CWDRSSTC / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

CWDRSSTC

1 2 3 4 next last

Move Thread

LAN_403

Dr. Dark Current

Fri Nov 11 2011, 07:53PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

Hi all,
I was planning to build a large SSTC, running on halfwave rectified unfiltered two-phase voltage (400 Vac), in the several kW range. However then I realised that the switching losses would probably be huge and left the idea.
But then came to my mind, what if I used a tuned primary with primary current feedback and zero current switching, which would greatly decrease switching losses? The primary impedance would, of course, need to be increased over a standard DRSSTC.

Before I start putting it together, do you think this is even practically doable?

Here are some preliminary data on the test setup:
The average input power was chosen at 4 kW, which means approximately 16 kW peak power at 560 V bus voltage. The coil would use an IGBT half bridge, so the output current at input power peak is 16 kW / 280 V = 57 amps. I selected a Q of 10 (this is the part I'm most unsure about) and calculated the primary cap to have 2800 volt peak on it at 80 kHz, which turned out to be 57 nanofarads.
The primary impedance is adjusted by tuning the primary on the lower slope of the resonant peak (=tuning and detuning below secondary resonance).

Does this sound right?

Steve Conner

Fri Nov 11 2011, 08:01PM

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

The first DRSSTC I ever saw was a CW one built by Richie B, with MOSFETs. The pulsed ones are a new thing!

He used his usual SSTC bridge with a 4.7nF high voltage ceramic tube cap from a transmitter. It lasted about 10 seconds before exploding.

DRSSTCs "sort of work" with a wide range of tank LC ratios. I've tried 10, 50 and 100nF at 220kHz. It doesn't seem to be terribly critical: 57nF at 80kHz would probably work.

Goodchild

Fri Nov 11 2011, 09:47PM

Registered Member #2292 Joined: Fri Aug 14 2009, 05:33PM
Location: The Wild West AKA Arizona
Posts: 795

You know basically what you are building is QCW type coil.

The tank impedance is high to help keep the current low, you are also using unfiltered half wave aka "the ramp"

VTTC run is basically the same operation supper high tank Z and a CW or half wave operation.

Marko

Sat Nov 12 2011, 01:02AM

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

Hi guys

Let me throw some of my experimental data. I've been playing with and improving my CW coil over time and am also about to take this route. But it's not as much because of switching losses alone, but the problem of having to drive such extreme reactive power through the low inductance primary.

My coil is so far still a normal SSTC and uses a full bridge of HGTG30N60A4D IGBT's which keep proving their extraordinary resistance to punishment, having ran up to 10kW input. They actually outlived all expectations and I have almost used up all my available socket current - yet it produces such amazing CW plasma that I just keep wanting more.

I installed a current limiter into my coil which monitors the bridge peak current, and it hit over 80A (in CW, so RMS current had to be around 60 amps!)

I didn't want to continue with any more of this completely unnecessary punishment for the IGBT's (well, I also had a problem of my cooling fans slowing to a crawl from interference, and have to replace them with AC fans). So I decided it would be time to consider a resonant primary topology. I had several considerations:

1. LCLR - I dismissed this soon after learning that it will actually not lead to much improvement in power factor compared to now.

2. Current fed inverter - Good PF, but unexplored topology for this power level and would require significant alterations on my current installations in this coil. I would very much like to explore this further though, especially for high frequency coils.

3. Series impedance matching network base feed - basically a DRSSTC, but with direct base feed of the secondary from the primary tank! The bridge drives a serier LC circuit of fairly high impedance, and the base of the secondary goes in between L and C. While this might look troublesome, a beefy inductor is actually far easier to design than a HV transformer (you can also finally put to use all those AM radio ferrite rods you have around). You could probably just use an air cored inductor like a spool of wire too, and C could be so small that you can use a big air variable cap for it (if you have one!). Would be very fun to try in my opinion.

4. High impedance series resonant - basically what you intend to do, use a high voltage MMC and a primary with many turns so you can drive this directly from your bridge. I was considering this heavily until I done some experiments - using a 100nF MMC which resonated well with my current primary, my input current skyrocketed even with just 20-30V input. So I don't think a 50 or so nF MMC will do for you, you will want more like the 3.3nF Steve mentioned earlier!

Note that this is also a tradeoff between your desired coupling and the tank impedance - the more weakly coupled primary will load your inverter harder, and you'll need to increase it's impedance to counter this. If you're building a coil in this topology from start, I assume you don't want the pointless risk with overcoupling (like it's done with normal SSTC's) so that is another thing to consider.

Only a good FEMM simulation or practical experimentation can lead you to the answer on what parameters you need.

The big drawback will definitely be the costly and cumbersome primary coil which will need to have several tens of turns of high current conductor (prefferably copper pipe) which you need to design into low enough profile so that it doesn't present a flashover risk.

I would try using several helical coils inside each other, possibly the inner ones being cable and only the outer one made of pipe to reduce the bending trouble.

The benefit here is that you can use transmitting doorknob caps for your tank cap instead of constructing a cumbersome MMC.

5. Low impedance series resonant: This is what I intend to do with my coil right now. Basically it's the same as above, but keeps the fairly normal primary and a DRSSTC-sized tanks cap. And instead of driving this directly, the bridge is coupled over a big step-down ferrite transformer. I got a nice ferrite toroid for this today and might carry out some quick experiments tomorrow.

With unity power factor on my bridge, it will be carrying only like 25 amps instead of those 60! So theoretically I will have no trouble pushing 20kW or even more considering I now have zero current switching... and you think that will make me satisfied? Not really - the next thing I'll probably want to do is replace the igbt's with some of those 1200V, 72A puppies whose dies are so large that no diode can fit in... but small external diodes can be used now since they have to carry only the transformer magnetizing current! The good thing is that garage breakers are unscrewable type, so I could temporarily hack in some 25A ones!

And when I truly end up with no more power available, that won't stop me ofcourse... I'll nicely ask at college for a high current 3 phase lab feed, and truly give an epic ending to this without caring if it turns into one big arc flash in the end. (BTW, No joke at all, I'm dead serious about seeing 50kW from that thing, if at least for few seconds. *epic laugh with arms raised and enormous CW sparks in background*

Anyways, enough of motivational talk - kilovolt, it's on you to decide what topology to choose (or whether you want to make your coil experimental and try them all!) Make sure to refer to Big Bad, the CW coil thread for more information!

One other challenge you'll probably face is also that there's not even a good, reliable SSTC driver circuit around that can keep lock with full and half wave rectified input well. I'm in process of designing one too!

Cheers, Marko

Dr. Dark Current

Sat Nov 12 2011, 08:25AM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

Hi Marko,
thank you for your comprehensive reply.
The problem with the small tank caps is huge voltage on them. The bridge output current cannot be changed and is a function of output power - in my case I calculated 57 amps RMS, if I used a 4700pf cap, there would be 24 kV RMS (34 kV peak) on it at 80 kHz !!!

I think the big increase in current with small voltage you observed is normal. The current should rise close to the maximum value at low supply voltage, and with increasing the voltage it should remain almost constant.
The key here is to properly tune the primary - if it's too far away from secondary resonance, the current will indeed rise to huge values. Tuning closer to secondary resonance (always tune below secondary resonance) makes it "suck" more energy from the tank, decreasing the current. The tank circuit must be tuned so that just the right energy transfer happens, and from VTTC experience this "sweet spot" is rather narrow. But I haven't tried this so far so I can't say for sure.

Marko

Sat Nov 12 2011, 11:52AM

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

Hi kilovolt,

well, yeah, it won't be easy to find the right RF doorknob caps for the application and will surely cost a lot. You might in the end resort to using an MMC, but it will be bulky as well with many small value caps in series per string. That's another reason why I opted for the transformer option.

Oh, I also didn't mention that I'm still using secondary feedback, and will continue doing so while tuning the primary slightly inductive.

I think Steve ward ran his QCW coil in true CW with about 20nF cap, but he used what I would think of as insanely high coupling (which apparently didn't matter anymore once the spark growth was actively controlled... but arcover in CW would certainly be a disaster)

By the way, I just bought some of these caps for my MMC, they are a great bargain (just don't buy them all! :) )

Oh, and have you even considered trying a base feed by a matching network? It sounds like a lot of fun and will be also confusing for most people to see a coil sprouting big sparks without a primary coil around it..

And heck, if you think this all is troubling and just want to have a reliable 4kW coil with no ominous greed for more power... just wind a secondary for about 100kHz and use a full bridge of HGTG30N60A4D's, I've tested this up to at least 10kW already! But I must warn you... 4kW CW sparks don't actually look so amazing as you would think, unless you make a smallish tabletop coil.

Marko

Dr. Dark Current

Sat Nov 12 2011, 12:16PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

Marko,
I need to use a two phase (400 volt) supply for the coil, the 230 volts here only go up to 16 amps (which means at most a 2.6 kW coil supplied from a half wave rectifier).
For this reason I bought 1200 V IGBTs, which are somewhat slower than 600 V ones, so I want to use a ZCS topology to decrease their switching losses.
I'm not sure about the base fed topology, as I need to use primary current feedback and here the tank would need to be tuned exactly to secondary resonance, I think detuning with spark loading would be a problem here.

Marko

Sat Nov 12 2011, 03:31PM

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

Hi Jan,

Well I've just made some of the first successful runs of the CWDRSSTC. I used a 27:9 transformer and a 125nF MMC, and actually this already seems quite reasonable for the coupling I'm using. So if you insist on using a half bridge and are OK with typical SSTC-like coupling your idea of direct drive with a 57 nF MMC might not be so far fetched at all, especially if you'll be minimizing your average power by using half or full wave input rectification.

I have a big 3 phase variac so varying supply voltage is not a problem for me, but you could still use 600V IGBT's and perhaps a buck transformer in the input if you're afraid of slamming full 560V onto them.

On the other hand, I intend to switch on 1200V IGBT's too, particularly these ones

and run them up to 1000V supply... they have much steeper temperature/current de-rating than 30N60's but are actually quite fast.

One thing I might in the end do is re-tying my H bridge to work as two half bridges in parallel, and leave out the transformer altogether for the most insanely powered runs... but we'll see what comes of this in the end. My camera has charged so going to do some more runs and post pics and vids here.

Cheers,

Marko

Dr. Dark Current

Sat Nov 12 2011, 03:51PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

Cool

Looking forward to the pictures.

The idea of this was to keep the coil simple, so no buck converters for me. Just a directly amplifying 74HC14 feedback and two gate drivers. The IGBTs I have are these

I intend to put 2 and 2 in parallel for the 4 kW half bridge.

Marko

Sat Nov 12 2011, 05:46PM

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

Hi sir kilovolt,

First, a video of a run at about 8kW, set the current limiter to some 40A peak and it never tripped until the full 400V input (limited by my DC link cap rating, really wouldn't want those to blow up)

Now, some pics of the setup.
Here's visible the ferrite transformer, as well as 125nF DRSSTC MMC I used as a tank cap. While the transformer did get warm in operation, the caps didn't even break a sweat at this power level (so the MMC is indeed fairly overkill)

The mess of the front of the coil, still with the driver board I'm unhappy with.

The whole system, ready to be fired up

Now, some more comments.

Jan - firstly, don't even think about using the 74HC14 input thing in your driver, it sucks. This is a Steve Ward's "invention" that has been plaguing people for years. I actually ended up feeding the base current sense CT directly into the inputs of UCC's, and this seemed to work quite well. But when I tried to recreate the same with other gate drivers, or using 74HC14 as a buffer, this resulted in overwhelmingly bizzare driver behavior and blown switching devices.

With the huge hysteresis 74HC14 introduces in the system, you need a powerful stimulus from a starting oscillator or interrupter to start it up (especially with half wave pulsed supply) and this seems to introduce huge problems because of later interference of the oscillator with the operation.

My coil still uses UCC's with a TL494 for startup (see the schematic in project thread) but it only ever seemed to work well in CW and after I introduced the dual resonance now the things got far worse, I can barely start the coil up and in short it's no good at all.

To make things worse, Steve ward designed a PLL driver which also had problems with modulated input (which he solved using the bilateral switch hack). It seemed to work for a while, but apparently he dismantled the coil because his mosfets kept blowing up (I think this is mainly due to use of isotop devices, to make worse, mosfets, which couldn't take the thermal load).

One could wonder, if we opt out using the Steve's PLL circuit, is there even a single reliable CW/HWR/FWR driver in existence?

The problem with the upper approach is the fact that input of 74HC14 is extremely poorly biased for small signals. We really want a circuit that will start at few mV input, and this is really a job for a good fast comparator. This is what Steve's very first SSTC's actually used.

For start, we can connect + pin of the comparator to say +5V and between it and another insert a CT shunt resistor between + and - . We're now having a huge loop gain compared to before and even tiny signals could trigger the oscillation. If noise is not enough, a very weakly coupled starting oscillator should do the trick.

External trimmers may be used to null the offset even further if necessary.

I'm in process of designing such a driver, and it will replace the old one from this coil in the scheduled refit.

Cheers - Marko

1 2 3 4 next last

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