Gate Voltage / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

Gate Voltage

previous 1 2

Move Thread

LAN_403

Graham Armitage

Mon Jan 12 2015, 04:33PM

Registered Member #6038 Joined: Mon Aug 06 2012, 11:31AM
Location: Salado, TX
Posts: 248

I agree. That's why I am hesitating on going beyond test phase. What I keep seeing is that the voltage on G1 looks much better. Without the bridge connected both legs of the GDT have perfect wave-form. I am wondering if something is going on with the second IGBT in the bridge. Tonight I might try and switch the two gate connections from the GDT and see if that changes anything. Both gate leads from GDT are 3" long and tightly twisted.

Steve Ward

Mon Jan 12 2015, 07:50PM

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

Is this just a half-bridge drive (not full H-bridge)? If so, its often necessary to "manually" discharge the primary capacitor with a bleeder resistors. My suggestion would be to try a pair of 20K, 10W power resistors placed from the C1 to C2E1 terminal and from the C2E1 to the E2 terminal. This should look like a resistor in parallel with each IGBT and the effect is that the MMC will be charged to 1/2Vbus between firings. This should happen naturally but often doesn't, and should the primary C have significant charge at start up, there can be a "DC" component that messes with the duty cycle of the driver (which is what i think im seeing in your scope shot).

If you look after many more cycles, does the Vce waveform look more symmetrical?

Differences in ringing on rising/falling edge can also be a matter of perspective. Assuming an isolated power source, you can alternately scope the Vce of the top switch by connecting scope ground to the +DC rail and "invert" the scope channel. The result is often that the turn-on looks cleaner across the measured switch, than the turn-off (assuming you switch before the current reverses).

Side note: You cannot easily measure Vge of the top IGBT, however. I should also say that putting scope ground on the switch-node is not a good idea in general when switching significant voltage. Special isolation type scope probes would be useful for measuring signals referenced to switched voltages.

Graham Armitage

Mon Jan 12 2015, 08:30PM

Registered Member #6038 Joined: Mon Aug 06 2012, 11:31AM
Location: Salado, TX
Posts: 248

This is an H-Bridge. Single brick with two transistors. Your point about the resistors is a great suggestion. I have 2 10W (6k) resistors across the bus capacitor at the IGBT. But that means they are not connected to C1/E2 - that could be significant. I will try fix that tonight. maybe go with 20k as you suggest before I run it at power.

I will check again, but I don't believe the waveform changed after multiple cycles. I can see how measuring Vge on the top gate is problematic. The idea of "referencing" on the +DC and inverting is clever. I will try that as well and see if it looks any different. Will post my findings later tonight. Thanks for all the suggestions

Antonio

Mon Jan 12 2015, 11:41PM

Registered Member #834 Joined: Tue Jun 12 2007, 10:57PM
Location: Brazil
Posts: 644

Do you have a capacitor in series with the primary winding of the GDT? This may cause asymmetries. Look if the waveform stabilizes after some cycles of the burst. And look carefully at where are the ground connections of the oscilloscope. You can, in principle, ground the oscilloscope at the negative rail (if it is insulated from the mains ground) and look at the outputs of the bridge and at the gates of the lower transistors. Asymmetries would appear also at the primary side of the GDT, where it's safer to observe them with a normal ground.

Graham Armitage

Tue Jan 13 2015, 02:43AM

Registered Member #6038 Joined: Mon Aug 06 2012, 11:31AM
Location: Salado, TX
Posts: 248

Making progress!! Tried checking the G1 waveform by inverting the signal and got the same weird behavior. Next I tried adding 6k resistors across the CE of the IGBTs - still the same. What was strange was as i changed the pulse width or frequency the phase would shift and it was difficult to get a proper measurement. So I tried adding a 10uF tantalum cap in series with the one output to the GDT - voila !! Immediately my GDT started singing like a canary (what I am used to hearing) and the bus current draw increased. Seemed to be getting much higher primary currents too. image below shows the Vge (smaller wave) and the bridge output with minimal ringing. There was quite a bit of ringing but as I tuned the inductor it disappeared.

I have not heard or read of the UD2.5 needing the 10uF cap, but I had it on a previous driver circuit. The UD2.5 has 2 1uF caps - would it make sense to remove those and replace it with the 10uF? This is my first UD2.5 build.

Graham Armitage

Wed Jan 14 2015, 12:34AM

Registered Member #6038 Joined: Mon Aug 06 2012, 11:31AM
Location: Salado, TX
Posts: 248

Well I figured out what was causing the problem on the falling edge on the gate voltage... initially thought it was adding a capacitor to the gate drive output, but it didn't matter what value I used. Then I realized I was switching the wires when adding the capacitor. So I removed the cap and switched the wires - everything works. Now I have NO idea why that would be, but I am leaving it as is for now. Maybe someone smarter than me has some clue.

When playing with the tuning I did see some other strange behavior. Increasing inductance would reduce ringing at the beginning of the oscillations but increases it toward the end. This 12s video

shows what I mean. Is this a symptom of something else? Could it be something the scope is picking up and not really an issue. I have low confidence in this scope right now, but have nothing else to set this ZCS.

The current on the primary seems to be behaving nicely

loneoceans

Wed Jan 14 2015, 12:42AM

Registered Member #4098 Joined: Fri Sept 16 2011, 09:26PM
Location:
Posts: 236

That's expected behavior. My page on the UD2.7 describes a general tuning procedure which will be useful for this. Your scope is fine by the way - I actually prefer analog scopes for catching spikes like these.

Graham Armitage

Wed Jan 14 2015, 01:17AM

Registered Member #6038 Joined: Mon Aug 06 2012, 11:31AM
Location: Salado, TX
Posts: 248

I totally missed that when looking at your site. This makes me feel much better - I see exactly what you show. Makes total sense to reduce the ringing at the end of the oscillations when current is highest. Glad my scope is not lying but I still think I need a nice new shiny digital storage scope

Thanks again to everyone for help on this. Hopefully post some photos of it in operation soon.

Graham Armitage

Wed Jan 14 2015, 02:30AM

Registered Member #6038 Joined: Mon Aug 06 2012, 11:31AM
Location: Salado, TX
Posts: 248

One last observation for the evening. The tuning seems to be fine and ringing eliminated at the end of the cycle. This is the bridge output and primary current with the bus voltage at around 100v. I would have expected the bridge output waveform to be a little more square and less spikey?

previous 1 2

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