Re: Bulk Order 15kW SSTC Boards

["Tesla list" <tesla-at-pupman-dot-com>]

Original poster: Dave Lewis <hvdave-at-earthlink-dot-net> 

In the case of an inductive load, switching time is not such a dominant
term on power dissipation.  The main term will be alone the lines of Pd
= Freq * 1/2 * L * I^2.   The IGBT or FET is going to have to suck up
what ever energy is tied up in the stray leakage inductance of the
circuit when it turns off.

Since this term increases with the sqaure of the current, at high power
levels proves to be a big problem to deal with.  Certain topologies
mitigate this effect handely like a phase shifted resonant bridge.. one
of my personal favorites.

Dave Lewis





Tesla list wrote:
 >
 > Original poster: "Jim Lux" <jimlux-at-earthlink-dot-net>
 >
 > switching speed has a significance over the max rate at which you can toggle
 > it.  While switching, the device is dissipating a LOT more power than when
 > fully on or off.  Slowing the switching speed may radically increase the
 > total power consumption.
 >
 > Consider an example: 100 kHz frequency (i.e. switching twice in 10
 > microseconds).. If you switch in 1 microsecond, you'll be in the "linear"
 > mode for 10% of the time.  For an idealized trapezoidal switch, the
 > dissipation is a parabola waveform, peaking half way through when the source
 > and load impedances are equal. (If the load is, say, 1 ohm, and you're
 > running 1 Volt, the peak dissipation is at 0.5 V output, when the pass
 > device is dissipating 0.25 W and the load is dissipating 0.25W (half a volt
 > -at- half an amp))  Making an (incorrect, but it's late) approximation that the
 > average power dissipation will be half that, we'll dissipate 1/8W during the
 > switch interval, and nothing during the rest... Our 1 W (peak) load will
 > dissipate 1/8*10% = 1/80W in switching losses
 >
 > Now, use a device that switches 10 times faster, so the switching duration
 > is now 1%... Now our device dissipates 1/8*1% in switching losses... if
 > you're talking kilowatts, it adds up quick..
 >
 > This was a very simple analysis for an on/off type switching. If you're
 > switching an Hbridge, then it's a bit trickier (the total power dissipation
 > has two humps, one as one device turns off, the other as the other device
 > turns on)
 >
 > This kind of thing might be insignificant if you were building, say, a 2kHz
 > PWM inverter to run an electric motor... but, when your switching rate
 > starts to get close to the transition time, you're getting awfully close to
 > Class A or Class B type power dissipations...
 > ----- Original Message -----
 > From: "Tesla list" <tesla-at-pupman-dot-com>
 > To: <tesla-at-pupman-dot-com>
 > Sent: Wednesday, September 24, 2003 8:03 PM
 > Subject: RE: Bulk Order 15kW SSTC Boards
 >
 >  > Original poster: jimmy hynes <chunkyboy86-at-yahoo-dot-com>
 >  >
 >  > Hi,
 >  > The IGBT is only slightly slower though. What about an IGBT like
 >  > HGTG30N60A4D? That one's rise +
 >  > fall time is shorter, but has a little more delay. It seems like the
 >  > difference in switching speed
 >  > is minimal.
 >  >
 >  >
 >  > --- Tesla list <tesla-at-pupman-dot-com> wrote:
 >  >  > Original poster: "Mccauley, Daniel H" <daniel.h.mccauley-at-lmco-dot-com>
 >  >  >
 >  >  >
 >  >  > IGBT's are still slow devices.  Also, at 100kHz, you would typically
 >  >  > only run this device at about 5-8A current.
 >  >  > Even so, even though I may build a coil to run at 100kHz (which is a
 >  >  > relatively big coil for a SSTC), others might
 >  >  > want to run at 300kHz etc...
 >  >  >
 >  >  > I'm no longer building the 15kW boards due to lack of orders and
 > minimum
 >  >  > quantities etc..., but am about completed
 >  >  > the redesign of the PlasmaSonic II SSTC which is designed for up to
 >  >  > 10kW.
 >  >  >
 >  >  > Dan
 >  >  >
 >  >  >
 >  >  >
 >  >  >  > Hi Dan,
 >  >  >  >
 >  >  >  > Why are you going to use IXFN44N50 MOSFETs? They cost $17.42
 >  >  >  > each, while a
 >  >  >  > IRG4PF50W IGBT costs
 >  >  >  > $4.43 from the same place. The IGBT is rated for 900 volts
 >  >  >  > instead of 500,
 >  >  >  > 51 amps instead of 44,
 >  >  >  > and also has less gate capacitance.
 >  >  >  >
 >  >  >  > They do take about 100ns (40%)longer to switch, but if you're
 >  >  >  > going to make
 >  >  >  > a big coil the
 >  >  >  > frequency should be down anyway, right?
 >  >  >
 >  >  >
 >  >
 >  >
 >  > =====
 >  > Jimmy
 >  >
 >  >