1 MHz GDT problems / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

1 MHz GDT problems

first previous 2 3 4 5 next

Move Thread

LAN_403

Dalus

Fri Oct 19 2007, 07:18PM

Registered Member #639 Joined: Wed Apr 11 2007, 09:09PM
Location: The Netherlands, Herkenbosch
Posts: 512

We've got first light, 2.5 inch sparks from 60 V input. Here's the gate waveform

And a small movie of it sparking (without sound)
]1192821467_639_FT32434_dscn7667.zip[/file]

Ken M.

Sat Oct 20 2007, 12:10PM

Registered Member #618 Joined: Sat Mar 31 2007, 04:15AM
Location: Us-Great Lakes
Posts: 628

Is there any other format you could post that vid in or on youtube?

Dalus

Sat Oct 20 2007, 12:45PM

Registered Member #639 Joined: Wed Apr 11 2007, 09:09PM
Location: The Netherlands, Herkenbosch
Posts: 512

Marko

Sat Oct 20 2007, 02:18PM

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

Dalus, that waveform doesn't look good at all. For a full bridge of IRFP450's, just two UCC's should give really nice waveform at that point, and probably even bipolar totem poles would work better than that.

Also, I really don't believe your circuit would dissipate 30W for long, because it would be cooked.
Just think about heat produced by a single 25W bulb.
Heatsink on your driver would need to be bigger than your H bridge heatsink at the time.
Your measurements could have been wrong.

Do you have any pics of entire circuit in current state?

I'd really recommend you to remove that pre-drive bridge and re-investigate everything from the beginning.

If you are using GDT and everything as it is on pics on your site, then you definitely need to fix that.

You absolutely need to reduce leakage inductance of your entire circuit.

First what you need to do is follow these instructions
to calculate the minimum core size and minimum number of turns for your GDT.
Any extra turns (which you probably have like 10x too much) will add leakage inductance.

You need no bigger ferrite core than one that just allows you to fit the minimum amount of turns through.
I would really recommend any GDT for <500kHz to be wound with coax. Those 2-cored coaxes for CD rom audio connections are great. Wind 2 on same core for a fullbridge. Shielded USB cable (4 cores) is great too.

Then, use the shield as the primary and cores for secondaries.

A bit thicker/more rigid cable is better since it minimizes the chance of insulation breakdown.

Make everything physically smaller and closer together; for connections, use back-wired ribbon cable or coax cable for connections.

Uzzors: Datasheets usually state the kind of load and conditions under which they test their transistors, and that is often guite close to operating limit.

Measuring them open will tell you nothing because just reverse transfer capacitance can be enough to spoil your measurement.

You need to give them real capacitive load and measure that.

If there is shoot-through you should see voltage dip/current spike on your supply.

You should have no problems using small signal transistors. Do you understand how to wire the circuit or I should draw a schematic for you?

Marko

Dalus

Sat Oct 20 2007, 05:47PM

Registered Member #639 Joined: Wed Apr 11 2007, 09:09PM
Location: The Netherlands, Herkenbosch
Posts: 512

I'm using the large gdt because all of it's leads are hot-glued to the circuit board of my H-bridge.
One thing about the waveform I'm using a really old 10 MHz scope (from 1960).This scope has quit high rise times (23 ns).The capacitance of the probes is (84 nF). So it's almost impossible to get a reliable wave form @ 1.4 MHz.
But you where right about the power usage my digital meters where influenced by the em field. With analog meters I measured 0.3 A @ 30 V this would be about right.
I'll post the pictures when my camera batteries are charged.

Marko

Sat Oct 20 2007, 06:03PM

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

I don't think your scope has 84nF probe capacitance - rather pF...

Scope may be creating problems, but even still I think you can do it just as good with UCC's.
You really need a different GDT here.

Are you feeding 30V (1:2 GDT?) to IRFP450's? don't do that.

Dalus

Sat Oct 20 2007, 06:26PM

Registered Member #639 Joined: Wed Apr 11 2007, 09:09PM
Location: The Netherlands, Herkenbosch
Posts: 512

I'm not saying it's better to use my approach then using the ucc's , if the hot glue wasn't there I would already be using a smaller gdt.
Here's the measured probe capacitance.

And here's the circuit in it's current state.

Why shouldn't I use a 6:12:12 gdt 30 V?

uzzors2k

Sat Oct 20 2007, 07:27PM

Registered Member #95 Joined: Thu Feb 09 2006, 04:57PM
Location: Norway
Posts: 1308

Dalus wrote ...

Why shouldn't I use a 6:12:12 gdt 30 V?

The gates can only withstand +/- 20V max. Any more and they become damaged. Since you're using a half-bridge, but 30V, a 1:1 GDT will give 15V at the gate which is perfect. Using a 1:1 ratio will improve the gate drive current too, meaning faster rise times.

Marko: I'm working on my site right now, and I haven't done any experimenting with the gate drive this last week. I think I wired the transistors together right, but if you don't mind a schematic would remove all doubt. I didn't have any small fets available so I couldn't use the class A pre-amp, but I think the 4049 can supply enough current. It drives the BD's well enough. As for the delays, do they increase under load? I'll measure the delays with the transistors loaded then.

Steve Ward

Sat Oct 20 2007, 07:30PM

Registered Member #146 Joined: Sun Feb 12 2006, 04:21AM
Location: Austin Tx
Posts: 1055

You cant measure a scopes capacitance in that manner... the 10Mohm input resistance will mess up your simple cap meter. Try this as a test: take a known capacitor, measure it with a huge series resistance (maybe try 1k, then 100k, then 1M). At least all of my cap meters give huge errors from this.

Marko

Sat Oct 20 2007, 07:52PM

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

Steve: I just wanted to say that too, but I measured my probes now connected and disconnected to scope with 1x and 10x divider;

I got about 40pF disconnected from scope and 12pF on the scope. Flipping the variable cap affected it for few pf.

That's a bit confusing since I know that my meter should be affected by the resistance.

But I don't think by any mean that this 80nF is right. This would probably make scope useless for anything above Hz range.

first previous 2 3 4 5 next

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