Multiple mosfets in parallel

Hi guys,

The title says it all. I have an idea of stacking mosfets to decrease on-resistance and increase wattage. Will it work? What is the side-effets of stacking mosfets in parallel?

I read that some car amplifier manufacturers configure their mosfets in parallel. Apart from that, the mosfets need to be 'matched' like cells in a batterypack if not, one will have to dissipate more current/heat from another.

Will a stack of mosfet work better in terms of efficiency in a switching power supply? Even with the low rds on.

vince.

Hi there Vince,

How about a 48v, 200 Amp power supply? IRF has one
somewhere in it's reference section where they parallel
lots of IRF150's.

Parallellllllling multiple mosfets isnt that hard.
The drains and sources are tied together, while each
gate gets its own drive resistor (like 10 or 15 ohms).
That's the trick...make sure each device is driven with
its own gate resistor.

For simple LED flashlights you probably could find a unit
that had very low Ron however. 0.008 ohms is low enough,
isnt it? Or do you need lower?

Two mosfets will require twice the drive power of one.

Take care,
Al

Ok if the power required to drive two mosfet is larger than current wasted going through 1 mosfet then I guess it's really useless stacking them.

Hi again Vince,

With our LED lower current requirements i would bet it would
be a waste, but then if you happen to have several laying
around you want to use up and you need lower Ron maybe
you could at least use them for something right?

Take care,
Al

One thing that helps parallel MOSFETs is that if one is lower resistance, it pulls more current, but then the current causes it to heat up, and that MOSFET's resistance goes up.

But you should be easily able to just find a better part, since at the tiny currents found in flashlights, there are thousands of suitors to choose from.

My fairly modest experience with paralled pass elements is you get caught up in beta (for bipolars) and Vgs (for FETs) differences. Without modest size resistors in emitters or drains to force current sharing the first one to conduct does all the work, or too much of it. The whole bank can go as the load gets transfered around as the transistors pop.

They can be matched, of course, and switching is likely to be less demanding than linear use, but there's a clue in the fact that switcher makers don't do this. They look for more appropriate parts?

Doug Owen

Funny timing.. I just this morning posted how i can't stand the fact that twice in two days i ended up massaging a kroll switch because it had an RDSon of 0.7 ohm which is about 2x the resistance of one of my FETs i use... and i made a mention in a post that i would love to come up with an electronic version of the kroll.

The first thing i had in mind was to use a pair of IRL 2502s which have an RDSon of 0.045ohm.. that should be as good as any switch..

A single 2502 can handle 4+A if kept cool.. and 3A up to 170C.. i just looked up the chart and it's voltage drop at 4A is less than .03V at 4.5V gate voltage.. i.e. if you used just one you could switch up to 4A through 3D batteries and lose only .03V on the FET.. which means .015V if you use two of them.

Well time to design my 'digi-kroll' i guess /ubbthreads/images/graemlins/wink.gif

You have to watch the gate cap loading (Miller Effect) but it's done all the time.

andrewwynn,

That does sounds like a fun project - the mechanical switch could be horrendous in terms of resistive load, but driving a low RDSon would mean you really don't care /ubbthreads/images/graemlins/grin.gif

If you put this switch on a tailcap, how would you get the right voltage to "activate" the FET? Or would this work better in a M*g style project where you would have access to both the "+" and "-" sides of the battery?

Will

the main reason there isn't a 'gimmie' electronic kroll is exactly as you ask here.. how to power the circuit... i've seen several solutions to this on CPF with electronic clickie switches.. some of them run a lead up along the battery (in the VIP there is so much extra room i could fit a 18ga wire in there).. but just a 30ga wire is enough.. but computer ribbon cable for toughness and thinness is what i'd be aiming for..

easier to put into a mag since both + and - are available.

[ QUOTE ]
Doug Owen said:
My fairly modest experience with paralled pass elements is you get caught up in beta (for bipolars) and Vgs (for FETs) differences. Without modest size resistors in emitters or drains to force current sharing the first one to conduct does all the work, or too much of it. The whole bank can go as the load gets transfered around as the transistors pop.

They can be matched, of course, and switching is likely to be less demanding than linear use, but there's a clue in the fact that switcher makers don't do this. They look for more appropriate parts?

Doug Owen

[/ QUOTE ]


Actually, paralleling MOSFETs is a very common practice in modern power supply designs. They don't utilize resistors in the drains, either. In fact, in many Pentium 4 motherboards, as well as Athlon 64 motherboards, you will find paralleled MOSFETs. The other technique is to have multiple switchers in parallel, but each ran out of phase, to share the load. This reduces inductor and capacitor requirements.

An example of a multiple phase setup is found here:

Multiple phase


Here is an example for even the older the Pentium III where there are two MOSFETs in parallel on the top, and three in parallel on the bottom. Take note of the x2 and x3 by the MOSFET.

Parallel MOSFETs

Another common reason for paralleling MOSFETs is that it is often desireable to spread the heat out over an area on a board, so you don't require heatsinks (reduces hotspots)

Hi again folks,

The timing of this thread is pretty good for me, as i've
been thinking of building a 120vac to 12vdc,30 amp converter
to run my cordless drill. On the other hand i'd probably
go with single MOSFETS, but then depending on price i'd
at least like to check into parallelling.
Then on the other other hand ( /ubbthreads/images/graemlins/smile.gif ) then i wont have a use
for my big jump pack again. It was kinda nice having a
good use for it after owning it for maybe 2 years and not
using it much except for finding out what i can run with it.
I ran a small 4.5v drill with it too, with a big bulb
in series with it. That drill never worked better /ubbthreads/images/graemlins/smile.gif


Take care,
Al

Yeah it's good timing for me too , looking at the bunch of cheapo mosfets sitting in my stash. Alot of info to absorb here, thanks guys.

Andrewwyne - please share with us your electronic kroll if you do make one. Here's my ikroll plug that I scratch. It works but not reliable because of the power routing. I have to hats off dj for his incredible Pic switching.

See this is what I meant by stacking fets...


webpage

Hi again Vince,

OH yeah, he he, that's really stacking all right!
I guess they ran all the gates in parallel too.

Take care,
Al

Koala.. i remember breifly seeing your iKroll.. to make your +feed 10x as reliable.. use a chip from a rare-earth-magnet to hold the power on the SIDE of the anode button.. or even design a plug that plugs into the little vent hole (there are usually 2 so no worries)..

i'm planning on something WAY simpler.. just a simple FET driven by a button Lithium battery that will be nested under the main spring i imagine.

it will only draw FET GATE voltage when on only.. so battery should last years.

I'm completely fed up with the horrible resistance on the krolls i've got in my lights.. i have had to reswork 2 and another needs it already.. how did you get the kroll apart is what i need to know. I need to re-do the inner spring.

-awr

There are at least 5 mechanical contacts in the kroll, I guess that's what makes the kroll a variable mechanical resistor.

awr - if you look at your kroll, there's a body spring that makes contact with the battery tube and threads. If you can insert a flat blade screwdriver under the spring(so the spring doesn't get fouled) and twist, the top part will come off. Don't take this as the best way.

Inside, there's 4 mounting post. 2 or 3 will be bound to break and if you are lucky, 1. I am not sure if they are secured by cyanoacrylate glue but at times, little pieces broken from the top part will remain attached to the bottom piece. No worries, there'll still be a tight fit without the mounting post. My ikroll picture shows the top part of the kroll without the 4 mounting post. You will probably need to shave them off to get more real estate.

-vince.