DRSSTC frustration / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

DRSSTC frustration

first previous 2 3 4 5 next

Move Thread

LAN_403

Marko

Mon Oct 01 2007, 08:38PM

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

About the mysterious arcs on your H-bridge, can you draw the schematic of your power board, draw a lightning bolt on it or something to indicate the two nodes that it's arcing between, and post it here?

Arcs happen between feedback-sensing CT stack hot output (yep, the one that goes into zener clamps and 74HC14 ) and + pole of the bus cap. Arcs are surprisingly small, and it was actually odd that it took so long for something to blow.

Why in world did I put low voltage traces onto the HV board like this??

I could have easily swapped the CT sets, I have no idea why haven't I done at least that. Then it would arc to OCD CT which has only 5 ohms to ground anyway.

The litycs are shunted with center-grounded set of caps to pass any HF to ground. GND of control board is grounded too. That's all that is odd, rest is nearly exact copy of steve's coil.

I can draw a schematic if really needed...

I haven't had time to investigate today, but it seems that failure took out random components on +15V rail. OCD NE555 and one UCC are found failed for now.

I believe this arc is what killed them. The failure happened as I turned off the interrupter; I heard a small pop and flash. (didn't notice where, but I suspect it was on bridge).

I curse this design now. I have so much better ideas to design this thing now. I don't know why I anchored everything to the bottom of the case, and not building the bridge and control into drawer-like construction I could easily pull out for repairs. The nice piece of wood ended perforated with holes and coil so hard to work on.

Steve wrote ...

In the upper mode, the induced voltage in the secondary due to the primary current is 180' out of phase with the voltage due to the secondary's own current in its self-inductance. This makes the voltage gradient shallower at the bottom of the secondary, where the primary's magnetic field is strongest.

Finn wrote ...
I tuned the primary to the low pole of the combined system.
Now, if the driver locked on to the low pole, it would arc.
If it locked on to the high pole, it would stop arching.

All I can do is tune the primary up or down respectively.

My primary current is already very high for the spark size I'm producing. The pic shown is the primary current before failure, hitting about 600 amps at least. It's running mere 7-8 cycles and although interrupter was <100us, OCD limited it to about 70. I had to see such monstrous dI/dt on a DRSSTC yet.

The surge impedance of primary is just around 4,8 ohms

My current is already very high considering poor sparks I'm producing. Other coils do like 5 feet with 600+ amps where I struggle for 3.

If I tune the primary higher, the current will get even worse. Remember that those IGBT's are rated for like 240amps of peak current in DS.

It is feasible for me tu build a conical primary with more turns, but I need to be absolutely sure that it's not a waste of time since masses of coils use it happily, along with as low Z as I have.

Steve Ward

Mon Oct 01 2007, 09:48PM

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

Marko,

Your suspicions of using too little surge impedance is very true as far as i can tell. At one point i had rebuilt my original DRSSTC using a .6uF tank cap. The primary current was very high (1100A) and i could only manage about 5 foot of spark from the machine. After increasing the tank Z (.45uF cap now), it runs around 700A for 6+ foot of spark. I still think that the impedance might be a bit on the low side, but its performance is pleasing enough for me.

The whole impedance thing is messy to analyze because the apparent impedance depends not just on primary components, but the secondary side (mainly the sparks) and the coupling coefficient. But in general, id say you shouldnt go with less than 6 ohms for Zsurge, and try to keep coupling as high as practically possible (usually its flash-over limited). This is of course really only applicable to my style of DRSSTCs. I think Conner's PLL stuff mucks things up and makes them even harder to analyze... almost as if the PLL adds another order or 2 to the system (its certainly not as easy to model as a proportional feedback control with a delay!).

The other purely experimental conclusion Ive come to is that a lower surge impedance only pays off if your IGBTs can support it. Thats to say that i think a lower Z only allows a longer max spark, but you need higher currents to achieve spark lengths possible with a higher Z at a lower current. Many of my coils have an extremely non-linear response of spark length vs Vin, Ipk. Typically Ipk will rise fast vs Vin (and only increase marginally after about 50% of Vin's max), but spark length takes an extreme jump near about 65% of Vin's max. Before that jump, the primary current is very high, and the output is very weak. Its possible that you are at this point, in which case your IGBTs are not capable of more power so you need to increase Zsurge to lower the peak current to get past the knee in the curve.

Marko

Mon Oct 01 2007, 11:27PM

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

Hi Steve

Yes, I have got way too mean about the low impedance thing and now I suffer the consequences.

I don't actually see what's wrong with using quite high impedance within reasonable limits, it's better for IGBT's. Steve C. used 50nF cap along with a bigger primary, and produced great sparks while not punishing his IGBT's for much at all. I don't remember now, but I think he didn't even drive them out of datasheet rating in normal runs and used 15V gate voltage.

200nF cap would give me about 7 ohms of tank impedance, still a bit low, I would actually target 100nF and 14ohms. This requires complete primary rebuild, though.
Cap will need to be rearranged to stand like 10kV at that point

Good things I've learned from all this, the small TO247 IGBT's are as capable as ISOTOP's when heatsinked well. 5 foot sparks should at least be feasible.

At this time flashovers are biggest concern, especially those at bridge. Could these odd racing flashovers be linked to too fast rate of primary current?
I don't believe even conical primary is going to fix them and something must be done.

I noticed many people use tighter coupling than me, helical primaries, throw sparks 4x secondary length and are fine about flashovers.

vasil

Tue Oct 02 2007, 05:13AM

Registered Member #229 Joined: Tue Feb 21 2006, 07:33PM
Location: Romania
Posts: 506

I have flashes on secondary myself when using cilindrical primaries until I turned to flater primaries forms, like this:
primary. It is easier to adjust just raising or lowering the primary.
It is possible that the flashes risk depends by the raport radius of primary/ radius of secondary. I observed that many drsstc that perform very well have some large secondaries it is possible that larger secondaries process power much better and have longer sparks.
The low voltage and the power stage have to be completely separated and the stages have to be powered one after another. I usually turn on the interrupter ON, then the driver, then I apply the power on the bridge. When closing, I crank down the variac, turn OFF the bridge, turn off the driver and the interrupter at last. When turning OFF the interrupter is possible to have some juice in the DC capacitors across the UCCs in the driver, turning the driver in CW mode and burning the UCCs.

Steve Conner

Tue Oct 02 2007, 10:55AM

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

Hi guys,

I am up to 100nF now and the output seems a little better. I still have the OCD set to the datasheet rating of 400A.

The PLL certainly makes the analysis different, but I don't think it's harder. Since I always know which of the two resonant modes it will be driving, I just analyse it as a CW system operating at that frequency.

This is why I talk in terms of real power flows, surge impedances and loaded Q's, because I see the whole thing as one big matching network between an inverter that delivers real power and a streamer that consumes it.

So, I start with a desired peak power, which I work out from the desired spark length using the Freau formulas and what not from classic coiling.

Then, the impedance of the streamer load is roughly the breakout voltage of the bare toroid squared, divided by the peak power I intend to deliver. This is a consequence of the non-linear nature of the streamer load, which adjusts itself to try to clamp the toroid voltage somewhat like a zener diode.

Adding a breakout point makes surprisingly little difference to the impedance, because the streamer growth is driven by the electric field in the entire space that it grows through, and the breakout point is just a small local disturbance at one end.

Now, the output impedance of the inverter is just the same formula but with 4/Pi times the DC bus voltage.

So, I design the primary circuit to have a surge impedance about 10 times the output impedance of the inverter, and the resonator to have a surge impedance about one-tenth of the streamer load impedance. I get the coupling as tight as possible, tune the primary to a slightly lower frequency than the secondary, and that's all there is to my design method. None of these values appears to be critical.

I know that the system is not CW, that the real power flow is not constant throughout the burst, and that you can't avoid exciting the other mode to the one you're trying to drive, to some extent. But I just ignore these things to make the analysis simpler, and I still seem to land in the ballpark :P

Marko, maybe your feedback CT is generating a high voltage because it doesn't have a proper burden? :|

Marko

Tue Oct 02 2007, 12:48PM

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

I am up to 100nF now and the output seems a little better. I still have the OCD set to the datasheet rating of 400A.

OK, 5 foot spark with just 400A of primary current? I'm beginning to realize how this has nothing with size of IGBT's, frequency pole or characteristic impedance. It's all about altering the reality by bending your perception of it.

I got completely lost trying to understand steve's post, I need to do a lot of googling before I just begin so.

All I can naively ask, what exactly I need to do to lower my primary current/spark size ratio and stop the damned flashovers?

You can consider increasing the tank impedance done.

tune the primary to a slightly lower frequency than the secondary, and that's all there is to my design method. None of these values appears to be critical.

This is contradictory with information you guys just gave about tuning the primary higher.

So how should I tune, higher or lower than secondary Fres?

Marko, maybe your feedback CT is generating a high voltage because it doesn't have a proper burden? :|

That would be odd since diodes are fine, and voltage should be clamped to 5V at most.

I need to disassemble the coil for this, and I fear I won't be doing that until weekend.

I can only appreciate all the theoretical information you guys can give me ^ ^

Marko

Steve Conner

Tue Oct 02 2007, 04:13PM

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

Hey Marko,

I guess it was unfair of me and Finn to bring up the PLL driver thing. The PLL driver has an extra degree of freedom. You can operate on either resonant mode, no matter what you tune your primary to.

Since you are using the self-resonant driver, you should probably tune your primary a bit lower than your secondary, because that's what most other users of this driver seem to do. Once you have raised the tank impedance, just play with the primary tapping point till you get the best sparks that your current limiter allows, and if they still aren't good enough, try a different size of toroid or something.

I don't really understand how the self-resonant one works. All I know is that if you tune the primary lower than the secondary, it will operate on the lower resonant mode, and vice versa.

According to my notes, the 5ft arc needed 450A peak, 300uS burst length and 200Hz rep rate, and over 3kW. :|

Tom540

Tue Oct 02 2007, 05:14PM

Banned on 3/17/2009.
Registered Member #487 Joined: Sun Jul 09 2006, 01:22AM
Location:
Posts: 617

It's a good think you're getting all this help. I learned this the hard way. I hacked up my primary at leat 12 times trying to get this part right.

Definitely raise that tank impedance. It sure saved me some IGBT's, of course after I had replaced dozens of them.
My coils couldn't run on a MMC capacitance greater than 35nF. My latest one uses 27nF and gets 24 inch sparks maybe more if I actually had taps on the primary. The FGA40n60's I was using can only handle 160 amps pulsed so adding more turns and lowering that MMC value was crucial.

Marko

Tue Oct 02 2007, 10:12PM

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

Steve, you do keep confusing me:

In the upper mode, the induced voltage in the secondary due to the primary current is 180' out of phase with the voltage due to the secondary's own current in its self-inductance. This makes the voltage gradient shallower at the bottom of the secondary, where the primary's magnetic field is strongest.

In the lower mode, the induced voltage is in phase, so the voltage gradient is steeper at the bottom.

Since you are using the self-resonant driver, you should probably tune your primary a bit lower than your secondary, because that's what most other users of this driver seem to do. Once you have raised the tank impedance, just play with the primary tapping point till you get the best sparks that your current limiter allows, and if they still aren't good enough, try a different size of toroid or something.

This is contradictory again. You advise me to tune to lower pole, but you also say that it will give steep voltage gradient and flashovers. I hope you didn't get me serious when I said I want the coil to self destruct.

I myself am completely unsure what happens with flashovers at higher or lower tuning at full power, I blew the coil before I could have done anything about that.

What I know is that when tuned slightly lower current rises almost linearily and spark output is not too good in overall.

At best tune the current heterodynes but peak value doesn't get much higher in same time period.

If I go significantly higher than resonant the current ramps up and output is poor.

My coils couldn't run on a tank capacitance greater than 35nF. My latest one uses 27nF and gets 24 inch sparks maybe more if I actually had taps on the primary. The FGA40n60's I was using can only handle 160 amps pulsed so adding more turns and lowering that tank cap was crucial.

That's a small tank cap, even for a smallish coil. What frequency/tank impedance were you running at that point?

Now how high is it feasible to go? If I used really high impedance I could pump lots of energy into the coil without actually exceeding IGBT SOA. Spark length/power input efficiency may be worse, but who cares about that as long as sparks are long.

Steve Ward

Wed Oct 03 2007, 04:39AM

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

The spark length will plateau earlier with higher tank impedance because you simply cant achieve the higher peak powers. Take the extreme example of the typical SSTC... my big one took about 7kVA to make 3 foot sparks but only had about 50A of primary current (at 400VDC). Of course, as you found the other extreme isnt so great either, and results in a poor impedance match to the output spark.

The tuning for the heterodyning of the primary current is the tuning i always go for. It implies a proper energy transfer to the secondary, and gives a nice point for shut down (since driving beyond that point doesnt seem to help much anyway). This tuning might be tricky to obtain with higher tank Z, but perhaps with proper detuning it could work (otherwise you will get a lot of beating).

Of course the best way to learn this stuff is to try it out and see what happens.

first previous 2 3 4 5 next

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