is necessary, as in the circuit, that each have its own gate resistor, otherwise it is possible to create very high frequency oscillation during the switching transitions, which greatly increases power loss and radio frequency interference.

The job could very likely be done with just a single FET without big problems with heatsinking. There are some amazing FETs on the market, and more coming along all the time. I would probably select something with a voltage rating of about 150 V. I haven't looked for anything in that voltage range for years, so I don't know what is available. If the battery were limited to 36 sealed lead-acid cells, a FET rated at 100 volts should be adequate. but it might be necessary to consider what would happen with high-speed coasting with the motor behaving as a generator.
 

MisterBill2

Joined Jan 23, 2018
12,692
To assure load sharing the resistor would be in the source lead so that as the current increased the Vgs would decrease, thus reducing the drive a bit. This negative feedback is a powerful tool for forcing current sharing and it works well for bipolar junction transistors.
The primary benefit of multiple device use is spreading the heat over more devices and area. In addition, lower current rated fets are cheaper.
 

wayneh

Joined Sep 9, 2010
17,171
In addition, lower current rated fets are cheaper.
That's true of course but I'm wondering if there's a sweet spot. In other words, suppose you need 500A of current capacity and a moderately high switching speed, say 100kHz to 1MHz (so that gate capacitance matters in addition to heat). What's the ideal MOSFET for that? Do you want five 100A MOSFETs or twenty 25A MOSFETs? Idle curiosity - I'm not building a 500A SMPS. :eek:
 

danadak

Joined Mar 10, 2018
4,057
Of course component count, interconnect and reliability inversely related. That's one
consideration.Minimizing mechanical volume another.

Regards, Dana.
 
Last edited:

shortbus

Joined Sep 30, 2009
9,525
That's true of course but I'm wondering if there's a sweet spot. In other words, suppose you need 500A of current capacity and a moderately high switching speed, say 100kHz to 1MHz (so that gate capacitance matters in addition to heat). What's the ideal MOSFET for that? Do you want five 100A MOSFETs or twenty 25A MOSFETs? Idle curiosity - I'm not building a 500A SMPS. :eek:
This is one of the things that marketing must have had a big hand in when data sheets were being written for mosfets. Promising big amperage use in a package that will not support it. In the mosfets that are mostly used by hobbyists (TO-220) they are still limited to ~75 watts. At least with normal heat sinks.
 

qrb14143

Joined Mar 6, 2017
112
This is one of the things that marketing must have had a big hand in when data sheets were being written for mosfets. Promising big amperage use in a package that will not support it. In the mosfets that are mostly used by hobbyists (TO-220) they are still limited to ~75 watts. At least with normal heat sinks.
One thing that our power electronics professor drummed into us during my degree.
"Read the voltage rating, read the thermal data, read the gate charge requirements. Just don't read the current handling capability because it's always garbage!" ;)
 

MisterBill2

Joined Jan 23, 2018
12,692
There is one brand of amplifier sold for HAM radio applications that makes a great deal out of the fact that their product uses eight transistors while the others products use just one transistor. The claim is made for much improved heat distribution as well as a cheaper repair price. Certainly distributing a given amount of power dissipation over many more junctions will spread the heat out over a wider area.So probably the whole thing is about trade-offs. Good engineering often includes carefully thought out trade offs. The decision thus must consider cost, size, and complexity, and in many cases also component availability. Single-sourced components can lead to a production halting disaster!!
 

ronsoy2

Joined Sep 25, 2013
71
More is more expensive in the production line but more is more reliable overall. Like someone suggested above the current ratings of the new super FETs are usually garbage, with a lot of impractical conditions applied.
Always consider the safe operating area curves of the FET when considering the power dissipation. Carefully read the conditions of the safe operating area. Some specify for a SINGLE PULSE only! Talk about garbage!
Some specify with a case temperature of 25C. Any FET dissipating high power will go way above 25C unless the heatsink is a block of ice!
It is much easier to cool a batch of FETs to a lower overall case temperature than a single super FET. Especially if the FET must be insulated electrically from the heat sink.
Source spreading resistors are necessary for power FETs just like bipolars. It is a misconception that the FETs can be carelessly paralleled with no spreading resistors. No resistors will work in low powered circuits, but become necessary as the heat load on the FET increases. This is especially important where the average current varies quickly over a wide range over time. (motor speed controls)
 

cmartinez

Joined Jan 17, 2007
7,764
More is more expensive in the production line but more is more reliable overall. Like someone suggested above the current ratings of the new super FETs are usually garbage, with a lot of impractical conditions applied.
Always consider the safe operating area curves of the FET when considering the power dissipation. Carefully read the conditions of the safe operating area. Some specify for a SINGLE PULSE only! Talk about garbage!
Some specify with a case temperature of 25C. Any FET dissipating high power will go way above 25C unless the heatsink is a block of ice!
It is much easier to cool a batch of FETs to a lower overall case temperature than a single super FET. Especially if the FET must be insulated electrically from the heat sink.
Source spreading resistors are necessary for power FETs just like bipolars. It is a misconception that the FETs can be carelessly paralleled with no spreading resistors. No resistors will work in low powered circuits, but become necessary as the heat load on the FET increases. This is especially important where the average current varies quickly over a wide range over time. (motor speed controls)
That was a lot of extremely valuable information in a few paragraphs ... thanks for posting!
 

Thread Starter

Xavier Pacheco Paulino

Joined Oct 21, 2015
728
I have read some papers, but still haven't decided if I should use parallel mosfets or not for my application. For example, let's say that I chose to use this MOSFET: SIHG80N60E-GE3, Datasheet: https://www.mouser.com/datasheet/2/427/sihg80n60e-1019450.pdf. I find that paralleling two of those MOSFETs is kind of expensive if I want to make something cost-effective. I attach an App Note that I found useful.

I have seen some designs that uses two mosfets in parallel @Id=50A. But there are plenty of MOSFETs that can stand 50A and higher so one mosfet will be enough I think.

I will control a brushed DC motor for treadmill. The motor is 3.5HP. PWM frequency: 25 kHz.
 

Attachments

MrAl

Joined Jun 17, 2014
9,337
Hello there,

Parallel mosfets was the only way to go a few decades ago when mosfets were starting to make their big appearance into the marketplace. Higher rated devices simply were not around yet.

The choice of part or parts though is sometimes not based on the most current technology, but sometimes what the guy happened to have laying around at the time. For an example a company i worked for back in the 80's wanted us to design circuits that used parts that were in stock already unless there was some very special case. That forced us to use certain parts unless there was some extreme circumstance.

If you do have the choice though then you should evaluate the efficiency using different combinations and pick the combination that works best. Parallel mosfets do have problems though unless done very carefully with attention to layout.
 

Thread Starter

Xavier Pacheco Paulino

Joined Oct 21, 2015
728
Hello there,

Parallel mosfets was the only way to go a few decades ago when mosfets were starting to make their big appearance into the marketplace. Higher rated devices simply were not around yet.

The choice of part or parts though is sometimes not based on the most current technology, but sometimes what the guy happened to have laying around at the time. For an example a company i worked for back in the 80's wanted us to design circuits that used parts that were in stock already unless there was some very special case. That forced us to use certain parts unless there was some extreme circumstance.

If you do have the choice though then you should evaluate the efficiency using different combinations and pick the combination that works best. Parallel mosfets do have problems though unless done very carefully with attention to layout.
I'm trying to avoid paralleling MOSFETs. I have just seen some applications which use them so that's why I started wondering myself if I should consider using mosfets in parallel. But as I said, some new MOSFETs nowadays claim to have better characterisitics so I guess that just one should work for my application, at least.
 
Top