Just how hot can things get?

[itsmoked]

Just designed a switching charger.

Input: 24VDC
Output: 5-30VDC
Output: 10A

My normal designs nurse nanowatts out of coin-cells and run for years on a battery. This thing funnels 300W thru FETs and blocking diodes etc with 28A thru an inductor.

Anyway, steady state conditions are shaping up to parts on the board under convection cooling hitting 80C. Nothing seems too mad about it but I am not used to feeling heat radiating off my happily running boards.

The hottest part is the blocking diode. It has the worst forward drop and hence the highest dissipation 80C. The part can run the junction @ 150C so I am well below that. I am not worried about torching anything, I am more worried about longevity. On the one hand this is a charger application so I expect maybe 24Hrs charging cycle. Initial time charging some dead battery, running the full 10A, and then tapering later to eventually full cutoff. Then not charging again except to make up again from an engine start every 2 weeks.

My question is where do you guys consider power electronics to be running too hot for reasonable life expectancy? A lot of off-line switchers have to do this endlessly when do you guys start reaching for fans and heat sink forests?

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[LionelHutz]

A couple of things;

You can get a Wakefield 217 series surface mount heatsink that goes over top of the D2PAK type of devices. You have a copper pad under the device and you just put solder strips on each side for the heatsink to mount to. Not as effecient as direct connected thru-hole I'm sure but should lower the temp.

Are you using thermal via's with the "fingers" connecting your plane to the hole? I don't know the technical name but thermal vias are done to keep soldering heat from being transfered to the planes. If so, try going to straight holes without the "fingers".

Have you looked at International Rectifier?

 

[MacGyverS2000]

Thermal relief pads ("fingers") should not be used for high-current applications in the first place, for obvious reasons. They're typically used when the components will be hand-soldered, so for production line reflow/IR soldered there's no need for them anyway.


Dan - Owner

http://www.Hi-TecDesigns.com

 

[Comcokid]

You could try using Silicon Carbide Schottky diodes. They are more efficient, faster, and handle higher heat than standard silicon. But, they are expensive.

Following link provides some data. This company appears to manufacture SiC diodes mostly for aerospace applications. There are other companies making them.

http://www.sensitron.com/Hi_brochure/SiCSelector.pdf

 

[btrueblood]

Isn't there a way to "wire" a FET (e.g. MOSFET) with the gate tied to the high side, and have it act as a diode? Seems like you can get very low Vf that way, but I'm probably missing something important. What is the drawback to that configuration, is there some circumstance where it smokes and stops acting like a diode? Is there a way to add some circuitry to avoid smoking the FET?

The wikipedia article for Schottky diodes mentions the reverse leakage going high as junction temp. increases being the main drawback to them, and then suggests an "ideal diode" (FET with control circuitry) as being a possible fix. The circuit they mention doesn't come anywhere close to smoked's requirements, so was wondering how you'd build something similar.

 

[BobbyNewmark]

"Isn't there a way to "wire" a FET (e.g. MOSFET) with the gate tied to the high side, and have it act as a diode?"

Probably, but it does not quite work. The rectifier's forward voltage would be about equal to the MOSFETs threshold voltage of 2 to 5 volts. Power dissipation would be very high.

What you can do is get a switching controller chip that can directly drive the MOSFET's gate at the appropriate time, turning it into a synchronous rectifier. This gives very low losses in the forward direction, and low leakage in the reverse direction. (Many designs also put a modest Schottky diode in parallel with the MOSFET. The diode reduces the voltage swing on the MOSFET's large capacitance, which saves a little power. It also is much gentler to the MOSFET's body diode.)

 

[jimkirk]

For silicon based parts and good reliability, I'd recommend a maximum junction temperature of 100C under normal conditions, and 20K below the rated temperature for fault conditions.

 

[itsmoked]

Yep Dan; you can find 2x20A diodes in SMD I will probably try that. I just figured out that as the I increases Vf increases which would force sharing by the two diodes in a package. BUT!! Simultaneously as the diode temp goes up the Vf plunges which has a grossly larger effect.. Meaning I have a 10A diode sucking down 10.1A while it's same die neighbor watches with amusement.. Damn!

Mr. Hutz; Thanks so much for that heat sink tip I shall look at it forthwith!
Yes I am not using "thermal reliefs" (the name you're looking for) for those mondo heat transfer vias. Haven't looked at I.R. parts directly - I will give that a go too!

Dan; Ah, I see you beat me to the "name". Good on you.

Comcokid: Interesting.. I have seen many diodes that purport to being higher rated but the prices jumped mysteriously from the $1 range to the 15-25+ dollar range(half my product budget) so I never pursued them. I will check, I bet that's the deal. They look really spectacular which means I bet their price fits that same description. LOL

btrue: This is an area I need to learn more about as I have five FETs on the board now running those currents that have no heat issues at all. One more, works well, in the quantity realm of manufacturing too. Thanks.

Bobby; Dang.. To reiterate: This diode is only to prevent a large battery from back feeding a broken charger... All my other FET's are high speed switching and have the parallel diodes specifically to deal with the slow FET's diode. I have a micro with about 30 unused PIO pins that could run an output FET diod replacement but fear a crashed CPU "allowing the beast to get in". Perhaps a High Side driver somehow controlled by a smaller voltage drop(much smaller) would work.... I already have a highside driver that works great doing a standard task on the board, another would fit easily.

I also have a non committed single sided OPAMP doing nothing in that same corner. (LM2903) Thoughts anyone?

David; thanks for the Topic fix!! (If I didn't thank you previously.) I shall remember that.

jimkirk: I need to see what those SiCarbide puppies cost. I hear you're admonition. Or were you speaking to std Si diodes which I have problems finding period, in that format. Or even the heatsinkable TO-220 flavor, probably owing to the nearly double voltage drop and greater yet heating.. Thanks.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[logbook]

International Rectifier 30LJQ045
30A schottky at 45V.
(Use a 40V if you can to improve the volt drop).
I think a single schottky is going to be better than a double paralleled for your application.

The one above is probabaly too expensive being a HiRel part, but it shows what is possible. Clearly the hard part is fitting the larger die into the package.

 

[logbook]

General Semiconductor MBRB2545CT
A new part. 2x30A at 45V in D2PAC.
That should do you.
Found on the RS website
rs

http://www.com

part number 215-7760
£2.19 each and in stock!

 

[jimkirk]

Hi Keith,

I was referring to standard Si semiconductors. SiC can generally handle much higher temperatures, but keep in mind that it's often the epoxy plastic outgassing within the package that can cause the reliability hit.

If you have to live with a hot junction, consider a metal or ceramic package.

Also, be sure to use a circuit board with high enough glass transition temperature or do something to mitigate that.

 

[MacGyverS2000]

I have used reverse-biased FETs in the past for power backup solutions (sounds like what you're doing here). Linear has an app note using a PFET, an NFET, and one of their voltage monitors to switch between "line" voltage and a battery backup when the main line drops too low... I'll see if I can dig it up.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[MacGyverS2000]

My bad, it was Maxim, not Linear:

http://www.maxim-ic.com/appnotes.cfm/appnote_number/3864

Dan - Owner

http://www.Hi-TecDesigns.com

 

[Madcow]

We call them or-ing fets.
Many chip manufacturers make them.

http://www.maxim-ic.com/solutions/power_non_portable/parts.mvp/scpk/1632

http://focus.ti.com/docs/pr/pressrelease.jhtml?prelId=sc07025

http://www.irf.com/product-info/datasheets/data/ir5001s.pdf

http://www.intersil.com/product_tree/Power_Supply_Support.asp

 

[Madcow]

Curse you Macgyver!
Your computer must be faster than mine.

 

[btrueblood]

HA! Thanks Madcow and mac. Silly mechanical type (me) saw an "ORing controller", and skipped it (wondering why anybody would need to control a circular rubber seal, they do just fine by themselves .

 

[itsmoked]

Ooooookay. I've ordered a potpourri of heat sinks partly to see how they'd fit this layout and partly to see how SMD heat-sinks perform. Pretty cool - you just pick-an-place the parts then P&P the heat-sink over the part and re-flow. {Should be fun to hand solder!!!}

I also bagged some TO-220s to keep the show on-the-road if the SMD sinks don't quite cut-it.

I also found a diode that is a "super fast" diode that has zilch reverse leakage - even when cooking. It has higher Vf but a more linear Vf-I that I can take advantage of because it's a 30A single diode in a package.

ORing or replacing the diode with a non-micro analog controlled FET needs more thought. I'd need to consider this carefully due to the complex power sources, initial conditions, blah, blahblah. The waste heat that would be involved would be pretty much null and void.. So it's pretty tempting. If a board rev crawls out of darkness I will think a lot about ditching the diode for a FET.

btrueblood; You think ORings take care of themselves?!?! You should see my Mom's Moen shower fixture. It has about 10 ORings in it and about every 6 months one of the ORings goes walk-about never to be seen again. Dribbledribble.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[OperaHouse]

This has been a far more interesting discussion than most posts here. A low cost heat sink that can work in some situations are 1/4 inch male spade connections that solder to the board (bare terminal without plastic spacer). These can be obtained which are 0.6 inches tall. Adding any hardware to a board is a pain in the assembly process. Just throwing that out as an option someone might use.

 

[Madcow]

110 degree junction temp. is the old military standard.
As far as the board, they are rated for different temps.
We moved to 130 degree boards for the surface mount.
SMT power resistors get real hot. Board temps also
become an agency issue if your planing UL/TUV/CE/ect..
Some time I just glue a hunk of heatsink to the part
with some thermally conductive glue.

The SiC diodes maybe worse for you. They are mostly
high voltage. The forward drop is close to 2 volts. A
6 amp 600vdc is running about 5$ in 1k pricing.They
are great for pfc/boost applications. I've gone to
400khz on a hard switched pfc. Prices may drop soon,
with the galium-nitrate diodes coming out.

Nasa thought those Orings took care of themselves too.

 

[rtronics]

Good Day Keith, just slightly off topic, but I wonder if you could tell us what charging regime you are using, ie. voltage, current, time, etc.
Back to topic, also, how does your charger respond to a battery with, say, one cell that has a partial short and will not come up to full charge?
I could envisage a prolonged bout of high current supplied in this situation.

Regards, Ray.

 

[IRstuff]

MIL-STD-883 parts were rated for -55º to +125ºC.

However, if you're looking for high temperature parts, you might consider looking for automotive rated parts, which are often specified to operate at 125ºC ambients.

TTFN

Eng-Tips Policies FAQ731-376