[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Fwd: [TCML] Magnifier topics
> Paul
>
> I would like to interject several additional points from Alex's 20kVA
> magnifier:
>
> 1. Quenching / dwell time is a critical component. When we first started
> operating this large machine, the magnifier was drawing 60A @ 280V
> with a 12 point ASRSG set as a 4point series gap; at this operational
> point approximately 8' (measured) streamer length could be obtained.
> I suggested we remove 1/2 of the electrodes on the rotary wheel and
> double the speed of the ASRSG (same BPS, 1/2 dwell time), no change
> in bang energy. Spark length increased to ~10' at same Vin and
> current
> demand dropped to ~ 40A. I know that kVA / ft of streamer is a
> "hokey"
> parameter, however it is difficult to ignore the below metrics:
>
> 16.8kVA/8' =2.1kVA/ft control
>
> 11.2kVA/10' = 1.1kVA/ft at 1/2 dwell time
>
> 2. With the machine in its present configuration, Alex would adjust the
> RSG
> for very low BPS (perhaps 40-60), streamer length is 12-18". By
> adjusting
> BPS _only_ to nearly burst speed of RSG, streamer length would grow to
> well beyond 10' and never showed any reduction is growth versus BPS by
> visual observation. I've seen similar results on a 1kVA machine,
> there the
> streamer length would slow due to transformer saturation. With 4 20kV
> 2.0kVA PT's, saturation occurs in the 20-22kVA range so is much less
> of a power input limitation. This strongly suggests that dwell/quench
> time
> for a given machine is likely a critical parameter.
>
> All experiments that Richard Hull, Alex and I have done with magnifiers
> strongly suggest us must "trap" the energy in the driver secondary, and
> decouple
> the "parasitic" load of the powering transformer/primary tank from the
> resonant
> secondary/tertiary. This requires super series rotary spark gaps, magnetic
> quenching, air-blast assist, and whatever edifice is available to force the
> gap
> to commutate off as soon as possible. Series gaps have additional loss, my
> comment to that is "oh well", I compensate by extremely high input primary
> voltage and smaller caps to boost bang energy.
>
>
>
> On Sun, Feb 14, 2010 at 5:48 AM, Paul Nicholson <tcml88@xxxxxxxxxxx>wrote:
>
>> So far I think the only documented example of multiple resonant
>> tuning of a TC is Antonio's 3:4:5 system described here,
>>
>>
>> http://www.coe.ufrj.br/~acmq/tesla/mag345.html<http://www.coe.ufrj.br/%7Eacmq/tesla/mag345.html>
>>
>> As a test of modelling software I set up a distributed model
>> of this system. Plugging in the dimensions of all the coils,
>> the 62pF 'transmission line' capacitance, and a guess at some
>> of the topload dimensions, reproduces the behaviour quite well.
>> I had to add an extra 2pF of top capacitance and I actually
>> needed 70pF on the t-line compared with Antonio's 62pF, in
>> order to get a reasonable tuning.
>>
>> Here is a comparison of lumped and distributed models, showing
>> the impedance response seen looking into the gap:-
>>
>> http://abelian.org/tssp/acmq345a.gif
>>
>> The red curve is from a lumped model using the component
>> values given by Antonio, plus a bit of resistance to simulate
>> a plausible loss factor. The green curve is the distributed
>> model, adjusted to match reasonably well by tweaking C2 and
>> C3 as mentioned above.
>>
>> From Antonio's web page you can see that the lumped model
>> gives a very good account and in this case the distributed
>> model doesn't have anything much to add.
>>
>> Here is an animation of the resonator voltages (800 kbyte,
>> 200 frames):
>>
>> http://www.abelian.netcom.co.uk/tssp/acmq345.anim.gif
>>
>> The 'resonator' in the top two graphs is the combined secondary
>> and tertiary, with the dividing line marked at position 29.
>>
>> Output voltage reaches a peak at 6.6uS which is frame 132.
>> If you can freeze it at that point, you see that there's
>> only a little residual current in the secondary and hardly any
>> secondary or transmission line voltage at all. Nor is there any
>> tertiary current - all the bang energy that survives the losses
>> is momentarily in the third coil's E-field. This animation
>> has a 10kV firing voltage and the peak output is about 210kV
>> (the lossless lumped model predicts 227kV).
>>
>> The step change in current between the top of the secondary
>> and base of the tertiary is due to the current diverted into
>> the added C2 - the 'transmission line' capacitance.
>>
>> You'll see some HF components excited in the secondary at
>> the beginning of the bang. These don't couple well into the
>> tertiary and are almost completely trapped in the secondary.
>>
>> It's fascinating to watch how the secondary coil drives the
>> tertiary resonance, whipping it up to a peak. Getting the
>> timing of that 'whiplash' correct is what multiple resonant
>> tuning is all about. You can see from the animation that it
>> really 'looks' right, but is not quite perfect because my model
>> is still slightly out of tune.
>>
>> If more overtones where brought in and tuned correctly,
>> it really would start to look like a whiplash at the peak,
>> with (momentarily) all the resonator voltage rise occurring at
>> the top end of the tertiary. Extra overtones can be tuned
>> by hanging capacitance onto the resonator at key points,
>> and/or splitting the resonator into yet more coils, each with
>> suitable reactance. What you end up with in the limit is a
>> pulse forming network involving stepped or graduated impedance
>> transformation. This could be described as a 'wide band' or
>> 'pulse' TC. It wouldn't be of any practical use because it
>> would be lossy and the final segment of the resonator has
>> to withstand the entire peak output voltage. However, the
>> principle probably crops up here and there in pulse forming
>> networks in power electronics.
>>
>> Note that the current distribution in both coils is almost
>> uniform, especially so the secondary. This means that the
>> effective inductance is very close to Ldc. Consequently the
>> voltage distribution in each coil is almost a linear rise
>> and this makes the Medhurst C valid for the equivalent shunt
>> capacitance. This enables Antonio's lumped model based on Ldc
>> and Medhurst to give such a good prediction of the resonant
>> frequencies.
>>
>> Apart from those little HF ripples, the distributed model
>> doesn't have a lot to add here, although it gives a nice
>> picture of the resonator in action.
>>
>> There are three resonant modes (almost correctly tuned) in
>> operation here. Separate animations of each:-
>>
>> 1/4 wave, 227kHz (232kHz measured, -0.4% error):
>> http://www.abelian.netcom.co.uk/tssp/acmq345-1.anim.gif
>>
>> 3/4 wave, 303kHz (307kHz measured, -1.3% error):
>> http://www.abelian.netcom.co.uk/tssp/acmq345-2.anim.gif
>>
>> 3/4 wave 383kHz (385kHz measured, -0.5% error):
>> http://www.abelian.netcom.co.uk/tssp/acmq345-3.anim.gif
>>
>> and just for fun, the next higher resonance, which is not
>> being tuned here,
>>
>> 5/4 wave 873kHz:
>> http://www.abelian.netcom.co.uk/tssp/acmq345-4.anim.gif
>>
>> This overtone makes 500V available at the topload at 20k ohms
>> reactive impedance.
>>
>> The errors of the distributed model (-0.4%, -1.3%, -0.5%)
>> could be improved if we took the trouble to measure all the
>> dimensions accurately and track down all the 'stray' wiring
>> capacitances, etc, but the lumped model with errors of (1.8%,
>> -1%, -1.3%) is already more than adequate for practical work.
>>
>> This 3:4:5 system really is a very good demonstration of
>> multiple resonant tuning. The added C2 is large compared with
>> the secondary coil's own shunt capacitance, which means that
>> Ldc and Medhurst C are good to use. The tertiary coil is not
>> quite so 'lumped' due to the relatively small topload but is
>> still very well described. These conditions are likely to
>> be present in most magnifiers and the well thought out set
>> of design equations and programs that Antonio provides are
>> completely justified for practical work.
>>
>> Hopefully the animations in this post will be helpful in
>> visualising what the multiple resonant tuning is aiming for.
>>
>> The challenge for the coiler is as follows: you build the
>> system according to the design equations and measure the three
>> resonant frequencies. They will not initially be correct
>> because the reactances can't be designed and built perfectly.
>> They need to be adjusted to the desired ratios.
>>
>> Once the coils are wound, the system has three tuning variables
>> to play with: topload height, added C2, and the primary tap.
>> Or perhaps four if coupling can also be adjusted.
>>
>> Actually I think it is quite a problem to know which to adjust
>> and by how much, in order to move the three frequencies into
>> a correct alignment. Perhaps it is possible to define a
>> tuning procedure? For example: C2 affects the 3rd frequency
>> more than the others; coupling mostly affects the distance
>> between 1st and 2nd (or is that 1st and 3rd); C3 affects the
>> 1st frequency more than the other two;
>>
>> If one can allow that it is only the ratios of the frequencies
>> that matter so that there are only two degrees of freedom,
>> then in theory only two of the reactances need to be tunable.
>> From considerations like these, can a tuning procedure be
>> defined?
>>
>> There is no quenching in these models, so after 6.6uS the
>> energy starts to transfer back to the primary. If anyone
>> is interested, I can animate the response with quenching,
>> and also with simulated discharges to ground from the topload.
>>
>> I'd like to post some more models of 3-coil systems, ones that
>> aren't so lumped. There's some interesting physics to be found
>> by looking at how the overtone mix, excited by the bang, changes
>> when the secondary is split. I hope to show that it is not
>> so much the higher k, but the extent to which the coupling is
>> concentrated onto a smaller fraction of the overall resonator,
>> that is responsible for higher overtone content.
>>
>>
>> Dex wrote:
>> > 1.Overall coupling in many magnifier systems
>> > is in 0.2+ range.That means magnifier is
>> > more efficient than a typical coil which
>> > usually has problems when coupling is that
>> > high.
>>
>> That's surely the simplest explanation for magnifier
>> performance. Simply a fast energy transfer being more
>> efficient.
>>
>>
>> > 2.Voltage shape of the outputs are not completely
>> > same even with the same overall coupling
>> > (perhaps some effect to spark is possible due to
>> > that difference)
>>
>> Another great possibility - the wave shape of the 3-coil is
>> just better for developing a good breakout.
>>
>> I think 'accidental' multi-mode tuning must come further down
>> the list of possibilities, and HF overtone heating effects at
>> the bottom. Having spent much time studying coil overtones,
>> I've still not found a use for them but I keep trying.
>>
>>
>> --
>> Paul Nicholson
>> --
>>
>>
>> _______________________________________________
>> Tesla mailing list
>> Tesla@xxxxxxxxxx
>> http://www.pupman.com/mailman/listinfo/tesla
>>
>
>
>
> --
> Dave Sharpe, TCBOR/HEAS
> Chesterfield, VA USA
>
> Sharpe's Axiom of Murphy's Law
> "Physics trumps opinion!"
>
--
Dave Sharpe, TCBOR/HEAS
Chesterfield, VA USA
Sharpe's Axiom of Murphy's Law
"Physics trumps opinion!"
_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla