[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Racing Arcs Explained???

Original poster: Vardan <vardan01@xxxxxxxxxxxxxxxxxxxxxxx>


Just to follow up...  If we take the example Dest gave:


and convert it to "inches stuff" with just minor length changes to get round numbers we get:


Solenoid - 6 inches high, 12 turns, 11 inches in diameter of 3/8 inch copper tube.

Secondary - 28 inches long, 1500 turns, 7 inches diameter. 0.015 wire.

Firing voltage 20kV.

But he does not mention Cprimary... That value is needed to know di/dt which is critical factor too...

So we add a 6 x 24 toroid top terminal with a centerline six inches above the top of the secondary. We can now go to JAVATC to find Fo and such...


91.67kHz and the primary is 39.462uH with a coupling of 0.306...

We will assume now that the primary is tuned 5% low as usual and find Cprimary as 85nF. A coil this size with 0.3+ coupling and 17 joules per bang "should" be a racing arc disaster!!!

Totally ignoring any "resonant" effects and totally relying on "pure transformer action", the voltage induced in the secondary is:

V2 = M x di(Lp)/dt

Since M varies drastically along the length of the coil, we have to chop it up and calculate it in little chunks or sections... We'll do every 0.1 inch since the computer is doing all the work ;-)) So I pump it into MandK and get this:


So we know the mutual inductance of each little section now...

To find di/dt we know Ip is in the first instant of firing:

I(t) = Vfire x SQRT(Cp / Lp) x SIN (t / SQRT(Cp x Lp))

If we take the derivative:

dI(t)/dt =  Vfire x SQRT(Cp / Lp) x 1 / SQRT(Cp x Lp) x COS(t / SQRT( Cp x LP))

If t ~~= 0 and with some simplification we get:

dI(t)/dt = Vfire / Lp  == 20000 / 39.462uH  ==  507e6 A/S

Now we can get out the spreadsheet and make some numbers:


and we can make pretty graphs:



So at 3 inches, the voltage stress along the primary is 35000 volts / inch in the first fraction of a microsecond after firing.... Dest got 40kV/inch too!!! We can look at this chart:


To see that it will jump 1.26 inches!!! We have an arc for sure, and super sure too since it is a surface creepage arc that will actually arc at even far lower voltage...

Worse yet, once the arc is established, it will look much like a short on the secondary and will very quickly absorb all the energy in the bang. Thus explaining why they are "bright" arcs that do a lot of damage...

On the good side, thick layers of insulation should protect against such voltage breakdown!! Put that poly coating in real thick ;-))

Of course, once the resonance is established, such effects will "ride" on top of the voltage on the secondary too... I imagine that will explain the arcs occurring at places other than the base of the secondary...

The voltage profiles Paul has at:


Will enter into this all too...

Of course, Gerry's paper provided the key to it all ;-)))


I CC'ed dest too incase he has comments...




Dest also looked into this:


But I don't think he did it quite right... The links are dead so I can't check now. But I think he did the di/dt wrong. If it was done right, I think his conclusion would be different ;-)))

I have seen miss-tuned coils do some very strange things. Fat 1 foot arcs from the lower six inches of the secondary to the primary... Primary to secondary arcs blowing holes though 1/4 inch G-10... And of course racing arcs... I think all this will be explained in a few days now :-))))))



At 10:17 PM 8/13/2006, you wrote:
Hi Terry,

Yes, I think I have enough info to compute the initial voltage profile. It will be interesting to also add the wave propagation to the model to see what happens over time and subsequent half cycles. My guess is that the problem is more complex than the initial voltage profile but modeling the secondary into managable number of segments may be a reasonable way to shed some light. Maybe we can share ideas on this and the modeling and I can proceed. You have a very intuitive perspective.

Gerry R.

Original poster: Vardan <vardan01@xxxxxxxxxxxxxxxxxxxxxxx>

Hi Gerry,

You may be able to resolve the racing arc issue now.

You know the di/dt of the primary coil.
You know the coupling to any arbitrary ring along the secondary.
You know the inductances.
You have a well known coil situation that is apparently close to the edge to study.

If we assume the racing arc occurs in say the first 1/4 cycle (probably in the first micro second!), then the other resonances and harmonics have no effect since they have not had time to setup yet.

Thus, the voltage gradient along the secondary in the first instant of firing is now taking into full account the primary to secondary geometry and coupling...

MandK can be set up to find the data in a single run and you can modify the MANDK.INI file for further control. You can just make the high increment the same as you coil section increment to calculate the right data directly.


The coupling/length output data file can probably be imported to a spreadsheet.

Thus, you should be able to graph the Voltage/UnitLength along the secondary for any primary to secondary configuration in the first instant the gap fires and from there determine if that is causing the racing arcs.

If one cleans the gaps, the initial di/dt in the primary coil my increase. Thus your gap cleaning may indeed have increased the change for secondary racing arcs.

I have always felt that the racing arc issue was a matter of pure transformer action. So maybe you can prove me right :D



At 11:53 PM 8/9/2006, you wrote:
Hi All,

Terry just put a writeup I did on my investigations using JavaTC into what may be a better primary concept. It is located at:


If the idea works out, it may allow for higher coupling without inducing racing arcs.

Please feel free to comment or criticize (hopefullly constructively) and If anyone is willing to try the idea, I would be willing to consult.

Gerry R