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

RE: [TCML] Chokes and Terry filters

Hi Terry,

It's not just when the safety gap fires; every time the main gap (in parallel with the safety gaps) fires, it will set an R-C-L filter into oscillation.

I recall that when I was simulating various filter circuits, I found that 1000 Ohms was not sufficient to suppress oscillation of an R-C-L filter.  It wasn't until an R value in the tens of KOhm was used that ringing was squelched, but losses in such a high value resistor would be unacceptable. At that time I was considering inductors in the several-millihenry range, so I'm not certain what the impact of using a much smaller inductor would be.

I think that if the inductor was on the NST-side of the Terry filter (I'm not sure what side is input vs. output), it would be a don't-care as far as oscillations go, but I think DC was suggesting putting them on the spark gap side.  If the choke is between the Terry filter and the main spark gap, each time the gap fires, the charged bypass caps will resonate with the choke.  But even if the advice was to put them on the NST-side, it is given in such loose terms that some folks are sure to conclude that if DC sees a benefit to using one, adding more will surely be better.  If there's no proven benefit to using chokes beyond "it probably does no harm if it is done in a very specific way", I think it's best to simply say don't use chokes, period.  And I think it's healthy to review this occasionally!

Regards, Gary Lau

> -----Original Message-----
> From: tesla-bounces@xxxxxxxxxx [mailto:tesla-bounces@xxxxxxxxxx] On
> Behalf Of piranha
> Sent: Monday, June 16, 2008 10:52 PM
> To: Tesla Coil Mailing List
> Subject: Re: [TCML] Chokes and Terry filters
> The 1000 ohm resistors of the filter will break the Q of an inductance
> on the output up to 500uH going back to the NST (the resistors are
> considerably inductive themselves). The danger of adding inductance
> though is when the safety gap fires, the added inductance and the
> primary caps oscillate. Just shorting the gap with added inductance
> basically fires the coil at an off frequency which might stimulate
> racing arcs or something odd. But typically the added inductance is so
> relatively low that it does not matter much.
> If inductors are place on the output of the filter, the filter will
> basically turn them into a "don't care" as far as the NST is concerned.
> If the inductors are used with just a standard NST alone, then there are
> many elements and parasitics to consider. In general at very high
> frequencies, the NST will have such a high impedance that the relatively
> low impedance inductor will not have much effect. Of course, just like a
> secondary coil, the added inductor will have it's own resonant frequency
> after which it will look like a capacitor. You are not going to stop
> much above 10MHz in general. But again unless you hit a bad frequency,
> the added inductor is probably not going to do much. Although a bit too
> simplistic, the 3000H inductance of the NST combined with a say 100uH
> inductor give a voltage division delivering 0.99999996667 of the voltage
> to the NST anyway :D
> The real danger is when you start adding serious RLC elements to a
> filter network in there and cause all kinds of giant secondary
> oscillations. But the two instance were are talking of here will not
> normally do that. Those big hot resistors fix a lot of sins ;) Ideally
> you would do twisted pair on the leads. But at our voltages the corona
> discharge is too enormous to do that wonderful trick. Works good at 12V
> 300 amps at 2.5MHz though :D
> So do the added inductors do any good. Not really, but if they make you
> feel better they will not do much harm either. Just don't go getting too
> creative and making a bunch of secondary resonant circuits. The source
> impedance (gap) is very low and the load impedance (NST) is very high.
> You need some serious shunt elements in there to stop anything, like the
> filter's normal RC. You probably could do some common or differential
> mode filtering with the legs going to magnetically coupled added
> inductors. But that is starting to get complicated :)
> Terry
> Lau, Gary wrote:
> > That rapid current rise/fall times is a source of NST stress and failure is
> something I've not previously heard.  Isn't it simply having too high a voltage that
> causes corona, ionization, and insulation breakdown?  I can't imagine how current
> changing rapidly in a conductor can degrade surrounding insulation.  Assuming that
> adding inductance actually does retard current rise times, given that an NST
> secondary winding measures in the thousands of Henries, it doesn't seem that
> adding a couple microhenries would make any difference.
> >
> > This is neither here nor there, but isn't the point of a Marx generator to generate
> very fast rise times?  I thought pains are taken to _reduce_ stray inductance
> towards that end.  I agree that adding inductance would slow down the voltage rise
> times; it just doesn't seem like a desirable thing.
> >
> > Regards, Gary Lau
> > MA, USA
> >
> >
> >
> >
> >> -----Original Message-----
> >> From: tesla-bounces@xxxxxxxxxx [mailto:tesla-bounces@xxxxxxxxxx] On
> >> Behalf Of DC Cox
> >> Sent: Monday, June 16, 2008 4:15 PM
> >> To: Tesla Coil Mailing List
> >> Subject: Re: [TCML] Chokes and Terry filters
> >>
> >> Certainly no new science in using RF chokes to stretch waveforms across the
> >> time dimension.  It's been in ham radio books since the 1915 era.
> >>
> >> A series inductance serves to limit the peak current in a waveform.  As the
> >> rapid risetime in current increases with a typical 250 kHZ coil, the current
> >> traveling thru the inductor MUST produce a magnetic field which tends to
> >> delay/resist the flow of peak current in a circuit.  This stretches out the
> >> pulse risetime and "chokes" off the rapid pulse by stretching it in time
> >> (x-axis).
> >>
> >> This procedure is very common with Marx Impulse Lightning Generators in
> >> which engineers want to slow down the rapid rise time of a capacitor
> >> discharge.  Any resistor or inductor this effect ---  that's why companies
> >> such as Resonance Research Corp and HiPotronics use resistors and inductors
> >> in Marx generators -- they slow the rapid nanosecond risetime of energy
> >> discharge caps down to a typical 1.5-50 waveform (1.5 uS with 50 uSec decay
> >> time) and 200-2000 uS switching surge risetime.
> >>
> >> In a Tesla coil circuit the rapid oscillations (typical 200-250 KHZ) must be
> >> retarded BEFORE they can reach the xmfr windings.  Terry filters work well,
> >> however, for over 50 years I have always just used a nice air core choke to
> >> produce the same effects.  Now, I use both a Terry filter and a pair of air
> >> core chokes to slow down rapid transient risetimes which can be destructive
> >> if reflected back into the xmfr windings.
> >>
> >> In over 50 years of which 40 of them I just used an air core choke, I've
> >> never lost a NST.  That speaks for itself.
> >>
> >> I have the empirical data to support my claims but unfortunately, due to
> >> business time constraints, just don't have time to type it all up.
> >>
> >> Dr. Resonance
> >>
> >>
> >>
> >>
> >>
> >>
> >> On Mon, Jun 16, 2008 at 8:43 AM, Lau, Gary <Gary.Lau@xxxxxx> wrote:
> >>
> >>
> >>> DC, as long as you suggest using chokes or inductors in protection
> >>> networks, I'm going to have to refute that.  This is IMO a Tesla coil myth
> >>> that needs to be put to bed, so here goes again, with deliberate clarity and
> >>> detail.
> >>>
> >>> 1.  Chokes or inductors cannot "clip" anything.  Clipping suggests a
> >>> non-linear circuit element, like a spark gap, MOV, or zener diode.
> >>>  Inductors are linear components.  The MOV's and safety gaps in a Terry
> >>> filter don't require any additional components to clip excessive or nasty
> >>> voltages.  You would be hard pressed to find an MOV application note
> >>> suggesting the addition of chokes.
> >>>
> >>> 2.  Adding an inductor in series with the resistors in a Terry filter turns
> >>> the R-C filter into an R-C-L filter.  This will only create a new L-C
> >>> resonant circuit with the filter caps and expose the NST to yet another
> >>> source of RF.
> >>>
> >>> 3. You describe the reason for using chokes in very vague terms, without
> >>> ever actually defining the nature and source of the nasty thing that needs
> >>> to be filtered out.  Allow me.  I believe there are two things that need to
> >>> be addressed.
> >>>
> >>> 3a) Low frequency (60/120 Hz) high voltage excursions.  These can be caused
> >>> by many things:  A static gap that is set too wide, a mis-wired (i.e. open)
> >>> main gap, an RSG that is turning too slowly and allows mains resonant rise,
> >>> a sync RSG who's phase is mis-adjusted and misses a firing, a sync RSG with
> >>> a too-small cap value, etc.  The properly set safety gap of a Terry filter
> >>> is capable of shunting such excursions without the need for inductors.  The
> >>> MOV's do the same thing; they're just there as a tamper-proof Plan "B" in
> >>> the event that the safety gap was not properly set.
> >>>
> >>> 3b) There is a seldom discussed source of VHF (in the tens of MHz) voltage
> >>> spikes that occur every time the main gap stops conducting, that produces a
> >>> transient of 2X the peak capacitor voltage.  The full explanation is lengthy
> >>> and is detailed on my web site at
> >>> http://www.laushaus.com/tesla/protection.htm  The good news is that these
> >>> transients are low in energy content and are easily attenuated by the R-C
> >>> filter in the Terry filter.  Again, adding an inductor is not needed.
> >>>
> >>> If there is some other source of nasty things that need to be filtered out
> >>> that I have overlooked, please bring it to my attention.  If you use the
> >>> word "kickback", please define exactly what that means - what is its source,
> >>> what determines its frequency and peak voltage, etc.  Sound engineering
> >>> practices require sound engineering analyses to back them up.  Otherwise I
> >>> consider it to be only so much hand waving.  Empirical data in the absence
> >>> of rigorous circuit analysis can be valid, but there needs to be a
> >>> controlled study of failures in IDENTICAL coils with and without chokes,
> >>> before any conclusions can be reached.
> >>>
> >>> Regards, Gary Lau
> >>> MA, USA
> >>>
> >>>
> >>>
> >>>> -----Original Message-----
> >>>> From: tesla-bounces@xxxxxxxxxx [mailto:tesla-bounces@xxxxxxxxxx] On
> >>>> Behalf Of DC Cox
> >>>> Sent: Sunday, June 15, 2008 4:52 PM
> >>>> To: Tesla Coil Mailing List
> >>>> Subject: Re: [TCML] Terry Filter Caps Suffering From Early Death
> >>>>
> >>>> Yes, contact me off-list and I will email the data to you.
> >>>>
> >>>> I use a pair of inductance chokes next to the output going over to the
> >>>>
> >>> spark
> >>>
> >>>> gap.  This tends to clip the most nasty of the spikes allowing the Terry
> >>>> filter to despike the rest of them.
> >>>>
> >>>> I also use one additional cap in each series leg of the Terry filter to
> >>>> raise the overall potential value of the cap string.
> >>>>
> >>>> Dr. Resonance
> >>>>
> >>> _______________________________________________
> >>> Tesla mailing list
> >>> Tesla@xxxxxxxxxxxxxx
> >>> http://www.pupman.com/mailman/listinfo/tesla
> >>>
> >>>
> >> _______________________________________________
> >> Tesla mailing list
> >> Tesla@xxxxxxxxxxxxxx
> >> http://www.pupman.com/mailman/listinfo/tesla
> >>
> > _______________________________________________
> > Tesla mailing list
> > Tesla@xxxxxxxxxxxxxx
> > http://www.pupman.com/mailman/listinfo/tesla
> >
> >
> >
> _______________________________________________
> Tesla mailing list
> Tesla@xxxxxxxxxxxxxx
> http://www.pupman.com/mailman/listinfo/tesla
Tesla mailing list