here's a very nice 300A mosfet

here's a nice 300A continous low rds MOSFET. Package would require some tricks to fit. Not expensive too. Avalanche rating much better than with irfp4110. Any drawbacks?

https://www.infineon.com/dgdl/Infineon-IPT015N10N5-DS-v02_02-EN.pdf?fileId=5546d4624a75e5f1014ac94680661aff

https://www.infineon.com/cms/en/product/packages/PG-HSOF/PG-HSOF-8-1/#!products

Looks pretty good. Only drawback is that it's out of stock at digikey.ca :lol:

I guess it's too new

I like the Rds...1.5 micro ohms...impressive!

Still it is a 375 watt mosfet. That's really cool that at 10 volts you can run it at 300 amps, but who makes anything that runs at 10 volts? We are using mosfets for motor controllers, BMS and other things that are running at 48-90 volts so that really means 7.7 to 4.1 amps per mosfet. In that regard...not so nice. Make that hunk of silicon work at lots more wattage please!!!

It does have a nice and fast rise time...30ns.

I wonder if they make this in a To-220 package?

Pros:
Fast and low Rds

cons:
still a boring old 370 watt mosfet...yawn!
Not a To-220 mosfet.

ElectricGod said:

Still it is a 375 watt mosfet. That's really cool that at 10 volts you can run it at 300 amps, but who makes anything that runs at 10 volts? We are using mosfets for motor controllers, BMS and other things that are running at 48-90 volts so that really means 7.7 to 4.1 amps per mosfet.

Click to expand...

How are you calculating that? You need to calculate using the voltage across the MOSFET, not the system voltage.

Looking at just losses from the RDS of 1.5mOhm we can calculate the current for 375W of dissipation. Rearranging the formula P = I^2*R :

I = sqrt( 375W / 1.5mOhm )
I = 500A

Of course in real life there are other losses and constraints so you wouldn't really be switching 500A, but that does set the bar pretty high.

Addy said:

ElectricGod said:

Still it is a 375 watt mosfet. That's really cool that at 10 volts you can run it at 300 amps, but who makes anything that runs at 10 volts? We are using mosfets for motor controllers, BMS and other things that are running at 48-90 volts so that really means 7.7 to 4.1 amps per mosfet.

Click to expand...

How are you calculating that? You need to calculate using the voltage across the MOSFET, not the system voltage.

Looking at just losses from the RDS of 1.5mOhm we can calculate the current for 375W of dissipation. Rearranging the formula P = I^2*R :

I = sqrt( 375W / 1.5mOhm )
I = 500A

Of course in real life there are other losses and constraints so you wouldn't really be switching 500A, but that does set the bar pretty high.

Click to expand...

And in comparison the Rds of the popular IRFB4110 is 3.7mOhm (miliohm, not micro ohm), and it has same max dissipation of 375W but then the max current is only around 100A.

So this one is really nice I believe. Much less heating and losses. I can't find anything in datasheet that would be significantly worse than IRFB4110.

I might order a few and try to swap IRFB4110s, the package is different so it's a bit of a mechanical challenge. I think we should go and design open source controller from scratch, since casainho and bunch are already crunching the software code, it's just the board design that we'd have to come up with, but there are plenty of examples. With these FETs due to lower dissipation we could have much smaller housing than normal 20x4x8cm.

ElectricGod said:

I like the Rds...1.5 micro ohms...impressive!

Still it is a 375 watt mosfet. That's really cool that at 10 volts you can run it at 300 amps, but who makes anything that runs at 10 volts? We are using mosfets for motor controllers, BMS and other things that are running at 48-90 volts so that really means 7.7 to 4.1 amps per mosfet. In that regard...not so nice. Make that hunk of silicon work at lots more wattage please!!!

It does have a nice and fast rise time...30ns.

I wonder if they make this in a To-220 package?

Pros:
Fast and low Rds

cons:
still a boring old 370 watt mosfet...yawn!
Not a To-220 mosfet.

Click to expand...

What do you mean by running it at 10 Volts? The Vds is rated at 100V so you can run batteries and motors at up to 100V theoretically. Idmax is 300A continous, more or less depending on thermal management. Perhaps you are looking at some Vgs? VgsThreshold is pretty standard, and range is -20 to +20V, also standard.

This MOSFET is more efficient than any other I've seen so far mentioned for controllers, at least for 100V. You can probably get lower Rds if you sacrifice a bit of Vds.

1.5 mOhm is a pretty substantial improvement over a 4110. Might hardly need a heat sink.
Mouser has 20k on order, $6.19ea in singles. Not bad.

fechter said:

1.5 mOhm is a pretty substantial improvement over a 4110. Might hardly need a heat sink.
Mouser has 20k on order, $6.19ea in singles. Not bad.

Click to expand...

digikey has 24 weeks lead time and 0 on stock. Mouser claims some 26 weeks factory lead time.

https://www.digikey.com/products/en?keywords=IPT015N10N5

http://eu.mouser.com/Search/include/aoo_popup.aspx?mouserpartnumber=726-IPT015N10N5ATMA1&fromCart=1

price is OK if it's possible to get away with less of them, for example only 6 should equal similar to 18 pieces of 4110 in terms of current carrying capacity.

justin needs to put this in the phaserunner then I don't have to design my own controller as I had to do with the spotwelder.

1.5mOhm means only cca 10W dissipation at 80A. That is both cool and cool.

Looks like a good mosfet, one disclaimer though:

Datasheet says 32A continuous if installed on a 6cm2 copper pad, 2 Oz thick. (40K/W). Thats very far from the claimed 300A at unpracticable conditions, so plan for a big heatsink if you want high currents.

This looks really promising! Thanks for bringing this topic up!
Does anyone know what kind of mosfets adaptto uses? I wonder if this could be a worthy upgrade to the current mosfets they're using

marcos said:

Looks like a good mosfet, one disclaimer though:

Datasheet says 32A continuous if installed on a 6cm2 copper pad, 2 Oz thick. (40K/W). Thats very far from the claimed 300A at unpracticable conditions, so plan for a big heatsink if you want high currents.

Click to expand...

I am not familiar with this RthJA spec, but with rds=1.5mOhm it has much much lower dissipation than the rest of normally used MOSFETs, so other MOSFETs like 4410 should be worse. With 80A this new MOSFET should dissipate only 10W which does require heatsink but quite manageable when mounted to the casing. Can you elaborate more?

10W is a lot of heat for something with the tiny thermal mass and surface area of a bare mosfet. There are also switching losses to consider as well as the conduction loss from Rdson

Punx0r said:

10W is a lot of heat for something with the tiny thermal mass and surface area of a bare mosfet. There are also switching losses to consider as well as the conduction loss from Rdson

Click to expand...

Yes of course, heat needs to go somewhere, I forgot about switching losses in the excitement of discovery. But anyway, still Rds and hence typicall dissipation should be some 60% lower than with 4110 FETs.

Punx0r said:

10W is a lot of heat for something with the tiny thermal mass and surface area of a bare mosfet.

Click to expand...

I've had soldering irons that are only 15W or less, so...yeah, 10W is a lot for a small component.

if it's an SMD part, though, the intent with those is to usually to have the copper layers of a multilayer PCB, with vias thru the layers under the FET body to carry the heat between the layers, then to have large exposed planes on the outer PCB layers to dissipate that (possibly thru heatsinks, if tehy're still needed).

The 4110 is a very old FET. There are many better. I recall TI had some that were quite popular years ago (CSD19536KCS), and a lot lower cost. IXYS has some excellent ones. Even IR has quite a few superior models like the 4468 (Adaptto) compared to the ancient 4110.

Infineon parts are often hard to get at decent prices for real people, they cater to large customers. Unobtanium parts don't do us a lot of good.

Look at the hot FET parameters, those are more important than the cold specs. We don't operate them at 25C. Look at thermal resistance, the ability to get heat out of the package. 0.4 K/W max is not great. Both 4468 and AOT290L's (PhaseRunner) are better there.

Don't forget to review the body diode heat at current, this often contributes more heating than the FET in motor control operations (depending on the design and software). 1.2V max at 100A may not best in class, though this spec is often measured at different currents so comparing is not straightforward, and this is a lot of heat to deal with.

Since heat is I squared R, improving R increases current capacity only by the square root of the improvement in R. So the improvement is tempered by the math.

There are so many specs on FETs, it is easy to overlook something important. The datasheets are carefully constructed to look good. Reality is often slightly different.

Not to say this Infineon is a bad FET, it is just not easy to compare all the factors at once and relate to actual real system performance, and availability may be an issue. Packaging is not a drop in for most designs, eg the PhaseRunner would require changes to be able to use it.

A good approach would be to use an aluminum PCB. I've seen a few of these around lately and seems to be a great solution to heat sinking SMD parts.
Here is a description: http://www.amitroncorp.com/printed-circuit-boards/aluminum.html

The aluminum PCB then gets bolted to a big heat sink. Thermal resistance from the die to ambient will be much lower than a typical TO-220 design.

If I were to use it I'd use A TON of vias, a very thin pcb and bolt the board on a large heatsink with heatsink paste. The best dissipation I've seen was a pcb made of 0.2mm ceramic substrate attached to an aluminum enclosure. Nowdays that is an integrated process and its how led bulbs are made, those led circuits seem much simpler than what you will find in a switching application though.

Anyway, I guess vias+thin pcb+heatsink gets fairly close to an aluminum substrate pcb. If you wanna do better just change FR4 by ceramic($), if you wanna do even better go for the integrated aluminum substrate process ($$).

You can calculate how many watts it will need to switch at 20KHz given that 200uC figure from the datasheet, its an important share of the total power losses.

That package has limited production use in large parallel arrays due to legs being so rigid and tending to have PCB trace fatigue and cracking from thermal cycling stress.

It's an excellent part if used correctly, but doesn't lend itself well towards an aluminum PCB due to cooling, and has limited alternative cooling options. D-pak7 has a little leg bend flex, but for ultimate power density it's a ratio between RdsOn loss (heating) + switching loss + diode loss, and the real packaged Rth you end up with on the system for cooling.

liveforphysics said:

That package has limited production use in large parallel arrays due to legs being so rigid and tending to have PCB trace fatigue and cracking from thermal cycling stress.

It's an excellent part if used correctly, but doesn't lend itself well towards an aluminum PCB due to cooling, and has limited alternative cooling options. D-pak7 has a little leg bend flex, but for ultimate power density it's a ratio between RdsOn loss (heating) + switching loss + diode loss, and the real packaged Rth you end up with on the system for cooling.

Click to expand...

good point about thermal stress and cracking connection between legs and soldering point, it wouldn't occur to me. Through-PCB legs are better suited for ebike with the thermo-mechanical stress.

Still, the overall characteristics on paper are quite appealing and I'll try to get my hands on some when they are out and convert one of my controllers to see how they behave. Perhaps I can test it to the limit by converting a small 6FET controller and push it to some 80-100DC/150-200A phase and if see if they are worthy. I'll open a thread if I get to that point.