[SOLVED] Need Help With Mosfet Packages

[meltbox360]

I am trying to design a motor controller (H-Bridge) capable of 100 amps continuous. I have switched mosfets multiple times since I am unsure of what type of mosfet would be best for this sort of job. I read DirectFET mosfets are good for use with heatsinks however they don't seem especially price effective. The mosfet I want to use is a sot404 d2pak mosfet model BUK664R6-40C,118 due to its combination of logic level gate and ability to carry a fair bit of current. The thing that is keeping me back is that I read d2pak conducts heat to the pcb. This would mean a heatsink on the package wouldn't cool it well?

If anyone has a better suggestion along with a driver that can drive a few mosfets in parallel at a time I would also welcome that but the main question is what are the advantages of surface mount high power vs through hole as well as the advantages of the various packages. Thanks :grin:

EDIT: Current plan is to use the lm27222 as a driver and a 7805 to power it limiting me to about 500ma-1a drive current.

 

[Tahmid]

Are you limited to using D2PAK package MOSFETs? Is it not possible to use Power MOSFETs with low-side MOSFET drivers in packages such as the common TO-220? This would give you a greater flexibility in choosing parts and using heatsinks as you can use external heatsinks and size them as required.

 

[meltbox360]

I am not limited. I simply though surface mount was the best solution except for the fact that they seem to conduct heat to the pcb which is not ideal for me. So from what I understand the best option is to go with a to 220 and mount a heatsink on the back drain?

Edit: I looked at some to-220 Mosfets and they all have higher junction to case thermal resistance. They also come packaged in d2pak which have lower thermal resistance to case. Doesn't this mean a heatsink would be LESS effective on a to-220 package?

 

[dustmote]

The only real advantage that surface mount parts have is that they are more easily and cheaply used in mass manufacturing operations due to their compatibility with pick-and-place machines and wave soldering processes..... They also may have a space savings over traditional through hole parts, but ALL of the higher power capable transistors must be equipped with heat sinks, and ALL of the higher powered transistors of any type are through hole {or larger} parts.

As was recommended above, please consider the use of through hole parts that can be easily interfaced to a sizable heatsink if you are going to need to be able to dissipate substantial amounts of power. From what I understand, the "logic level drive" MOSFETs tend to have a higher on state resistance, and therefor dissipate more power in use than another equivalent traditional MOSFET...... Also, traditional MOSFETs of similar current carrying and voltage ratings are cheaper than the equivalent "logic level" substitutes.

Another thing to be mindful of here is that when you mount these transistors to a metal heatsink, they will often need to be electrically isolated from the heatsink. This entails the use of either purpose made mica insulators, or other varieties such as silicone rubber sheets, ceramic insulators etc etc etc.... You can easily find these by searching for the package e.g. "TO-220 Heatsink Mica"..... You will also probably need to use non-conductive IE nylon shoulder washers for the mounting screws so that the screw which is used to attach the transistor to the heatsink does not act as a conductive path.

I am also a beginner in working with transistors, and this is some of the helpful information that I have learned largely from this website over the last few months..... This is my first post here, And I would like to thank you Tahmid, for all the help that you have provided on this website. It has been very educational and enlightening to me over the last few weeks.....

Many thanks my Friend!

- - - Updated - - -

Also please consider if the use of "logic level" MOSFETS is really necessary in this design.... Although it may seem easier to interface to a microcontroller etc, the logic level MOSFETs often present disadvantages of efficiency, cost, and robustness....

If you have a higher voltage supply available, perhaps you could incorporate a 7812 regulator or similar so as to gain the ability to use traditional MOSFETs.

 

[meltbox360]

Ahh I see. Thanks for the replies. I wanted to go logic level because of my concern with turn on times. I am interfacing it to an atmega168/328 28 pin package. From what I understand The turn on time should be quite a bit faster than the pwm signal which would be best kept out of the audible range. I figured that logic level would require the least burst current to drive and so I went with them as I would like to drive an entire side with one driver. My power source will be 12v for the motor and another smaller 12v for the driver and logic with a common ground.

Here is my current understanding. Please correct me if anything I write is incorrect.

1. Surface mount and chasis mount Mosfets are best for high power motor pwm applications.

2. A PCB can carry 100 amps with a reasonable trace width and 30C temperature rise. (As per trace width calculator on google)
Of course assume I solder copper wire along the traces to double the capacity or more. I think it should work fine. No? 1oz board assumed. Perhaps a board heatsink will be in order.

3. Switching current can be roughly calculated by (switching time)*(switching current)=(gate charge)

End of assumptions.

Isn't the isolation of the heatsinks optional and simply a precautionary measure? I will probably still do it though, thank you for reminding me.

Edit: Wouldn't electrically isolating the heatsinks also decrease thermal performance rather significantly?

Edit2: Currently looking at the aot240l which looks like it can carry 40 amps when heatsinked and possibly up to 45. I am looking into non logic level now.

 

[dustmote]

Whether heatsinks must be isolated depends on the situation..... IE What else is connected to the heatsinks, are they connected to/integral to the case?? It is usually against all kinds of UL and CE safety ratings to have a large heatsink electrically live like that LOL.... Of course it all depends on the voltages involved, and what the particulars of the situation are... Use common sense to determine.

You asked if electrically isolating the heatsinks would decrease the thermal performance... YES IT WILL... It has a large impact on thermal performance and care should be given in calculating heat rise of the transistor to NOT FORGET the increased thermal resistance added by the isolator..... You can find pretty decent estimates of the thermal resistance introduced by the various kinds of isolators by searching Google....
-------------------------------------------------------------------------------
As to your other questions:

1.Surface mount MOSFETs are certainly NOT the best for high power applications..... Specifically, the more effectively you can cool a transistor, the more current it can handle.... If you had some magic way to infinitely cool it's semiconductor junctions, then you could push an infinite current through it. This is the idea with superconductors.

Through hole components are almost always MUCH more effectively cooled via the proper use of heatsinks than any surface mount component can ever be through it's circuit board.....

Calculate how much current needs to be handled under the worst case scenario.. IE Highest temperature, motor fully loaded or even seized, THEN determine whether it makes sense from an economic and design perspective to use surface mount components or not. In certain cases where the final size of the device is critical, metal core circuit boards are made to help cool the SMT parts..... Most of the time for a home project, you can give yourself a lot more "elbow room" and ease in design concerning the thermal parameters of the board, in addition to a wider selection of high powered parts by using traditional components.

2. Short answer: This is totally unreasonable. Long answer: While a PCB trace can theoretically carry any amount of current when properly sized, a 1 OZ PCB trace would have to be at LEAST a couple of inches wide in order to carry 100 amps of current.... Also, how would you attempt to make a connection to this trace???? This is problematic. You can easily calculate the {DC} current carrying capacity of a PCB trace of a given width and thickness by considering the conductivity of the copper of which it is made. 1 OZ PCB uses a copper conductor which is about 35 micrometers thick.

Keep your high current paths as short as possible to try to eliminate as much heat production as possible, and consider that if this board is inside a case, that it is not a simple matter of the temperature rise in the copper being defined as the temperature rise of an uninsulated conductor in free space. All the heat builds up in the case, which causes higher resistance, which causes MORE heat to build up..... This same idea applies to ALL of your other components such as capacitors and transistors as well.

Also, at the point that if you have soldered a big wire all along the trace, why not just use an insulated wire to begin with?????

3. Switch current is the total current flowing through the MOSFET..... This is of course dominated by the current from drain to source, usually abbreviated I{subscriptDsubscriptS} Most of the time, when considering the switch current, the gate to source current is proportionally so small that it is disregarded for "quick and dirty general use" calculations...
In most cases, you figure the switch current as the duty cycle ratio multiplied by the current from drain to source while fully on.

......... I think that what you are trying to get at here is the gate current, as that is where your gate charge becomes important, and is the figure you need to know when deciding on a proper driving method to use a MOSFET. The gate current is usually largely affected by the use of a resistor in series in between your driver {In this case, the pin from your microcontroller} and the gate of the transistor. This series gate resistor will prevent your microcontroller pin from attempting to source too much current at once and burning out. The higher the current it can safely source, the lower a gate resistor you can use, and the faster the MOSFET will switch, leading to less "switching losses" during the turn on of the MOSFET.

Remember, the magic of MOSFETs is that they have very small resistance from drain to source when they are "fully on" and saturated. While they are in the linear region of their operation --IE where the voltage drop from drain to source is linearly inversely proportional to the gate to source voltage-- they have high losses. You want to turn these things "on and off" as quickly as possible to avoid dissipating large amounts of power and heat inside the transistor.

This is largely why special mosfet driver circuits and ICs are used in these applications rather than just driving the gate directly from a microcontroller. Higher current being used to drive the MOSFET = shorter time to overcome gate capacitance = faster MOSFET turn on = higher efficiency overall.

Also, do not forget the use and purpose of a pull-down connection at the gate of the MOSFET as well in order to help it turn off quickly and "completely" once the drive voltage is removed.... Forgetting this can cause the MOSFET to slowly turn off, or otherwise to oscillate and cause all kinds of problems.
------------------------------------------------------

Go ahead and use logic level components if it is going to make more sense in this particular design. Although they have some disadvantages, they have upsides too, and otherwise they wouldn't be manufactured at all. Some of them are designed to switch quickly, having a low gate charge, and with tolerable "on state resistance" as well..... Look at the Fairchild FDB7030BL for example.... Logic level MOSFET comes as either a surface mount part or through hole, with a 60 amp continuous current rating....
Here's the datasheet: https://www.fairchildsemi.com/ds/FD/FDB7030BL.pdf

ALSO please keep in mind that when a datasheet says something like 60 amp continuous current rating, that this is ENTIRELY dependent on ideal conditions as far as thermal dissipation and switching profile is concerned, and that actual power dissipation and continuous current will be lower depending on your heatsink and particulars like the applied gate voltage and switching frequency.....

There is a lot to know when you get into working with high power semiconductors, especially when it's at higher frequencies and not purely being used like a light switch or something.....

Good luck my friend, I hope this has been helpful!

 

[meltbox360]

It has been very helpful and I also believe the thread title question has been answered so I will keep my other questions for another thread so other people can find answers more easily. Thanks everyone.