New Pick and Place machine arrives at Castle

[Pdelcast]

Originally Posted by NoControl

Hey pd ever think of making an esc out of fets like sure this ones dead but what do you expect when a EMI kickback of 3kv and 20a shorted arcross while switching a direct feed from a wall out let, but anyways I mean these things would make a mamba system next to industructable, with ratings of 340v @ 100A.

They may look indestructible, but in reality they are not suited for the type of controller we build.

That is a 200V, 46A (rated) FET you have pictured there, with a resistance of 55 milliohms. (in practical terms, this FET can handle about 10 amps with a large heat sink -- about 5.5 watts of waste heat.)

The FET we are using in the Mamba Monster is a 30V, 191A (rated) MOSFET, with a resistance of 2 milliohms. (in practical terms, this FET can handle about 25 amps with a minimal heat sink -- about 1.25 watts of waste heat)

We are using 36 mosfets (6 per phase-half) in the Monster, so the actual resistance is about .33 milliohms.
Using 36 of those IRFP260s in the same configuration would yield a resistance of 9 milliohms, and would need a shoebox sized controller... seriously.

At 150A, the Mamba Monster dissipates about 15 watts of heat from the FETs.

At 150A, the IRFP260s would need to dissipate 405 watts of heat. Ready for a 25 pound heat sink bigger than your truck??

And, it wouldn't be indestructible -- quite the opposite. The dies would run at VERY high temperature (due to junction to heat sink thermal resistance,) and the controller would have a short life.

It would, however, run at much higher voltages. So, you would need to run, say, 40S lipos (electrocution danger?) at low current (10 amps or so...)


In practical terms, the best configuration for an ESC today is lots of large-die, small package MOSFETs in parallel. The package we are using (SO-leadless) handles a huge die, but is packed into a small package. This allows us to place lots of parallel MOSFETs into a very small area, and maximize efficiency. Large die, large package MOSFETs like the one you have pictured would require us to use fewer MOSFETs per controller, and would have significantly higher resistance -- thus higher losses, larger heavier heatsinks, and lower efficiency. In the long run, this means higher cost, heavier controllers, and less power to weight ratio.

What we try to do in ESC design is minimize waste heat generated in the controller -- heat is the enemy of controllers. Heat losses are equal to (current squared) times the resistance of the MOSFET (this is known as "I squared R losses", often written as I^2R.) So in designing an ESC, we try to minimize R to the lowest possible level, while still making an affordable ESC.

[Pdelcast]

No video yet of our new machine running (I have to figure out how to get a camera into a position where you can see something actually happening...)

But here is a video of an AC-30L running, which is the same as our OLD machine. The AC-30L (which we traded-in on our new machine) is a single head rotary drive machine with 30 spindles. Our new machine is a four head, 120 spindle machine, and is about 20% faster per head due to the linear drive.

Remember when watching this that our new machine runs like this one, but has FOUR heads working at the same time.

http://www.youtube.com/watch?v=oEz4L...eature=related

And here is another video of a Genesis (linear drive) single head machine running:

http://www.youtube.com/watch?v=Wd-DU...eature=related