GDT testing on table with IGBT at 225 Khz, please check / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

GDT testing on table with IGBT at 225 Khz, please check

1 2 next

Move Thread

LAN_403

serge petiot

Thu Aug 18 2011, 01:11PM

Registered Member #4054 Joined: Sun Aug 14 2011, 01:39PM
Location: The boonies, France
Posts: 26

Hello everyone,
As I mentioned in my other thread, my TC secondary being 4" by 22 " 1150 turns of AWG 26 should run at about 230ish Khz with a 12 by 3 toroid. So, having made the GDT and while still designing the board, I wanted to test the GDT to asses whather it's ok as well as startplaying with a single IRG4PC50UD IGBT. Of course, the prototype being what it is:
Setup

There are stray inductance and capacitance everywhere! The driver is Steve Ward's supplied by a Frequency generator @ 225Khz, and a single GDT secondary is connected to the 30 V supplied IGBT loaded with 10 Ohms to GND.
The waveform supplied to the GDT primary is very clean. The GDT output waveform, with no IGBT connected is this (10V/div):
GDT OUT NOLOAD

Now, with the IGBT connected, the GDT is now loaded with the gate capacitance (no IGBT 30 V power supplied):
GDT OUT LOADED

please notice the C charge time on the rise and the C discharge on the fall. Would this work? How does this look to you experienced hands?
Then with IGBT power supplied, it's switching:
IGBT TEST

I know a resistive load is not representative, but I'm a bit worried about the overshoot on IGBT switch off. I did tune the gate resistor and the GDT is a 10 turns Cat 5 cable classic. I also have a basic snubbing 1uF Cap across the IGBT supply which allows rise time and fall times of 200 nSec, give or take.
Feedback and remarks appreciated...

Serge

axelro

Thu Aug 18 2011, 05:51PM

Registered Member #3640 Joined: Sat Jan 22 2011, 12:16PM
Location: Germany close to Heidelberg
Posts: 39

This looks pretty ugly. I think you basically won't get reliable results this way.
If I can move myself, I'll make some shots of my gates - I am also using kind of a Ward driver, on prototype board.

axelro

Thu Aug 18 2011, 06:09PM

Registered Member #3640 Joined: Sat Jan 22 2011, 12:16PM
Location: Germany close to Heidelberg
Posts: 39

Sorry for the double post, seems I can't update an existing post with a picture.

Here's the picture of my layout + Gate image, 187kHz.

serge petiot

Thu Aug 18 2011, 06:51PM

Registered Member #4054 Joined: Sun Aug 14 2011, 01:39PM
Location: The boonies, France
Posts: 26

Hi Axelro,
your setup is just about ready to use! "small signal" electronics on one board, IGBT driver MOS on another, and your Half bridge is complete... Soo tidy! This is the reason why your countrymen are so good engineers!
Mine was just to validate the GDT I made as the boards are on order. Speaking of which, why have you got 2 ferrites on what appears to be your GDT? Also, with such a low scope timescale, my gate signal looks very similar to yours. What IGBTs were you using when you recorded that signal? What load; gate resistor value, and IGBT supply voltage?
I am aware the prototype wiring is going to cause all sorts of stray capacitance/inductance, but I'd like to know whether that GDT is going to do the trick...
Thx for the pictures, it's great!

PS: scratch the GDT ferrite question, that was just the terminal off the main cap playing games with my eyes!
Do you know what GDT ferrite model you're using (looks like 10 turn tri-filar winding wise)?

Serge

Th3_uN1Qu3

Thu Aug 18 2011, 08:08PM

Registered Member #2614 Joined: Sat Jan 09 2010, 08:57AM
Location: Bucharest, Romania
Posts: 26

That waveform isn't pretty but it'll do for a test run. Could use a PCB tho... What are you using to drive the GDT? Schematic please.

Arcstarter

Thu Aug 18 2011, 08:18PM

Registered Member #1225 Joined: Sat Jan 12 2008, 01:24AM
Location: Beaumont, Texas, USA
Posts: 2253

I do have a few bits of advice to get a better waveform on the gate. First of all, what are those power resistors for? If they are in series with the GDT, or the gate resistor, get rid of them! They are wirewound and have quite some inductance. Breadboards are the worst. I can almost guarantee it is adding ringing. Build it on a protoboard or etch a PCB (or have one etched from your design for you).

Another thing is that the wires coming from the GDT leading to the IGBT are long, and far apart, which creates a lot of inductance. Shorten these wires, and twist them together (or just use the wires from the CAT-5 to connect straight to the gate) to decrease inductance. What kind of DC blocking cap are you using? This could also be a factor.

axelro

Thu Aug 18 2011, 09:32PM

Registered Member #3640 Joined: Sat Jan 22 2011, 12:16PM
Location: Germany close to Heidelberg
Posts: 39

Hallo Serge,
thank you for the nice words, but actually the setup is pretty dirty compared with some of the experts here on the forum.

Yes, the GDT is a 6400AL core with 3 pairs of ethernet wire strings wound together. Got it for 20ct. I injected a signal from a generator at 187 khz or so, but it looks equally good up to 400khz. The gate resistor is 10 Ohm - with that the raise time is approx 100ns. Lowering them to 5 Ohm would make it faster, but there would be more overshoot.

The IGBT's are: HGTG20N60B3D. They were cheap - just 1.50EUR per piece, if I recall right. They are a little slow for the purpose, but it works.

The coil is not on the foto - the setup actually works. With this half bridge setup I get about 30cm sparks. Not great, but probably not yet perfectly tuned either yet. With a full bridge it would probably be significantly better. With a predictor circuit like Finn's or Steve's UD2 it would probably be a lot better.

PS: As you do a parallel setup: If you wanted to do something sophisticated, which I haven't seen yet, you could combine an extremly fast, weak IGBT with a slow, strong IGBT in a parallel setup. Then switch the slow IGBT off earlier by the tail time - the fast, weak IGBT would then carry the load for just the tail time, and due to it's shorter tail, overall the results should be better. Understand how I mean it?

PSS: Saw your web site - impressive. Sounds like you live in a good skiing area as well

serge petiot

Fri Aug 19 2011, 06:06AM

Registered Member #4054 Joined: Sun Aug 14 2011, 01:39PM
Location: The boonies, France
Posts: 26

Hello Th3 uN1Qu3, arcstarter;
this setup is just a very rough setup to validate the GDT ferrite of unknown origin coming from a battery charger... The driver basically is Steve Ward's "Universal DRSSTC controller" (available on his website) with a squarewave injected into it for testing. As mentioned, only one of the 4 GDT secondary drives a single IGBT loaded to GND with 10 Ohms. These big power resistors you see are just the IGBT load! There is no DC blocking capacitor, just straight resistors...

Axelro,
Do you know how much current your GDT is taking to drive your half bridge (if you've measured it)? My GDT is taking about 0.2 Amps per secondary used with a 10 Ohm gate resistor @ 200Khz.
Hmm, your suggestion about // a big slow IGBT and a fast one is interresting and to be tried, but I think you still need very similar pairs of IGBT for the paralleling to work. In the meantime, I think I'll try tidying up the wiring and test an inductive load on the IGBT.

PS. Thanks for browsing my website. There has not been too much updating to it this summer due to the crap weather over Europe... Oh yes, for skiing here, the possibilities are endless!

serge petiot

Fri Aug 19 2011, 10:44AM

Registered Member #4054 Joined: Sun Aug 14 2011, 01:39PM
Location: The boonies, France
Posts: 26

Hello.
I tidyed up the wiring a bit, and now the Gate voltage looks alot prettier:
GDT OUT LOADED1

I reckon the GDT is definitely good for service!

Serge

axelro

Fri Aug 19 2011, 06:25PM

Registered Member #3640 Joined: Sat Jan 22 2011, 12:16PM
Location: Germany close to Heidelberg
Posts: 39

Hallo Serge,
yes, this looks absolutely fine now. Question is, what happens, when you have 4 IGBT's to drive. It will not ring, but the flanks will be less steep. OK, then you can still reduce the gate resistor value down.

As for the current the driver takes - that would be difficult to compare, as I am using an interrupter, so the driver is actually off the most time. The (average) current would depend on the duty cycle. The weak driver IC's I use would probably overheat, if I ran the GDT driver non-interrupted and the 10 Ohm/0.25Watt Gate resistors would start smoking.

Some time back, when I ran an SSTC continuously, I had 10 x 100 Ohm resistors in parallel in front of the gate, and they still got pretty hot (actually that was at 500khz, where there are twice as many switchings then at about 200khz).

1 2 next

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