DRSSTC half bridge switching scoped - interpretation of signal changes / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

DRSSTC half bridge switching scoped - interpretation of signal changes

Move Thread

LAN_403

axelro

Sun Aug 21 2011, 11:12AM

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

I scoped my half bridge (It's raining - that makes one creative how to spend the time). I am particularly interested in the non-obvious switch points in this.
Perhaps one of you is familiar enough with IGBT switching to help interet - I marked the significant points in the pictures with numbers.

First some background:
- DRSSTC, Half Bridge
- Approx 180khz
- About 30V into the bridge at time of measurements
- Scope probe as short as possible at bottom Emitter & Collector of IGBT.
- IGBT: HGTG20N60B3D (relatively slow)
- Overshot and ringing is typically higher at low voltages (relatively) then at higher, but that shouldn't matter so much for this analysis.
- Gate drive with 10 Ohm and parallel diode, to slow switch on. Timings in last attachment.

Attaching 4 pictures:
- One period in total
- Detail switch on
- Detail switch off
- Gate drive wave form with rise/fall times as a reference, what drive was present.

Points 2 and 5 are obvious - switch off and on.

What are points 1,3,4,6 and 7, 8 (7+8 only marked in the detail). The scale in pictures might not be optimal, but I did look at this at higher amplification - there is an obvious signal change at this timings.

Interpretations welcome. Also hints, please, if there are points not marked by me, but also relate to some phase of IGBT switching.

Thank you!

1313925113 3640 FT0 One Period Of Half Bridge Wave Form

1313925113 3640 FT0 Switch Off In Detail

dude_500

Sun Aug 21 2011, 05:28PM

Registered Member #2288 Joined: Wed Aug 12 2009, 10:42PM
Location: Cambridge, MA
Posts: 179

These are all my best guesses, it's hard to know without being able to probe stuff in person. A shot of the bridge on channel 1 with the gate drive on channel 2 would be extremely helpful in confirming all of my suspicions.

1. I'm thinking this may be a deadtime where both transistors are not in a conducting state. It looks like your gate drive signal is quite a bit less than 50% (maybe around 40%), which means there is 20% of the time that no transistor is on (going with the 40% assumption). I'm fairly confident that transition 1 is the gate drive turning off. Now to speculate as to why the voltage jumps up a tiny bit, during this period, the coil primary's back-EMF is being channeled through the recovery diodes. These diodes could have a lower v-drop than the IGBT's forward on-state v-drop.

3. This just looks like the beginning of some general ringing. Any inductance in the bridge's DC wiring will cause ringing since V=L dI/dT and there is huge dI/dT in a DRSSTC bridge. I'm going to go out on a limb and speculate that maybe the little peak right at #3 is caused by passing the miller plateau of one of the transistors, but that may very well not be the case.

4. Same as #1, just the other half of the cycle.

5. This little jump and up from 5 to 6 and delay at 6 definitely looks like a miller-plateau phenomenon to me.

7. Not sure what you're pointing out exactly here, it's just a bit of damped oscillations finishing out in the deadtime.

8. Kind of same as 7, it's just a little more evident that it is ringing in deadtime.

So a lot of the things here are more visible due to your less than 50% duty cycle, not that it's necessarily a bad thing.

axelro

Sun Aug 21 2011, 07:23PM

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

Thank you very much for your efforts!

Actually, the non-50% duty cycle has escaped my attention so far. No clue why this is. I am using Steve Ward's driver circuit. I will need to carefully monitor voltages at higher input voltage into the bridge, to see, if this moves towards 50%, as it probably should.

This non-50% duty cycle is probably connected to another phenomenon - my DRSSTC does not start operating, until I trigger it by shortly feeding a signal from a signal generator. This is primary feedback with 2 CT's in series with each 33 windings - this should actually have plenty of feedback even at low amperages.

Investigating ... edit: no clue - could not replicate, new measures show a 50% gate duty cylce, at a somewhat higher Vin to the bridge. But why should that matter...

teravolt

Mon Aug 22 2011, 03:52AM

Registered Member #195 Joined: Fri Feb 17 2006, 08:27PM
Location: Berkeley, ca.
Posts: 1111

from what I have seen your zero crossing is off or your dead time is to small. Does that board have phase leading for zero cross timeing adjustment.

axelro

Tue Aug 23 2011, 01:24AM

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

No phase lead on this one, it is basically a Steve Ward clone. As mentioned in the edit above, I could not replicate the odd duty cycle during a second try.
There is no explizit dead time feature. I can only guess that Vin was too low to create enough voltage in the feedback CT for a 50:50 signal. I perhaps dialed the Variac too low first time, perhaps 20V in.

serge petiot

Tue Aug 23 2011, 06:54AM

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

Axelro,
the way your voltage look from (1) to (2) looks like a very steady Vbus with no current being drawn. like you are switching at (2) when you should be switching at (1). What would be interresting is to take a measurement of your feedback CT signal, that would allow to comfirm the late switching.
As for those voltage spikes at switching time, your TVS caps are not doing their job. Watch that you do not exceed Vce Max on your IGBT because of the spikes like it happened to Steve Ward... Generally, TVS diodes are too slow to catch it
My guess anyway, but I'm by no means an expert...

Rgds

Serge

axelro

Wed Aug 24 2011, 01:35AM

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

Hallo Serge, yes, i will repeat measurements at the next weekend, as I am currently travelling. Something did not lok right.

Not sure I understand your comment regarding the TVS diodes (you actually said caps?). The diodes would only kick in at 400+V, but I took measures at a very low voltage. They are not slow, actually to my knowledge they react in picoseconds, minus any delay by inductance. The overshoot also typically proportionally gets smaller with higher Vin.

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