Few DRSSTC questions / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

Few DRSSTC questions

previous 1 2

Move Thread

LAN_403

HV Enthusiast

Sat Apr 22 2006, 12:50AM

Registered Member #15 Joined: Thu Feb 02 2006, 01:11PM
Location:
Posts: 3068

The problem is you have too much leakage inductance. This is what is causing the sinusoidal looking waveform. You need to increase the coupling between windings. Wind them bifillar / trifillar. You said so before, the primary is somewhat separated from the secondary . . .

Marko

Sat Apr 22 2006, 01:33AM

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

Sounds like the IC is latching.

Can you build multiple gate driver stages in parallel?

Very strange, I didn't expect ICL's to be particulary powerful but this really suprises me.
Seems like internal halfbride in output stage of IC latches up with both transistors conducting for some reason, removing all the load doesn't help at all and only way to save the stupid self-destructing IC is to cut the power.

When powered up again it works and drives mosfets like nothing happend.

You can see in first schematic that idea was to drive main GDT directly with bunch of stacked ICL's. (I didnt mention possible stackage in schematic and drew only two for simplicity).

I think ''only'' about 4-5 of them on each side would be enough, but theese latchups totally discouraged me of trying such a thing.

The problem is you have too much leakage inductance. This is what is causing the sinusoidal looking waveform. You need to increase the coupling between windings. Wind them bifillar / trifillar. You said so before, the primary is somewhat separated from the secondary . .

Yes I wound some better GDT's. Primary waveform was relatively good and I realised that I need more coupling.

I used motor to wrap wires together tightly, and gos somewhat better results.
Waveform is still somewhat sinusoidal-ish, but quite better than the first.
I' post a pic of it tomorrow.

Now i'm trying something with mosfet halfbridge, but that requires another GDT and isn't promising.

Terry Fritz

Sun Apr 23 2006, 03:54AM

Registered Member #393 Joined: Tue Apr 18 2006, 12:30AM
Location:
Posts: 297

Hi,

Those stupid ICs can lock up internally if "any" voltage goes out of "spec". You need to insure that the power supply and inputs never go out of the specified range even for a microsecond... If the base silicon wafer gets a bad signal, it will bias the thing into oblivian...

BTW - If you need IC's, we can probably get them to you even in Croatia

)

Cheers,

Terry

Sulaiman

Sun Apr 23 2006, 07:54AM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

I haven't seen this prblem myself,
but I put diodes across the outputs of my GDT drivers
(theoretically not required, just 'insurance' against the unexpected)
Diode 1 Anode-Ground, Cathode-Output
Diode 2 Anode-Output, Cathode-(Driver +Supply)

I use schottky diodes becaue I have them in stock
any fast 1A or more diodes will do.

This arrangement prevents 'lockup' and will help with your problem
although a better GDT sounds in order

(have you accidentally used a low-mu (Iron Powder) core?)
(the cores in common-mode line filters are good)

Marko

Sun Apr 23 2006, 12:08PM

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

With deepest regrets I have to pronounce that on Sunday, 23 April 2006, theese poor ICL's decided to take life of their own.

They killed themselves in really unique way: I disconnected them from the H-bridge of BJTs (used to drive GDT's) trying to do some more tests. I put a scope probe between their outputs, and noticed thet LED on interrupter goes dim when I touch them (with probe (!) and im pretty sure I didn't short anything as I used separate wire and aligator clip)
Whle I realised that probe actually latched up (at least one of them) and cut the power (I didn't hear transformer buzzing this time from overcurrent, as it would usually do on latchups).

Foolishly I didn't have diodes across outputs as I just driven the bases of bipolar transistors.
They lowered output a bit in such circuit so i removed them.

I always put these diodes (4148's) when driving anything else, but they certainly didn't stop the latchups nor affect the circuit in any other way.

I blew 4 ICL's (2 are from before) and just before I wanted to take the ppic of the waveform.

Nothing good, more like sinewave with cut top and huge transistent times (500 - 600ns).

You can also see GDT im using (OK i hope i'm not that stupid to use powdered iron instead of ferrite cores).

I used motor to wrap a bunch of litz wires together for biggest opssible coupling. It works perfectly, when I hook it up to ICL's I get very nice waveform on the output, and it stays unchanged until they latch up. Bipolar transistors work with 2x 20nF loads but make large delays themselves.

I wound a bit more turns than actually needed, maybe I could get some more current if I unwind some

Anyway it just seems that ICL7667 is totally unusable and wrong IC for the job.

HV Enthusiast

Sun Apr 23 2006, 01:29PM

Registered Member #15 Joined: Thu Feb 02 2006, 01:11PM
Location:
Posts: 3068

wrote ...

Anyway it just seems that ICL7667 is totally unusable and wrong IC for the job.

Yep, as i said before, you are going to need a driver that has higher peak current output.

You also try simulating your gate drive design before building it. This will provide powerful insight to your design and allow you to vary all the parameters etc... before buying or building anything.

Marko

Tue Apr 25 2006, 11:18PM

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

I finally made some progress.

This is with single 120nF load, at 20V input (GDT is 1:! and it will be powered with 33V).
It is driven with a push-pull fullbridge (center-tapped primary) with two IRFZ44 mosfets, and with enough coupling it seems to work fine.

rise times are somewhat high, resistor slows them to about 400ns total (note the fast falls because of diode, 200ns, almost the delay of mosfet itself).

When I don't use resistor I get even faster transistents but it rings much more.

Another big problem is ringing after interrupter shuts the whole mess of, as mosfets don't short the primary (like full-bridge of driver chips would do) but just stay high Z and let the rinigng go.

I lowerd it somewhat by 100 ohm resistor across the primary but I think I'l need to shut it off actively using another mosfet that shorts the primary when interrupter signal is low.

This seems best idea until I get some more powerful driver chips.

The last failure of ICL's was caused obivously by shorting the output temporarly to ground, as I left another probe connected and this caused massive latchup and failures.

Now I rebuilt half of the circuit, replaced the 74HC14 and latchups have ceased, I could load ICL's with big caps and they would normally heat up and deform the waveform, but never latch up. I don't know what on breadboard actually caused that, some voltage spikes bad signals (unfortunately I couldn't scope things while they were blowing.

Steve Conner

Wed Apr 26 2006, 12:43PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Hi Firkragg

Most gate driver chips will latch up and self-destruct if you try to drive a low impedance inductive load. (eg, a GDT with a low-mu core or a secondary shorted, or a short circuit with a small enough loop area.)

It's caused by a high current that builds up and then kicks back through the body-drain diodes inside the chip. As with any other CMOS IC, these diodes are part of a parasitic SCR-like structure that will latch on and short the power supply. The UCC3732x are supposed to be protected against this, but only up to about 1A or so, and you can easily exceed that with a short circuit. The UCC can drive up to 9A, and the same 9A will kick back.

I managed to latch up and kill a few "UCC"s until I started putting the Schottky protection diodes on them like Sulaiman recommends, and since then I haven't killed a single one. I imagine the ICL7667s are even more sensitive, so the diodes would help them a lot more. They have to be Schottkys though. Ordinary diodes have too high a volt-drop and won't divert the current away from the onboard diodes.

Desmogod

Thu Apr 27 2006, 01:07AM

Registered Member #139 Joined: Sat Feb 11 2006, 11:01AM
Location: Perth, Western Australia
Posts: 358

Steve Conner wrote ...

I managed to latch up and kill a few "UCC"s until I started putting the Schottky protection diodes on them like Sulaiman recommends, and since then I haven't killed a single one.

Once again, Elmer fixed Bugs' problem before he even asked it.

previous 1 2

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