Gate drive transformer output droop ... / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

Gate drive transformer output droop ...

Move Thread

LAN_403

Dylan

Fri Sept 26 2008, 05:33AM

Registered Member #1024 Joined: Sun Sept 23 2007, 10:56AM
Location: Northern NSW, Australia
Posts: 95

Just testing the new driver and have hooked up the GDT to do the phasing.
The GDT is driven by a normal H-bridge , as per Steves new driver, capacitively coupled with 50uF electro and 470nF film caps.
The output looks ok :
2889463354 Fef79bb7e5

And judging from the low current draw with no load, cross conduction in the bridge is under controll.
But , over the duration of one burst the voltage swing droops down from +/- 2x the rail voltage to +/- 1x the rail volatge, ie it starts at 36v p-p , but droops to 18v p-p :
2889463764 509ff46b82

I'm a little hung over , and the brains not with me today, but I dont get it.
I've not noticed this on my past drivers, but maybee I just havn't noticed!
Does this look normal to anyone ?
Whats going on ?
Whats my name ?
Where am I?

================== UPDATE =====================

Ok, turns out the SUPPLY RAIL is actually drooping, traces below are :
-top : supply rail
-bottom : gate voltage from random bridge fet
2889504056 794f689468

Wierd , seems the three terminal voltage regulator is just not supplying the required current ( <300mA ! ) , it is however happily regulating the voltage under no load conditions.

ragnar

Fri Sept 26 2008, 07:07AM

Registered Member #63 Joined: Thu Feb 09 2006, 06:18AM
Location:
Posts: 1425

Take those electrolytic capacitors out of your primary circuit and put them on the power pins of your regulator/driver chips instead =-P

Dylan

Fri Sept 26 2008, 07:14AM

Registered Member #1024 Joined: Sun Sept 23 2007, 10:56AM
Location: Northern NSW, Australia
Posts: 95

, Indeed!

But seriously, there is loads of filter caps on the rails and drivers.
I've looked a bit closer , and it appears I do have a cross conduction problem in the bridge.
I'm going to try lowering the bridge gate drive voltage and see if that helps, failing that I'll start cutting gate tracks and inserting zeners and resistors and other mumbo-jumbo on the bottom halves (N-ch) of the bridge.

Teach me for not using matching P and N-CH devices in the bridge!
I'm using what I had lying around ( IRF9Z24NS and RF1S25N06 ) there close-ish but certainly not 'matched'.
Fun and games !

===== UPDATE =====

Well , dropping the bridge gate drive voltage to 8v (as per Steves new design) did the trick:
2889815262 B70e723b6f

Eevrything running cool now , no droop

.
I should have listened to Steves notes in the first place!

Steve Ward

Sat Sept 27 2008, 04:57AM

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

Ive had this P/N driver work OK with fairly different P and N fets. I found that the drive voltage for them was fairly sensitive, and that lower drive voltage (lower than the 9V i suggest) can sometimes help avoid cross-conduction even better. I wouldnt go less than 5V for that gate drive.

For what its worth, i dont really see any drooping with multi-mS long pulses in the 100's of khz range on the gate drive supply when using the mosfets i have spec'd out, and im only using 660uF total filter cap (330u before and 330u after the regulator).

Dylan

Mon Sept 29 2008, 09:20AM

Registered Member #1024 Joined: Sun Sept 23 2007, 10:56AM
Location: Northern NSW, Australia
Posts: 95

Well , I loaded it up today with the final output bridge :):
2898458152 91b7451168

And the waveforms are now all bad again :( , all the following are waveforms at the input side of the GDT.

First, no load (final output bridge not connected), very nice:
2898444674 Ed606c804f

Then with load, not very nice :
2897602339 36c3b1673b

Then referenced to ground this is from the side of the gdt driving bridge with no dc blocking caps:
2897602673 93920b023a

And this abomination is from the side of the GDT with the blocking caps:
2898446050 5fc882980a

So, I have tried upping the DC blocking capacitance to 50uF , no change, I have replaced all fets in the GDT driving bridge, no change.
There is no droop on the supply rails.
This waveform looks like what I was getting when my driver was ending the burst with one driver high and one low, but that is not the case here, this driver ends the burst with both N-ch fets in the bridge ON, and the P-ch off.

Does this look familiar to anyone else ?

========== UPDATE ===========

How very embarasing, just had another look, and I have stuffed up my PCB layout, I'm driving both P-ch fets in the bridge from the same driver! DOUP!

Its tempting to clear this post!
But I will leave it here for reference if anyone else make the same stuffup

Steve Ward

Fri Oct 03 2008, 02:36AM

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

Hah, way to go! :P

I just wanted to mention that if you feed the input with a signal generator (or some other non-synchronous signal) then there can often be a small ripple in the gate envelope due to the mismatch in cycle length of the first half-cycle and the rest. It will basically look like your gate drive wave form is waving up and down at a lower frequency. Typically this will completely disappear when using feedback as normal. I also notice that at really low input voltages (and hence, low primary currents) i can get the thing to oscillate, but the first few cycles of RF are botched up, so it causes the gate waveforms to jump around a bit. Again, this should go away at 15% or less of the intended operating voltage of the inverter. If bad waveforms do persist up to higher voltages, check that your feedback signal looks good at the diode clamp circuit and at the input to the hex-buffer chip.

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