[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Terry's DRSSTC - The Bridge...



Original poster: Terry Fritz <teslalist@xxxxxxxxxxxxxxxxxxxxxxx>

Hi Dave,

At 09:24 PM 12/10/2004, you wrote:
Hi Terry

..........
 The TLP250 Is a 2500 Volt gallium arsinide high
> speed opto isolated thing with like 1pF of transfer capacitance...  It is
> powered by a simple "LED"...  Perfect for gate drive signal goofiness to
> follow later...  But be sure to switch those simple LEDs "fast" and no
> cross conduction ;-))

Some other things I learned about these devices I would like to share with you and list...

1. If you use a "speed up" RC network in parallel with normal 10mA dropping resistor on LED side of otpocoupler will _significantly_ speed up operation. I am getting turn on times of 150ns and turn off times of 350ns with IRL3803 (110A 30V FET's) with a FPGA controlled high speed load bank at work so you CAN drive these guys fast (at least to 100kHz). R should be sized for ~10X normal LED current (i.e. 100mA), C of 4.7nF. With 12V in I was using values of
~91 ohms, with 4.7nF results in a TC of ~470nS, but rise time on LED current is
"fairly fast" [ like <20ns to 10mA, <50nS to Ipeak of 80mA].

Great idea!!!!


2. Make sure they (the LED signal side) is either ON or OFF. If you let the LED float, the high gan amplifier on the IGBT/FET drive side will oscilllate, the output drive amplifier will go into thermal run-away and WILL fail. Guess how I know about this... :^? And this occurred with proper bypassing and input voltage of +/-12VDC.

Ok. The LED inputs can't really be left open anyway since we have 250kV and say 200 amps of RF near ;-)) If the 200+ joules in the buss caps dumps into one these little TO-247 IGBTs, it will go off like a rifle shot!! Probably needs to be in a Lexan shield still like our old rotery gaps ;-)) A game between dead times for the current transfer and getting close to the crossover to cut down switching loss... If the IGBTs switch off "too fast" the Miller shoot through will kill them too. They are run at 1/3 the voltage rating which will help that.



3. A totem pole drive (like a GDT driver driving the OC) will also speed up response significantly, especially with turn off since you are hard reverse biasing the OC LED, which generally has reverse recovery times of <<100ns.

I will probably just use high current IGBT drivers like Dan's UCC37322P drivers and Schmidt trigger everything. Their very low impedance drive will fight noise really well like when a streamer hits the LED drive wires :o)))


BTW - Here is what the AC wave form looks like going into the thing:

http://hot-streamer.com/TeslaCoils/MyCoils/DRSSTC/DRSSTCinputWaves.gif

Just the 120VAC with ~~40 amp pulses charging the buss caps. It would be cool if some triac or other simple circuit could control the voltage here and eliminate the variac. I think you could control off either voltage or limit the current pulses. Maybe it could even do the "soft start" for the buss caps... Although this is the last of my concerns at the moment, it will certainly be something to look into once the thing is working.! Getting rid of the variac is one thing almost begging to be done!!

Cheers,

        Terry




-----------------------------------------------

> http://rocky.digikey.com/WebLib/Toshiba/Web%20Data/TLP250.pdf
>
SNIPPO
>
> Cheers,
>
>       Terry
>
------------------------

Contact me off list if you need more "poop"...  :^D

Best Regards

Dave Sharpe, TCBOR/HEAS
Chesterfield, VA. USA
>