First SSTC, Ward's Mini-SSTC Design / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

First SSTC, Ward's Mini-SSTC Design

1 2 next

Move Thread

LAN_403

lvl7hound

Sun Sept 26 2010, 06:19AM

Registered Member #2477 Joined: Sun Nov 22 2009, 10:56PM
Location:
Posts: 4

I'm new in a sense I haven't done much. Funny enough I'm not that new to this site (I've been on IRC a bunch), but every time I start on this project I end up getting delayed by something. After the latest delay (getting up, moving across the country for a new job and a new coastline) I finally find myself with some time to finish this project. I've actually gotten it to the point where I need to ask questions before I can really proceed!

This is my first tesla coil project. I've done some projects with ignition coils and tube amplifiers before, so I'm not a stranger to high voltage. I also work on electronics for a living, although of the lower voltage variety.

As my starter coil, I'm using Steve Ward's Mini-SSTC as a reference design (someone on IRC recommended this as a good starter point). I've added a few deviations, but nothing major. I changed a few decoupling cap values, added some jumpers, split the board into controller and h-bridge side, and probably most importantly decided to use FGA30N120FTD IGBTs instead of MOSFETs. I have a feeling the latter might bite me in the butt later, but my reason for doing so is that I eventually want to do some experiments with audio modulation, and I figure now is the time to get some super-beefy transistors to deal with that.

I've split this coil up into 4 parts, all of which I have some questions on. I'll go into details of the construction in case anyone see's any major flaws as well.

Controller
Gate-Drive Transformer/IGBTs
Half-Bridge Driver/Caps
Coil

The Controller

Prettymuch exactly what Steve Ward's schematic shows, I don't anticipate may problems here. I did replace a few of the diodes due to issues of part availability. In particular, the 1N60's were replaced with 1N4001 diodes (I realize 1N60's have a low voltage drop, I figure actually having a slightly higher drop may be beneficial for the Schmitt trigger turn-on. Plus the voltage rating is higher.

Aside from that, the only major change is that there are jumpers between the gate drivers and the 555/7414 circuits. I added these because I want to experiment with some PLL drivers after I get this coil working, and it's a way to get a known-working coil and start hacking away.

The only real questions I have here are what kind of duty cycle am I looking for out of the 555? I've been seeing conflicting numbers on what the duty cycle should be, I assume it should be pretty small (10% or less) to prevent excessive heating no? Then again, with the GDT in play it seems like my duty cycle is more limited by that, as I can crank it up to 50% and still see about 5% or so on the actual output of the H-bridge.

The GDT and IGBTs

This is also mostly identical to what Steve Ward's schematic shows, with the major exception that I've moved to using FGA30N120FTD IGBTs, instead of the IRFP260's that Mr. Ward calls for. I chose this because I want experience with IGBT designs, as I want to eventually move into DSSRTCs and audio-modulated designs. Plus this seems like a much more robust transistor, and it gives me an inventory of these bad boys to play with later. Is this going to be a major mistake? Or am I safe? (I'm pretty confident I'm safe, but it never hurts to ask!)

For the gate drive transformer, I am using a Ferroxcube TN25/15/10-3E25 (which took a very long time to get in), with an inductance factor of 5620nH, size 25.8mm - 14mm diameter x 10.6mm depth. I have wrapped 3x 20-gauge magnet wire strands 13 times around the core. I made a spreadsheet to help calculate the parameters here, and so far I see:

Volts = 12 volts
L = Torroid L Factor * Turns^2
Ipk = Volts * Cycle Time / Inductance

Inductance - 949.78 uH
Peak Magnetizing Current (50% @ 100kHz) - 0.063 A
Peak Magnetizing Current (50% @ 300kHz) - 0.0211 A
Peak Magnetizing Current (50% @ 500kHz) - 0.0126 A
Peak Magnetizing Current (50% @ 700kHz) - 0.0090 A

I was a little confused about tradeoff of number of turns vs. saturation density, so I'll ask how this matters? I was looking at the High Voltage Wiki GDT Page. Using that formula in my spreadsheet, and targeting 0.2 Tesla, I get:

Target - 0.2 Tesla (Not sure where I actually got this, I think it's in a datasheet somewhere hidden)
Min Turns = Volts * Cycle Time / (Cross-sectional Area * Target Teslas)
Min Inductance = Min Turns^2 * Torroid Inductance Factor
Mag Current = Volts * Cross-sectional Area / Min. Inductance

Frequency	Minimum Turns	Minimum Inductance (nH)	Magnetizing Current (A)
100kHz	6.13496932515338	211524.709247619	0.28365480427046
300kHz	2.04498977505112	23502.7454719577	0.85096441281139
500kHz	1.22699386503067	8460.98836990478	1.41827402135231
700 kHz	0.87642418930762	4316.83080097183	1.98558362989324
1MHz		0.61349693251534	2115.24709247619	2.836548042704631

I'm not sure what this is telling me. The minimum inductance is far higher than what I can provide but it seems absurdly high? I think I'm just confused in general what I should be doing here.

The IGBT's themselves have an input capacitance of 5140 pF, and a total gate charge of 208 nC. I've been seeing the transistor switch on and off with my oscilloscope and the device both unloaded and loaded with an ignition coil as a tesla coil standin, so I don't think there's going to be issues there.

Also, is there a way to calculate the resistor between the transformer and the transistor gate? I assume the hard way involves modelling the transistor as a capacitor and trying to hit or exceed that magical gate charge number. Any rules of thumb?

Half-Bridge Driver/Caps

Getting out of that mess, the Half-bridge is pretty identical to Steve Ward's design. I've purchased Cornell 942C20P15K-F caps (10 of them, I'm not screwing around). I've installed two of these for the half-bridges, one on each side. I did this for testing, I'll install more once I get a coil up and can start adding/subtracting them from the equation. Any advice on a housing for multiples of these caps? I envision maybe 3 steel plates with holes in them, for soldering?

Coil

I'm following directions at this site. Is it trustable? If so, I have a 3" and a 4" PVC pipe, both 2 feet long. I plan on using the 4" pipe as the secondary, and wrapping 20" of wire around it, unless someone tells me not too. Good idea/bad idea?

For the primary, I have 12-gauge galvanized steel wire. For the secondary, I have 2000' of 28-guage magnet wire. I calculate I'll need about 1500 feet for a 4"x20" tesla coil. Wire gauge is good, right?

I'm going to hook it up and take some scope screenshots. Don't know if it will be all that relevant without a coil in place, but I figure there may be some information to be gleamed from it. I do have a Variac and a 100x scope probe, so if I run at lower voltages I can probe the output of the bridge! I just got the Variac today though, so I've just been using a DC power supply up until now for debugging.

Thoughts, Answers, Pitchforks?

william L

Sun Sept 26 2010, 06:55AM

Registered Member #3093 Joined: Mon Aug 09 2010, 11:40PM
Location:
Posts: 68

I went through tis coil as well.
What I got on my first light was 4 inch streamers, much like that of a vttc.
The problem was that I added filter caps to the bridge later on, but they were backwards!
After lots of solder, ringing problems and bogus mosfets from ebay, I've been left with a pile of scrap that can barley make 1 inch.
Some Advice I have for you:
*Decoupling caps are your friend. Love them.
*Be careful what you scope.
*ALWAYS ALWAYS ALWAYS check polarity of items.
*Don't let stuff get too warm, as I have had many such problems.
I wish the best of luck to you.

On the topic of hooking it up without a coil:
you probably wont get much besides your interrupter, and some 50/60 herts noise.
Also, I would say that for a 20" by 4" coil, this driver probably won't be too impressive.
Once again, GOOD LUCK!

EDIT: Those IGBT's have a pretty high delay. I don't feel like math right now, but you may want to switch to something a little less sloth-like.

lvl7hound

Sun Sept 26 2010, 07:44AM

Registered Member #2477 Joined: Sun Nov 22 2009, 10:56PM
Location:
Posts: 4

Haha polarity, I'm trying to keep the number of electrolytics down. I'm not using any tantalum caps, because I always screw those up :p (Plus even when they do work I haven't found them to be terribly forgiving capacitors). Notice I haven't installed the huge 2000uF+ caps yet... All of the others are ceramics, 0.1uF or 1.0uF. The cap in the sub point is a 2uF.

I've tested the interrupter mostly, I've also tested the antenna with the 'finger test' (touching it and seeing the 60Hz noise). The rest will wait until I get a coil running. I tried to get loopback with an ignition coil, but I didn't have enough power to drive it at the time (stupid DC supply).

I realize those IGBTs are quite slow. I may try upgrading to faster models or even the original MOSFETs, I think they are pin compatible. I did find some awesome automotive MOSFETs the other day, hrmm....

Do you think a 3"x18" coil will be easier to drive? Significantly easier?

Marko

Sun Sept 26 2010, 11:57AM

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

For the primary, I have 12-gauge galvanized steel wire. For the secondary, I have 2000' of 28-guage magnet wire. I calculate I'll need about 1500 feet for a 4"x20" tesla coil. Wire gauge is good, right?

Uh, no, don't ever even think about using steel for high frequency conductor, not even small connections. It shouldn't be that hard to find some decent copper wire for your primary.

I do have a Variac and a 100x scope probe, so if I run at lower voltages I can probe the output of the bridge! I just got the Variac today though, so I've just been using a DC power supply up until now for debugging.

No, you can't - bridge explosion is assured if you connect your scope straight to it's output. You'll need an isolation transformer to do that, or use two probes in differential mode.

Regarding your igbt's, I looked at the datasheet, and they seem to be rather slow standard speed ones with huge fall times. I wouldn't use them for this design.

It's possible to use latest generation hyperfast igbt's for SSTC's, though they may end being less efficient than the mosfets you intended to use, but not necessarily (at high currents they beat mosfets by far in conduction loses). I'd say up to 100kHz is optimum for igbt's.

What voltage are you running at? Unless you're going to be supplying like 800V to the coil it makes little sense to use 1200V devices anyway. If you're running from 120V mains IRFP260 are probably among the best devices for the job you can get.

Finally regarding your secondary, it looks rather large in my opinion if you're going to use entire tube. For a 2xIRFP260 coil I'd use like half of that PVC tube at most. Stubbier coil also gives better coupling than the long one for same primary position.

Marko

Marko

HV Enthusiast

Mon Sept 27 2010, 03:21PM

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

I wouldn't use steel wire. Steel is very lossy at high frequencies. Use copper, or aluminum. No reason you can use copper as its so readily available, especially in 12 AWG.

Steve Ward

Mon Sept 27 2010, 10:10PM

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

Not sure if anyone caught this, but a 1N4007 is absolutely nothing like a 1n60!!! A 1n60 is an ultrafast response germanium diode made for high speed signals, the 1n4007 is a crappy rectifier diode and simply wont work properly here. A 1n4148 or 1n914 would be suitable replacements for the 1n60, which is indeed hard to come by now days.

HV Enthusiast

Tue Sept 28 2010, 03:18PM

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

LVL7Hound -

I have some 1N60s.

If you need some, let me know and i'll send you some at no charge (assuming you are located in the US)

lvl7hound

Sat Oct 02 2010, 02:44AM

Registered Member #2477 Joined: Sun Nov 22 2009, 10:56PM
Location:
Posts: 4

"SteveWard" wrote ...
Not sure if anyone caught this, but a 1N4007 is absolutely nothing like a 1n60!!! A 1n60 is an ultrafast response germanium diode made for high speed signals, the 1n4007 is a crappy rectifier diode and simply wont work properly here. A 1n4148 or 1n914 would be suitable replacements for the 1n60, which is indeed hard to come by now days.

Ahh, good catch. Mouser still has some stock, and so I purchased a few along with some IRFP260s. Thanks for the offer EasternVoltageResearch, Mouser wasn't too pricey though.

"EasternVoltageResearch" wrote ...
Posted: Mon Sep 27 2010, 08:21AM
I wouldn't use steel wire. Steel is very lossy at high frequencies. Use copper, or aluminum. No reason you can use copper as its so readily available, especially in 12 AWG.

I wasn't too sure what to expect with the steel besides higher losses (didn't realize that climbed at a higher rate than copper with frequency, unless I misread that). Just for fun, I'd like to see some data on that if anyone has some.

"Marko" wrote ...

No, you can't - bridge explosion is assured if you connect your scope straight to it's output. You'll need an isolation transformer to do that, or use two probes in differential mode.

What's the mechanism for this, arcing through the probe or is current really that high? I was planning to use this at lower voltages (~20VAC), but I'll abstain from this practice alltogether now.

I decided to switch over to IRFP260s, I'll save the big IGBTs for another project. I'm going to do some more research now on the sizing and coil design now, I found some websites with good information (especially

it looks like).

ScotchTapeLord

Sat Oct 02 2010, 03:43AM

Registered Member #1875 Joined: Sun Dec 21 2008, 06:36PM
Location:
Posts: 635

To scope something, you need to ground one side of the load and probe the other. If you ground any part of the AC bit of an H-bridge, you short the top transistor of that leg to the ground!

lvl7hound

Sat Oct 02 2010, 06:02AM

Registered Member #2477 Joined: Sun Nov 22 2009, 10:56PM
Location:
Posts: 4

Oh, I was more thinking of removing the grounding strap and just using the probe tip with the scope on AC coupling. Probing something this high current with the ground connected would be a bad idea. I'm aware tip-only gives terrible quality signal, it's better than nothing though.

1 2 next

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