DRSSTC waveform troubles / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

DRSSTC waveform troubles

1 2 3 next

Move Thread

LAN_403

DoS

Sun Jan 23 2011, 09:05PM

Registered Member #360 Joined: Sun Apr 02 2006, 12:13AM
Location:
Posts: 12

Hi,

I'm building a small DRSSTC and ran into some troubles.
The board is Steve Wardish, but with onboard Atmega for interrupter and two TC4429s instead of discrete H-bridge output.
It's half bridge IRG4PC40UD, 75nF primary cap, 3in secondary, Fres about 270kHz.
Primary feedback is two 1:17 transformers in series.

The thing is it kind of works. I have so far tested it with 60VDC on the bridge and it generates about 4in sparks (think it was about 350us on time and 10BPS). But the waveforms look really ugly and noisy. On the scope picture, the green line is primary current (two 1:33 transformers, 10ohm resistor), magenta is lower half bridge IGBT gate and blue is noise (same on both 12V and 5V).

First of all I noticed that the 12V supply dropped about 1V during burst and added more capacitors (2200uF on 12V + 2x10uF tantalums and 0.1uF ceramics on TC4429s), but now it's ringing like seen on the picture.

I've also attached my board layout, maybe it's some kind of design fault?

Gregory

Sun Jan 23 2011, 09:34PM

Registered Member #2922 Joined: Sun Jun 13 2010, 12:08AM
Location:
Posts: 226

Some noise is normal.. tesla coils put out a lot of RF waves.. Try put your circui in a metal box, like a faraday protection

DoS

Sun Jan 23 2011, 09:37PM

Registered Member #360 Joined: Sun Apr 02 2006, 12:13AM
Location:
Posts: 12

And what about the fact that it does not seem to switch the bridge during zero current?

axelro

Sun Jan 23 2011, 09:42PM

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

I don't think it's the board layout. Also the bypass values for the TC4429 should do, if they are close enough to the IC.

Some ideas:
- Have a look at Steve Ward's circuit - he has a 10 Ohm resistor accross the 10uF to the GDT.
- Check for the GDT's leckage inductance - esp. as you are uplifting the output voltage.
- Do you have long wires to the GDT? As short as possible and twisted they should be. If I remember right, a coax to the GDT input might be beneficial, if it's a longer wire.
- What gate resistors do you have? Try 10 Ohm between the secondary of the GDT and the gates. Slows it down, but dampens parasitic effects.
- Last - is your probe connected to ground as close as possible to the source of the lower IGBT? Otherwise you might have measuring errors.

DoS

Sun Jan 23 2011, 09:56PM

Registered Member #360 Joined: Sun Apr 02 2006, 12:13AM
Location:
Posts: 12

Thanks for the ideas, will check them out tomorrow.
Why is the 10ohm there for?
The GDT is 1:2:2 and wires are twisted (about 2-3in on each side).
Gate resistors are 10ohm and yes, the ground of the scope might have been too far when making that particular measurement.

I found another picture I made today, red line is upper IGBT (measured directly on IGBT between gate and emitter, primary a bit out of tune) and it seems there is less noise there. But the primary waveform still has some ringing during the switching.

And still, why is there a phase difference between zero crossing of the bridge and the primary current? Is this ok or something to be worried about?

axelro

Sun Jan 23 2011, 10:20PM

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

I believe the 10 Ohm is meant to dampen parasitic oscillations between the 10uF and the GDT primary, but this is only a guess. Perhaps Steve Ward shows by and explains it.

The signal already looks pretty good to me, now. The switching noise could be, because a) your frequency is pretty high and IGBT's are stretched a bit and b) the switching doesn't occur exactly at the zero current point. If it switched exactly at the zero current point, the conduction caused by Toff and Tail off time (or however that's properly said) would be minimized.
Have a look at Finn Hammer's predictor thread (see Home page featured projects) - there this is pretty well explained.

The phase difference is caused due to the fact, that each circuit has a propagation delay - so until the signal is processed and gives it commands to the power stage, it's a bit out of phase (delayed) already. From the experts I understand, that it's better, they switch on a little after the zero current point rather then switching off before the zero current point (of primary current) is reached.

Hope I got that all right.

Goodchild

Mon Jan 24 2011, 04:51AM

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

If you look at your green primary current trace you will note some noise near the peaks of the sine. If you compare this with your drive signal it's clear that you are not switching at the zero crossing but rather almost when the current is at it's peak. That spiky noise stuff is most likely from your IGBTs trying to interrupt the current when it's at it's peak.

I think a couple other people mentioned this on this thread but some possible cusses of this are:
A) you are tying to switch your IGBTs faster than they are rated for.
B) you have some extra delay in your driver board that is making you switch to late, long after the zero crossing.

I think the most probable explanation is B, because if you look at your gate drive waveform there is a delay between when it turns on and when you see the spiky junk in your primary current. This delay is the Td Tr of your IGBTs (or any switch for that matter).

If you really want to be sure put a probe on the output of your half bridge and compare that to your CT output and primary current. This will tell you if the delay is coming from your board or your bridge.

hope this helps

DoS

Mon Jan 24 2011, 02:56PM

Registered Member #360 Joined: Sun Apr 02 2006, 12:13AM
Location:
Posts: 12

Well the delay between primary current and half bridge output is about 500-600ns.
Why is the output like it is (not going up and down straight, but has something like steps)? Same can be seen on previous gate voltage waveforms.

Adding 10ohm resistor parallel to the GDT primary DC blocking cap made the waveforms look a bit nicer, but they're still not how I'd like them to be. I'm still not sure if there are two separate issues (delay and steps on gate waveforms) or are they connected (delay causes ugly gate drive)?

EDIT:
Sure enough they're connected.
I added external function generator to the feedback instead of the current transformer and got perfect switching and everything looks ok. So now I should look for the cause of the delay (maybe that kind of delay is normal for this circuit?) or build a variant of the prediktor which seems to be the new hot word around here :)

Goodchild

Mon Jan 24 2011, 05:28PM

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

If you want post your schematic and maybe someone can pick out what is casing your delay.

DoS

Mon Jan 24 2011, 05:37PM

Registered Member #360 Joined: Sun Apr 02 2006, 12:13AM
Location:
Posts: 12

Here it is, as I said basically the same as Steve Wards.

Some relevant component values that are missing:
D10 and D11 1N5819
D5 and D6 BZG05C4V7TR3
R5 1K
D7 and D8 1N4148

1 2 3 next

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