Paralleling Mosfets in smps for Cooler Operation

Yes.

Yes :-)

You need to make sure that the driver has enough gusto. If it doesn't then the slopes flatten a bit and you might not gain anything, or worst case even lose efficiency.

I won't get into that battle of opinions again. It's like arguing ground planes. Or, ahem, politics.

While a gate resistor or bead can avoid oscillation they may not be necessary with clever placement and layout. I usually was able to do it sans resistance. A gate resistor will cost you in switching losses. To me they often look like taking Maalox everyday (some people do that...) instead of changing to a proper diet.

This here designer has done that. Many times. Not so much for temperature reasons but for more efficiency.

Ahm, that would require more data from your side. A lot more data. If this is a 5kW switcher I'd be inclined to say no, if it's 5W then probably yes.

--
Regards, Joerg

http://www.analogconsultants.com

Switching losses will increase with paralleled fets, because more capacitance means more switching energy dumped. How important this is depends on the switching freq and such.

10-30 is a reasonable range to kill birdies.

Even single mosfets will often make rf bursts on rising/falling edges. I'm not sure what determines this, but there's no obvious (to me) relationship between "good" and "bad" layouts here.

John

True. It will increase the losses in the driver itself because it has to charge more capacitance. What I meant were the losses caused by sluggish transitions if the driver he has now is a bit too wimpy for more than one FET.

Agree. But for a big FET that carries a penalty.

A big sin is sloppy layout at the drain node. Long traces that are too thin and stuff like that. Same for long gate traces, those can cause real grief. Most "singing" that I noticed was due to that. It can also happen when the driver is wimpy and can't hold the gate up while the drain node roars down. In one case they didn't want to change the parts placement and there wasn't any space to begin with except for a gate trace of 1/2" or so. I put a pnp/npn follower there which made a huge difference and also notched up the efficiency by a couple percent.

Another factor is the transformer/inductor. In most cases I wasn't happy with catalog parts and we rolled our own. Surprisingly that often ended up being cheaper for production than the catalog part (when made in Asia).

--
Regards, Joerg

http://www.analogconsultants.com

Wow..that was a fast answer :) Thanks... This will keep me thinking for awhile.. Very interesting bit about the gate resistors too..

About the paralleling up to no heat sink: It's for a 130watt Cuk converter running at 100khz. Maybe it's possible to "make a heat sink out of mosfets" :) I could prepare a LTspice file and put on my website for download?

Heat sinking question: My estimated mosfet total power loss is 3Watts at 25C on paper. I suspect my single mosfet heats up due to switching loss causing Rds(on) to rise leading to more conduction loss.. Heat making heat. Does this happen? Efficiency depending on heat sink quality?

D from BC

A schematic would be nicer but I won't have much time since I am in the middle of a large design. With digital stuff in there, yuck...

But maybe someone else can take a look at it.

I can't imagine that happening unless you are smoking them. But quite frankly I wouldn't want to run that without any heat sink. Can't you bolt them to the chassis via isolation layers? While maintaining safety, of course.

--
Regards, Joerg

http://www.analogconsultants.com

Inductors and transformers are like sheet metal and heat sinks, often a lot cheaper for custom parts than for catalog stuff. That's kinda counter-intuitive.

Tried Minntronix? Excellent magnetics house. Ask for Butch.

John

Not yet but after a recent post from you I added that link to my wiki. It's always good to know a source that others were happy with. Right now it's mostly optics stuff I am doing but there sure will be more transformers. Then I'll ask for Butch. And whether there are any of those cookies left...

--
Regards, Joerg

http://www.analogconsultants.com

I used a crappy temporary heat sink on the mosfet (1/2"x 4"x 1mm aluminum strip) for out of the box bench testing. No grease. The top of the TO220FP case almost reached 100C after 1minute. Then I unplugged it..So that's go me freaked.. The control is stable. It does get bolted to a thick aluminum extrusion once fully assembled. I was just trying a crude way of evaluating efficiency with the temp heat sink. I did think about dumping the whole smps in water and measuring a temperature rise..

I should be able to research the rest confident that it's not a dead end road. I'll work on the math for optimum mosfet specs for paralleling and make sure gate drive as good as it gets..

Maybe post some of that yucky digital stuff on here :) Thanks again.. D from BC

Well, that sure ain't much of a heat sink. Maybe a wider piece?

Your best friend is the scope. Check out what happens upon turn-on and turn-off. Also across the Isense resistors (hoping there is one). That gives you a good indication of whether something is out of whack. Make sure you don't fry it with any spikes. If in doubt use a resistive divider so some more sane voltage level.

Most of the efficiency problems I saw were due to sluggish turn-on and core saturation.

It's not ready yet. I am wrestling with ADC, DAC, SPI, addressing and all that. It is like eating oatmeal soup. I don't like oatmeal soup...

--
Regards, Joerg

http://www.analogconsultants.com

calorimetry can be a lot harder than you expect.

its often easier to measure the DC input & output power - if you have reasonable input & output filters. if the current is all nasty and pulsating, this is not quite so easy.

one gotcha with FETs is the positive tempco of Rdson. If you dig up a copy of the Siliconix MOSPOWER applications book, there is a paper on it in there (by Rudy Severns IIRC).

in a SMPS the FET current tends to be independant of Rdson, os Pfet = I^2*Rdson.

causing dTj

causing Rdson to increase

causing more Pfet....

there is a critical heatsink thermal resistance above which you get thermal runaway (little dutch timebomb, tick tock boom)

below Rtheta_kaboom, it doesnt run away, *but* the junction temperature will end up higher (sometimes a lot) than you expect.

For a prototype, just use a monster heatsink until you have the whole thing running, and can make an efficiency measurement.

Then try the real heatsink, and make another measurement.

That'll tell you straight away if its thermal runaway (kaboom) or thermal step-away (efficiency drops, Tj higher than you expect)

Cheers Terry

[snip] [snip]

Huhhh...I wish ...I made a damn offline smps. Can't scope ground it or poof! But it's gotta be done... (Using a Hickok scope from 1968!! Only 8 knobs :) I think I have to buy a >100W power transformer (or equivalent transformer combo) to isolate the smps supply under test. That's going to cost $$$.

Or brainstorming ideas such as: Make an optocoupler probe.. Make an op amp differential probe.. Make HF transformer probe.. Buy an isolated digital scope D from BC

Those sneaky mosfets.. :)

I'll try a Pin Pout measurement/comparison to see if I'm getting mysterious losses.. The continuous mode Cuk has easy waves to work with.. Triangle Iin& Iout currents and Vin has 120Hz ripple and Vout is nearly DC. Thanks D from BC

when I test smps I power them up from a lab supply (to smps startup)- this lets you check your feedback loop (use another supply to crank up the output, at the desired setpoint your duty cycle should drop to zero), gate drive, oscillator etc.

Then use another lab supply to slowly wind up the DC bus. If you scalethe switching frequency down, you can organise for the right peak current in your magnetics, and therefore test for saturation.

or get really keen and get a variac (I bought a big 3-phase variac, dismembered it and sold one phase for slightly more than I paid for the entire unit).

light bulbs are also very handy, and cheaper than exploding FETs

its low power, so scab a pair of transformers from something, and hook them up sec-to-sec.

I've got a 2.5kW isolation transformer I bought from a 2nd hand store for about $25.

waste of time really. besides, when looking at a nasty waveform, how can you tell if its the circuit or your probe?

the new Tek scopes are fabulous in this regard.

Cheers Terry

[snip]

I didn't like any of my probe ideas either.. So it's down to surplus pwr xformers or new scope.. Mmmm chunk of iron...or new toy... :)

D from BC

I think the issue of oscillation using paralling MOSFETs is relevant when you operate in the linear mode of the MOSFET. When you drive it hard on, then there should be no problem (the internal of a trench MOSFET is parallel devices, isn't it?)

In a current design we are using 8 small MOSFETs in parallel instead of one big MOSFET. The losses are lower even though the resulting RDSon is the same, since the MOSFETS cool to the PCB (so there is less thermal resistance) and there is no hotspot on the PCB (there was before since the PCB has significant thermal resistance). Since the hotspots are gone, the individual MOSFET runs cooler. Remember RDSon is about 2 times larges for a MOSFET running hot). Lastly, we can avoid the heatsink, since the MOSFETs are distributed on the PCB to allow the entire PCB to act as a heatsink.

Regards

Klaus

It would be helpful if you told us the MOSFETs and drivers you are using, and some of the other operating parameters.

At the frequency you are working at it, it is normal usage to place a Zener between the gate and source in bridge based switchers to reduce the oscillation problem. If overheating worries you so much use MOSFETS with inbuilt protection circuitry. The PHILIPS TOPFET range (BUK100 ) has thermal protection incorporated, and probably protection against short circuits and other stuff too.

For paralleled use though derate the current rating by 20% for 2 MOS's and as you add more increase that. Differences between them will lead to different currents flowing in them. Though this isn't a big problem in most cases because of the positive temperature coefficient of the drain to source gates, be prepared for the worst.

Andy

Take a look at the multiphase converter topology. International Rectifier

and Texas Instruments make parts for this approach. Basically, you have a separate inductor for each mosfet so the current is evenly divided among them. The multiphase part refers to the fact that they are sequenced to be on at different times so that the magnitude of the ripple current is less.

How does a zener do that?

John