[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Shottky Diodes in SSTC ---> Not A Necessity
Original poster: "Terry Fritz" <teslalist-at-qwest-dot-net>
Hi Justin,
You may appreciate this scope trace of Ic on an IRG4PF50WD IGBT with
antiparallel diode internal to the TO-247 package
http://hot-streamer-dot-com/temp/OLTC08-30-02.gif
http://hot-streamer-dot-com/temp/irg4pf50wd.pdf
711 amps peak. I turned off the IGBT at 45uS after the waveform
begins. The current drops way down until the anitparallel diode picks up
the waveform again. Once the signal goes back to the off IGBT at 53uS it
stops dead.
This was a repeating signal firing at about 1Hz. The IGBT ran just warm
until the 15 volt gate drive pooped out and the IGBT started to turn off at
the high currents. I could get to about 800 amps with more gate drive, but
800 amps is pretty much the limit for this device. At 800 amps the silicon
physics just starts to run out of steam. Then it starts to warm up, but I
can see that on the scope and stop before it gets too hot. The setup looks
like this. Just discharging a cap (4.7uF 800V) into a single turn copper loop.
http://hot-streamer-dot-com/temp/OLTC08-30-01.jpg
At the ends of the loop, I use transorbs to take to voltage spike when the
IGBT shuts off in mid cycle causing a big voltage spike.
The gate drive signal is shown here in red:
http://hot-streamer-dot-com/temp/OLTC08-30-03.gif
I use TLP250 drivers which do seem to be able to resist shoot through and
provide great drive signals even with super currents and about 800 volts
Vds here (2 amps drive). They also have 2500 volts of opto-coupled isolation!
http://hot-streamer-dot-com/temp/TLP250.pdf
Even though the max speced collector current is just 204 amps, one can
really drive them far beyond that. One may have to push the gate drive
voltage to about 25 volts since Ic does start to limit just from the ~40
transconductance. Note that I limit the IGBT Toff and Ton with about a
50ohm resistor on the gate. You don't want the IGBT turning off too fast
or the gate spikes will get bad and pop the gate (the transorb should stop
that).
I should point out that I stick transorbs everywhere and I have never blown
an IRG4PF50WD that I did not mean to deliberately. In the case of the
OLTC, transorbs take to voltage spikes when I turn the primary off with
energy still stored in teh primary inductor as current. That can be like
5000 amps at 400nH for 5 joules of spike =:-)) The transorbs actually take
that fine. I think because the turn off is rather soft and other losses
take the power too.
Just wanted to mention this encase you want to play with more current
;-)) IGBTs are really fun and pretty cheap. They can handle far higher
currents than a FET with much lower loss. Note the tiny heat sink used in
this 570000VA test ;-)
A thought about protection - In this case, it would be very possible for
the IGBT to blow (I am powering the circuit with a MOT). I really did not
need to have 12 transorbs and three opto-isolators between the $1800 signal
generator and the IGBT after all, but I did!!! ;o)))
Cheers,
Terry
At 03:04 PM 3/14/2003 -0800, you wrote:
>Hi All,
>
>Many of you may remember me posting a message with a similar subject
>line, several weeks ago. To be honest, I am still not happy with the
>general consensus that these shottky diodes are even remotely
>necessary. So, I'm back to bug ya ; )
>
>The shottky diodes I am referring to, are ones placed between the
>MOSFET drain pin, and the ultrafast "antiparallel" diode cathode
>pin/lead in typical SSTC schematics including ones shown on my
>website. These schematics will be revised soon, just haven't had time
>lately.
>
>Four Very Good Reasons to exclude series shottky diodes in your next
>SSTC project:
>
>1. Huge amount of testing by a friend (Aron) and I. I have personally
>tested the no-shottky SSTC design in several completely different
>SSTC's that include a). +2kW IRFP460 H-bridge running with 16 inch
>streamers. b). IRF640 half-bridge, 120V input with only 30V headroom.
>c). IRF840 H-bridge, 10" arcs -at- 340VAC input.
>
>Aron has also tested several different setups, which include a
>IRFP460 half-bridge, making 12" arcs with 340VDC input and a IRF840
>full bridge -at- 11" arcs.
>
>2. Have you ever noticed that you have dead shottky diodes in your
>SSTC circuit, only to realize that everything still works fine? I
>have, many times. If you are using 1N5822's in a powerful circuit,
>chances are, one or more of these diodes are dead and things are
>still working just fine.
>
>3. MOSFET body diodes are EXTREMELY SLOW. An IRFP460 MOSFET body
>diode has a rated reverse recovery time of: TYP=580nS and MAX=1200nS.
>This is like hours in an SSTC circuit. Most ultrafast diodes blow
>body diodes away in Trr (rev. rec. time), hence no need for a series
>shottky to isolate the two diodes. I do agree with Jan Wagner, and
>Ken Herrick, that this diode should have a forward voltage drop
>significantly less than the MOSFET body diode (we are talking forward
>recovery time/response time here, which is similar for both diodes,
>so Vf would be important *breath*). I didn't agree with this before,
>but I do now.
>
>4. I don't see any commercial electronic designs using series
>shottkys. Not in motor control OR in power conversion, the two big
>ones. If someone can find a design using the shottkys, please email
>it to me!!
>
>----
>
>(Refer to #1 above). For all testing on small SSTC's less than 500W,
>Aron and I used "antiparallel" dioes with 50 and 75nS reverse
>recovery times, in DO-201 packaging (same package as 1N5822). Max
>avg. forward current rating was only 3 amps for the smaller SSTC's.
>For our larger ones (0.5 - 2KVA), we used 16A TO-220 packaged ones.
>
>You're not going to believe this, but I can *almost* say that
>back-to-back zeners from gate to source are unecessary too. This is
>pushing it, I know, but I can't ignore my own findings. I have gotten
>16" arcs without any zeners in a IRFP460 full bridge. The gate
>signals were still very clean with no Vspikes past + and - 20V. I
>did, however, use a very powerful driver I designed that has an
>output impedance of less than 100 milliohms.
>
>Anyhow...SSTC's are cool!
>
>Take care,
>
>Justin Hays
>KC5PNP
>Email: justin-at-hvguy-dot-com
>Website: www.hvguy-dot-com
>
>__________________________________________________
>Do you Yahoo!?
>Yahoo! Web Hosting - establish your business online
>http://webhosting.yahoo-dot-com