CWDRSSTC / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

CWDRSSTC

first previous 4 5 6 7 next

Move Thread

LAN_403

Goodchild

Wed Nov 16 2011, 07:19AM

Registered Member #2292 Joined: Fri Aug 14 2009, 05:33PM
Location: The Wild West AKA Arizona
Posts: 795

These suckers are fairly nifty APT45GP120B fast and high current rating.

Although there lack of a co-pack diode is incredibly annoying...

Marko

Wed Nov 16 2011, 01:00PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Goodchild wrote ...

These suckers are fairly nifty APT45GP120B fast and high current rating.

Although there lack of a co-pack diode is incredibly annoying...

Woo, nice. I realized though that if I end up wanting to use external diodes and yet use sigma-delta QCW, the little 18A ones might not be enough! :|

Kind of hard to find a diode I can easily incorporate into my existing bridge design as I don't have any more space on the heatsinks...

Marko

Dr. Dark Current

Wed Nov 16 2011, 01:40PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

Marko,
The school is eating up some of my free time, so the progress is somewhat slower. I have the control circuit and bridge lashed up, just need to wind a primary with taps and I can start testing the coil.
However something a bit "stinky" turned out about the IGBTs... I fired up the controller at 80 kHz (no bridge power), and the two TO220 gate drivers got very hot running the four IGBT gates! There is a figure of 3200 pf gate capacitance in the datasheet, but the real energy-related capacitance is probably around 4 times that.

Hard switched design is out of question, even though the swithing times look pretty good on paper, the switching losses don't and I'm pretty sure I would be burning huge amounts of power on switching. And add to that the huge gate capacitances and resulting slower rise/fall times...

I can't see any theoretical problem with this primary driving method, we'll se how it works in real life.

Marko

Wed Nov 16 2011, 01:50PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Hi kilovolt,

I'm actually pretty sure the IGBT's will work in hard switched too because I have ran 10kW continuous at over 100kHz so far that way without problems. Still running resonant will make it all far easier on devices and is not a big problem to do once you get the tank caps.

I initially used UCC's in my design with mosfets, and alter just swapped for 30N60 igbt's and they kept working... I don't know how hot they actually get but they seem to survive, some TO220 gate drivers are internally biased so that they waste a lot of power even when not driving anything and that's actually the main source of heating - put them onto a big heatsink and they should be fine. Also make sure your GDT is not saturating, you'll need quite a lot of turns (20..30) oon it to prevent it from doing so. UC3710's especially sucked with this self heating as I found out.

If you don't use overcurrent protection and keep insisting on HC14 feedback though, I think there are big chances your design will blow up anyway. SO better watch for the developments in this thread! (Heck, you have enough info already!)

Marko

Dr. Dark Current

Wed Nov 16 2011, 03:11PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

Marko,
the GDT is not saturating, I checked by disconnecting the gates and the input power dropped 6 times.

It seems to me that the 1200v IGBTs are kinda "sluggish", even though they claim fast switching times in the datasheet. The total switching loss at 600V Vce and 40 amp current, is 3.4 mJ. If the bridge did run filtered CW at 120 kHz, each IGBT would be burning 408 Watts from switching losses! Running halfwave unfiltered power, the value is 2.82 times less (assuming constant current sink) or 4 times less (for resistive load), but still this is too much. The tuning and/or feedback with hard switched TC is also more problematic.

The gate drivers are TC4422/4421 and they don't heat up without GDT connected.
I just checked by replacing the gates with 5 nF caps, and the input power dropped to half! So this means that the effective input capacitance of the devices is in the order of 10 nF, or maybe even a bit more.

Marko

Wed Nov 16 2011, 04:52PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Hi kilovolt,

I had the same grim predictions for 30N60's, but they outlived all expectations, I guess partly because of they really switch only the magnetizing current and not the full peak value, as well as that my heatsinking was quite well done with very low case-heatsink thermal resistance.

The predictions of very conduction losses of mosfets turned out far more grim actually, at those 60 amps the igbt's have been withstanding they would dissipate over 700W of conduction losses alone, while for igbt's this was less than 100W or so!

I don't know about the 1200V igbt's, they may be slower after all and if I was to build a 4kW coil I'd use 30N60's because I know they would work (and buck the voltage a bit with a buck transformer). So in reality no matter how bad the igbt's might be the mosfets are actually far worse.

But I guess implementing a resonant cap is always a good idea and would be neat to compare losses with one and another setup. If anything you could run at lower power temporarly with hard switching to set everything up before your tank caps arrive.

BTW, have you ordered your tank caps and where from?

Marko

Dr. Dark Current

Wed Nov 16 2011, 06:24PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

Marko,

I need to use a two-phase power because the standard 230 Vac sockets here only go up to 16 amps, and I would need around 25 amps rms. Our three phase socket has no neutral, just protective earth so I must draw power from between the phases.

So far my tank cap is made from Arcotronics MMKP caps, .082 uf at 1600 volts, they are 8 in series and 2 parallel strings. I have no idea if they're going to survive the full power. If they won't take the power, I'll need to start looking for other caps. I have 50 pcs of .015 2kV MKP caps, maybe something can be made out of these.

Dr. Dark Current

Thu Nov 17 2011, 02:31PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

The feedback circuit I designed was just tested and works perfectly. The circuit tunes to resonance at maybe 1-2 volts (!!) on the bridge, and the ZCS is almost perfect. So far I made some little sparks with it (~20 cm) and nothing blew up, even with the driver not shielded.

Marko

Thu Nov 17 2011, 04:54PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Dr. Kilovolt wrote ...

The feedback circuit I designed was just tested and works perfectly. The circuit tunes to resonance at maybe 1-2 volts (!!) on the bridge, and the ZCS is almost perfect. So far I made some little sparks with it (~20 cm) and nothing blew up, even with the driver not shielded.

Hi kilovolt,

nice work, but please do add overcurrent protection (just an LM311 and a NE555) no matter how unnecessary you think it might be.

I assume this run is already a tuned one with primary feedback? Please do some pics in any case

Marko

Dr. Dark Current

Thu Nov 17 2011, 05:29PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

Marko,

yes the run is with the resonant primary, otherwise the feedback would not work.

So far I went up to some 30 cm sparks, with the bridge supply input peaking at 230 volts and primary peak current 50 amps. The bridge peak voltage at full power will be around 560 volts. The tank circuit is close to a constant current sink, with increasing bridge voltage the current first quickly rises then the rate of rise slows down greatly. The caps so far hold up fine and don't heat up.

For the sake of keeping things simple, I omitted the over current protection. The gate drivers don't have enable pins and the only gate drivers with enable I have are in DIP packages. I have used 4 of the 16 IGBTs, so if they pop its not a big deal, but of course its better not to pop them.

I can post pictures if you want, but I warn you that my room is a real mess...

Edit: Some data which may be of interest

first previous 4 5 6 7 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob