Hi,

I am trying to build an H-bridge using MOSFETS (how many threads start with this?).

I have decided to use P-channel FETs for the high side rather than N-channel (I will admit to not being able to deal with the additional gate drive complexity for an all N-Channel bridge).

I am trying to determine which MOSFETS to purchase with an aim to being able to handle 12 Volts and 30 Amps (and minimal dollars).

Looking at the data sheets for various MOSFETS, I was delighted to find that, in many cases, Ids max will often far exceed my expectations for a mere $1.50 !

Then I discovered this:

http://www.mcmanis.com/chuck/robotic...FET-power.html

and have been depressed ever since .

Given that, for most of the “Through-hole” N-Channel MOSFETs that I have looked at:

• Rtheta Junction to Ambient is 62°C/W
• T junction max is 175°C
• 25°C ambient is almost typical for southern California

am I really doomed to spending $5.00 - $10.00 for a MOSFET with and Rds_on of <2 miliOhms in order to meet my 30 amp goal?

(This is based on:

P = (Tj max – T ambient)/Rtheta Junction to Ambient = 2.4 Watts

Using P = I^2R and a factor of 2 to cover for the increase in Rds_on at Tj max:

2.4 Watts = 30^2 Amps^2 x (Rds_on x 2) Ohms

Which gives a target Rds_on of 1.3 milliOhms.)

I understand that I could “stack” the MOSFETS to try to over come this but I was hoping to keep the final package as small as possible.

Thanks in advance for your thoughts / comments.

 

Paralleling MOSFETs is very common, as it reduces the Rdson by a factor of n, and further reduces the power per MOSFET by the same factor. Hence, two MOSFETs can handle four times the current as one.

 

Same experience here. Use heat sink, paralleled mosfets, and a commercial driver chip. The latter will allow you to use N-channel on top, which have lower RDSon to begin with. John

 

Thank you both.

So the approach outlined for determining the true current capacity of a MOSFET is correct?

(I will still need to be able to assess this whether I stack or use an all N-channel with driver.

 

Is there a reason you can't use a heat sink? Rtheta Junction to Ambient only applies to a device without a heat sink.

Also, I don't understand this:I guess you meant 902 amps^2?

__________________
Ron

 

Roff - sorry it should have read 30^2 Amps^2 - I will correct it.

With respect to the heatsink - I was hoping to keep the weight to a minimum (this is for an RC car application). I will need to start reading up on the heat transfer aspects of this to determine "how much" heatsink I need.

 

What about FDP8860 or 8441?

 

A lot of speed controllers use fans instead of heatsinks. Very effective, but the math is probably a lot harder. See the controllers by "victor".

Or, you could use CPU heatsinks with fans. Some of them are good for more than 100W.

 

Be aware that the simple drive requirements for high-side P-channel MOSFETs is lost if your "motor" or "primary" voltage exceeds the maximum GATE tolerable by the MOSFET (usually 12V or 15V)

If this is the case, the gate will be destroyed with simple pull-up/pull-down, ground reference level shifting methods to switch the gate since they are linked to the primary voltage and will exceed the MOSFET max gate voltage.

Thus, you would need speical gate drivers, and in that case you might as well use NMOSs isntead which are faster, more efficient, and more common.

Congrats on being the first person in the forum that I can remember since I joined that bothered to REALLY find out where manufacturer's MOSFET current ratings come from.

PMOSs are hard to for high current ratings. Much easier for NMOS...
http://www.irf.com/product-info/data...324s-7ppbf.pdf

Don't forget about the losses incurred when switching the MOSFET on and off (on top of the losses from just conducting). THey are often (and ideally) equal to the conduction losses. This definately limits the speed you can switch your MOSFET at, and the gate drive circuit you use (the more current it can provide, the faster the MOSFET switches and the less switching losses there are per switch).

You could just use an IR2010 or some other bootstrap gate drive circuit to drive your high-side NMOSs (add a diode and boostrap capacitor to allow the IC to float the voltage and you're off!).

__________________
NO, that picture isn't me so don't bother asking if we can be friends.

 

Wow - as always, lots to think about.

Ubergeek - thanks for the suggestion (8860 and 8841) - these are pretty close to the 30 amp mark (albeit at $2.50 a pop).

Mneary - I am missing something here.
I get the Rdson part but can't quite get my head around the power reduction per MOSFET part.

Dknguyen - - High praise indeed - Thanks.

I was going to use a simple gate drive for the high side P-channel MOSFETS (pulled down through an NPN and pulled back up using a pull up resistor). Although the pull up would be slow my switching frequency will not exceed 2kHz. The gate driver side would be running at approx 6-7 volts while the drain source voltage would be 12 (or is that -12) volts. Would I still encounter the problem you are describing?

It looks like I will need to spend some time figuring out how these gate driver ICs work as an all N-channel bridge (with parallel MOSFETS and heatsinks) seems to be the consensus opinion as to the best way to go.

Thanks again everyone for your help and advice.

 

Since you are inquiring about paralleling MOSFETs, you should probably know that most of the low RDSon products, are really just a bunch of paralleled mosfets inside a single case.

__________________
www.ps3toothfairy.com -- Full IR control for your PS3

 

If the gate voltage can handle the voltage at the source, it's no problem. Just think of it like this, if |Vgs| max is 12V, if you use a non-floating voltage method to trigger the gate, and the source is connected to a main voltage higher than 12V (liek 15V), when you turn it on by pulling the gate to ground, suddenly |Vgs| > 12V and it dies.

__________________
NO, that picture isn't me so don't bother asking if we can be friends.