Autopsy of a successfull failure... / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

Autopsy of a successfull failure...

1 2 next

Move Thread

LAN_403

Finn Hammer

Wed May 12 2010, 08:58PM

Registered Member #205 Joined: Sat Feb 18 2006, 11:59AM
Location: SkÃ¸rping, Denmark
Posts: 741

5 years ago, I built 3 coils for Danfoss Universe, a science theme park in Denmark. Steve Conner helped me immensely during the construction and installing phase of these coils.
One of these coils was a replica of his OLTC coil, the rave of itÂ´s time, at least on this side of the pond, and this coil runs on a daily basis to this day.
The 2 other coils failed miserably the other day, and I have brought the first one home to the lab for an autopsy, and repair.
They are both SSTC's, based on Eastern's driver, and my coil design, here seen prior to the fit of the front part of the faraday cage:

They were designed to produce this display, in short bursts, to simulate the experience of faint and distant discharges inside of a thunder cloud:

At that time, I was absorbed by the idea, that internal primary coils was the safe way to avoid strikes to the primary.

This goal was acheived, but another problem arose: The voltage gradient btwn. pri. and sec. got so big that corona was formed on the primary coil.

A closer look

The corona started to carbonize the inside of the secondary coil, and this carbonized spot started to act as the plate of a capacitor. Trough this plate, charge was transferred from sec to pri, and in the process, the plate expanded towards the high voltage end. Finally, the IGBT's were overvolted, and failed.
I knew this could become a problem, because when I tested the coils in CW mode, the effect was pronounced. However, since the coil were producing short bursts in the display, I took the chance of ignoring this, and it took 5 years to develop into a fatal failure.
Around that time, I took the initiative to code a LUA script for FEMM, which automated the drawing of a coil structure like this, aimed at producing voltage gradient plots in FEMM, to evaluate how to design potting and shielding techniques for this type of coil. This script was shortly rewritten by Father Dest, a russian coiler, and it became a well functioning tool for the task at hand. I still have the potting compound, but as so many projects, it had to be scrapped due to lack of time and urgency.
A significant observation is that there is no puncture of the coil former, so a repair consists of scraping off the tracking, a lacquer coat, and construction of an external primary coil.

Cheers, Finn Hammer

Sulaiman

Wed May 12 2010, 10:53PM

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

Interesting, sad but interesting.
Thanks for sharing.

Is it an established fact that the Lichtenberg / electrical tree figures are propagated by ozone?
(in this slow-motion version)(just curious)

I don't have your experience but I suspect that a thorough cleaning and no laquer coat may be best
- coatings eventually degrade or detach and absorb humidity etc.
- future repairs (if any) would be simpler without a laquer coat too I guess.

Hazmatt_(The Underdog)

Sun May 16 2010, 04:27PM

Registered Member #135 Joined: Sat Feb 11 2006, 12:06AM
Location: Anywhere is fine
Posts: 1735

That damage looks pretty extensive to me. I think you're going to need a new coil form, but you ought to do some Mohm readings from a coil lead to the tracking and several other strategic measurements to make sure its salvagable.

Steve Ward

Fri May 21 2010, 05:10PM

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

This brings up an interesting point, rather a difference between the world of professional HV design (of which i am a part of) and hobbyist HV design (of which i used to be, and still sort of am a part of). Basically, most hobby designs are non-conservative at all, especially when it comes to corona discharge, or the possibility for it. Many hobbyists are not familiar with what even un-seen corona will do to insulators given enough years. Partial discharge testing is used to ensure that devices are *corona free* in the operating region. Youd be surprised how even a structure at 5kV can be difficult to be corona free, it ends up looking like what a hobbyist would assume is good to 50kV. And the fact is, it would work just fine at 50kV for a few 10s of hours, but not for a few 10's of years like we need. Youd be surprised at how things degrade slowly, over years and years of use.

I rely on using FEMM for 2D modeling of field stress, but also have a 3D software package at work when the need arrives. Generally, a design intended for long life will operate with the maximum E-field being just 1/3rd of the specified breakdown of the material (oil, air, silicone...). Based on many of my FEMM simulations of my tesla coils, i find that even stuff near the very bottom of the secondary coil is still near the 1-2MV/m field stress! Yikes... good thing its not installed in a theme park running day after day! In fact, im not sure exactly how to design your typical tesla coil without exceeding 1MV/meter field stress! I guess thats why DC Cox has huge coils that make wimpy-ish sparks, enough failures have led him to understand how to make things last in the long run.

Finn, could you report back on what your guess of the field stress was on your secondary/primary structure using your 2D FEMM? Id be curious to see if its more than 1MV/meter (which is about where we'd draw the line at work). But just looking at the geometry of the primary coil inside there, im not the least bit surprised it finally gave out, i am surprised that the damage to the coil former isnt worse and didnt happen sooner!

Finn Hammer

Fri May 21 2010, 08:44PM

Registered Member #205 Joined: Sat Feb 18 2006, 11:59AM
Location: SkÃ¸rping, Denmark
Posts: 741

Steve Ward wrote ...

Finn, could you report back on what your guess of the field stress was on your secondary/primary structure using your 2D FEMM? Id be curious to see if its more than 1MV/meter (which is about where we'd draw the line at work). But just looking at the geometry of the primary coil inside there, im not the least bit surprised it finally gave out, i am surprised that the damage to the coil former isnt worse and didnt happen sooner!

I would like to, but unfortunately, I donÂ´t have FEMM installed at this point in time, so this will have to wait.
I am also in the midst of a divorce, so time is limited to the most urgent tasks at hand.
But there is no doubt that corona was sproutting off those primary windings every single time the secondary approached peak voltage.

I have now rebuilt the coils with *standard* external primary's, and they are ready for another 5 years of daily abuse.
Interestingly enough, the second of the coils did not sustain any damage to the power electronics: It failed due to a broken transformer, powering the interruptor. Saved by the bell.

Cheers, Finn Hammer

quicksilver

Sat May 22 2010, 04:21PM

Registered Member #1408 Joined: Fri Mar 21 2008, 03:49PM
Location: Oracle, AZ
Posts: 679

The idea of an internal primary is attractive in other ways also. But does this mean that internal primaries are too problematic for they're return?
Would this have been such a problem with a much smaller coil?
Is the corona issue solvable with some type of insulation or strike-rail configuration?

Dalus

Sat May 22 2010, 05:42PM

Registered Member #639 Joined: Wed Apr 11 2007, 09:09PM
Location: The Netherlands, Herkenbosch
Posts: 512

Wouldn't a oil filled secondary with the primary and oil internally be a solution. Think it's the only way of making a coil operate for long stretches of time without failure.

Finn Hammer

Sat May 22 2010, 07:18PM

Registered Member #205 Joined: Sat Feb 18 2006, 11:59AM
Location: SkÃ¸rping, Denmark
Posts: 741

I did a lot of simulations with FEMM, working on a potting solution, where the secondary contained an electrostatic and grounded shield between primary and internal secondary. The object was to get high coupling and the option of adjusting the coupling on the fly.
I can attest to the fact, that it is not a trivial task to avoid field gradients in corona forming magnitudes, even when the shield was formed with the largest possible radius of curvature at the top end..
Oil is an atractive option, untill you start to work on practical ways to contain it.....
Given the later proven fact, that a small bypass cap from primary to ground solves problems associated with flashovers, and strikes, from secondary to primary, the attractiveness of the internal primary started to fade...

Cheers, Finn Hammer

Hazmatt_(The Underdog)

Sun Jun 13 2010, 04:58PM

Registered Member #135 Joined: Sat Feb 11 2006, 12:06AM
Location: Anywhere is fine
Posts: 1735

How about back-filling it with SF6. Even at low pressure you could fill the tube as if it were water and it won't leave the tube so long as your electrode ends are fully sealed.

Arcstarter

Mon Jun 14 2010, 12:21AM

Registered Member #1225 Joined: Sat Jan 12 2008, 01:24AM
Location: Beaumont, Texas, USA
Posts: 2253

At least the internal primary works for OLTC's, due to the really low coupling.

Yea, i know how much corona can get in the way, but i haven't the slightest idea on how to protect and minimize corona.

1 2 next

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