0

I'm trying to design a BLDC motor controller rated to run a 200kW electric motor, 240v dc ~833A. Looking at MOSFETs I've found a TO264 n-channel which is rated for 500V 100A. How are the legs not going to immediately act like fusible links and vaporise?

Edit: I've found a more suited MOSFET [IXFB210N30P3][1] which is 300v 210A Rds of 14.5mOhm continuous power of 1890W.

[http://ixapps.ixys.com/DataSheet/DS100463A(IXFB210N30P3).pdf][1]

Edit Dos: it appears, in the data sheet, that there is an external lead current limit of 160A and that the chip capability is 210A

Tanenthor
  • 250
  • 2
  • 10
  • 1
    Can you link the datasheet? – tangrs Jun 16 '16 at 12:46
  • Assuming like 5mm lead length, one would assume 500µΩ resistance, thats about 5W, hardly a vaporizing fusible link – PlasmaHH Jun 16 '16 at 12:48
  • @PlasmaHH What about Rds(on)? That's usually in the tens of milliohms range right? In that case, could the MOSFET be dissipating a significant amount of heat? – tangrs Jun 16 '16 at 12:50
  • @tangrs: First, there is no "usually", there are different FETs for different purposes, and secondly, you specifically asked about the legs. For the Rds, check the datasheet, we can't. – PlasmaHH Jun 16 '16 at 12:52
  • Slightly unrelated, but have you considered looking in to IGBTs. They tend to be higher voltage/current devices. – Tom Carpenter Jun 16 '16 at 12:57
  • @PlasmaHH That's fair enough. Btw, I'm not the OP. I was just curious – tangrs Jun 16 '16 at 13:02
  • 4
    @tangrs: I am an old mips C compiler, three chars of the identifier matching is good enough ;) – PlasmaHH Jun 16 '16 at 13:12
  • The latest device has a 300V Vds rating. You said 500V. What is the supply being used? 12V 100V ? ...? | And what max current do you wish to operate at? As power dissipation rises with I^2 it makes a substantial difference. – Russell McMahon Jun 16 '16 at 23:30
  • @tanenthor you could parallel devices assuming that the switching is matched sufficiently (if one turns on first to fast then you have the same problem). Paralleling is a good way to solve thermal problems – Voltage Spike Jul 20 '16 at 17:51

2 Answers2

3

Datasheet link & device name please. !!!!
You will often see the notation"package limited" for very high energy devices.

Perhaps a IXFK-FX98N50P3 Rdson <= 50 mOhm - wow. 500W dissipation at 100A :-)

About 0.2 C/W Rth_cs
So 100 C junction rise above heatsink.

I pulse max is 245 A T junction limited. This is "really" a 49A part and still rather unhappy at that level.
About 2.25V Vds at 25C at 50A and about 5V at 125C Tj (figs 1-3)

So still 250W at 50A once it gets hot.

The only reason you'd use a device with this apparently abysmal spec is its 500 V Vds rating.

________________

For about double the price ($US27/1) STY139N65M5
17 milliOhm at 25C. About 2x that at 125 C. Still an inadequate package. __________________

For $US59/1 you get an Isotop pkg that starts to work. 650V 143A. Rdson is 12 mO typical at 25C and a bit under double that at 125C. Tjc is about as above but the package can tolerate heat and current. melting the screws would be a challenge - not an unmeetable one :-).

STE145N65M5

enter image description here

Russell McMahon
  • 147,325
  • 18
  • 210
  • 386
1

Assuming that the legs are made of steel, that gives a resistivity of 3*10-18 ohm metres. Going from the dimensions of the legs, that means that the resistance of each leg is roughly 300uOhm/mm, meaning that each mm adds 3W of heat. As another person said, this is quite a lot, but probably not enough to melt it. Bearing in mind that this is the absolute maximum, and you're already going to have very good heatsinking otherwise the mosfet itself would break.

The IXFK98N50P3 has a pretty low Rdson of 50mOhm, and to keep it that way you'll have to keep it at 25C, meaning you'll have to get rid of 500W of heat. Looking at the junction thermal resistance, that means keeping the case 48C colder, so -23C. This obviously means a lot of active cooling and some serious thermal design.

BeB00
  • 5,303
  • 2
  • 18
  • 34
  • onsemi has a nice AN about TO220 leads (should be comparable here) which have 70-140mΩ/m, making 100µΩ/mm a good rule of thumb, thus at 100A 1W per mm. – PlasmaHH Jun 16 '16 at 12:56