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

Re: [TCML] DRSSTC tuning at high power



Whoops, forgot about no attachments.  I'll try again.  Links to pictures are
at the bottom.

Hi Steve, I was hoping for a reply from you in particular

On Tue, May 17, 2011 at 5:16 PM, Steve Ward <steve.ward@xxxxxxxxx> wrote:

>
> This behavior is very much like what ive seen in many of my coils,
> particularly the big ones, but at the same time ive also managed to find
> tunings that, while maybe not having a primary notch, did not have the
> primary current blow up at the end (so just plateau for long enough).  So
> the basic phenomenon is that before breakout the secondary back-drives the
> primary (as the current phase shifts from zero to 180*, or is that -90 to
> 90?) and so the current goes back to zero (notch).  Once streamers provide
> an exit for the secondary energy, less and less gets "reflected" back to
> the
> primary after 1/k cycles, and instead you end up pumping the primary even
> harder.  The thing i havent quite come to grips with yet is why the
> performance can often be so pathetic despite the rather huge reactive power
> in the system.
>
> I think a critical measurement to make is secondary base current.  Id
> suggest watching that along with primary current as you play with tunings.
> I should take my own advice and do this next time i get a chance to run
> large DRSSTCs.  I think the base current is going to be a strong indicator
> of spark performance, though i realize the spark growth is not an
> instantaneous thing so talking purely about peak secondary amps is not good
> enough, you have to look at how long the current is sustained to feed the
> spark.

I've never really taken the time to look at secondary current.  Is it
actually possible to infer when breakout occurs on the topload from looking
at secondary current?  If so that would make it very useful.  Otherwise I'm
not sure what I'd really be looking for.

>  I havent got a good grasp on how this works yet, but at least with
> my small coils ive done some various tuning studies to look at the
> relationship between primary current, secondary current and spark
> performance, and there does seem to be some subtle "sweet spot" that
> maximizes I(s)/I(p), whether or not I(s) is at its maximum seems to be
> another issue.  Basically, the most efficient tuning may not be the one to
> make the biggest sparks, so there can exist less efficient tunings that
> consequently use more power and make bigger sparks.  Im using "efficient"
> loosely as the ratio of I(s)/I(p).
>
Is/Ip seems like a reasonable figure of merit, but only if you're not
getting premature streamers, right?  I've always assumed that you want to
build up as much charge in the topload as possible before breakout.  But if
a streamer breaks early, you'll see secondary current rise more quickly,
correct?  And wouldn't this give a false impression of good energy
transfer?  That is, you're dumping large amounts of power into the
secondary, but it will be going to little streamers instead of big arcs.

>
>
> How is the 10kW measured?  Whats the power supply input?  Is power factor
> considered?
>
We're using PFC supplies we've built specifically for this purpose.  Power
factor is generally >0.95 at decent power levels.  We're measuring input
current from the 240V line.  And the supply should be around 95% efficient,
so I think our power estimate is accurate.  But keep in mind that the 10KW
figure is at high repetition rates.  At around 100bps, we draw more like
3KW.

Have you noticed similar changing in the tuning or waveforms at high rep
rates?

>
> You have changed k?
>
We're mainly changing Lp when tuning (though k does change a big along with
it).  Right now our k is likely around 0.12 or 0.13.  I haven't really
played with it much because I assumed it wouldn't cause our waveforms to
look nicer, but rather just cause the energy transfer to happen faster or
slower (which isn't my main concern right now).

>
> Ive wondered about this, i think this can eat up >10% of the total power,
> id
> estimate up to 20% on some of my systems is lost in components (including
> the silicon).
>
At one point we had an aluminum screen RF ground laid under the coil, and
our H bridge assembly was sitting on top of it.  The cables leading from the
bridge to the primary were lying directly on top of the screen.  After a
running the coil for a while, we noticed that the parts of the cables lying
on the screen were getting very hot, so we lifted them up.  Apparently were
inducing tremendous eddy currents in the screen, which not only caused the
heating in the cables but also burned the wood underneath the screen.  We've
since solved that problem by propping up the cables, but we're still on the
lookout for other eddy current losses.  Hopefully we'll get a thermal camera
in at some point...
However I really doubt this is the issue.  None of my simulations show that
increasing primary resistance can cause the waveforms I'm seeing.  If
anything it should cause the peak Ip to get lower, but I'm seeing higher Ip
than a lossless system.

>
>
> >   3. Bad topload design.
>
>
> Seems unlikely to be the problem.
>
Why is that?  I was actually hoping it would be the topload, since we're
already planning to squish the rings into a toroid shape.

>
> >   4. Bad breakout point.  We're using a 12" long breakout point on the
> side
> >   of the topload.  Maybe it's breaking out too early?
> >
>
> Also seems unlikely, though i do find that i always prefer to use the
> shortest breakout point that reliably does what its supposed to.  Im not
> sure that the performance difference is due to lowering breakout voltage,
> because the hot streamer channel seems to persist between firings and
> re-ignite fairly early anyway.  I do have some suspicion that the spark
> starting in a higher electric field (closer to the toroid) might help, but
> this is just a guess.
>
This makes sense.  What bothers me about our topload/breakout is that even
with such a big, over exaggerated breakout point, we will still get many
streamers forming all over the topload (until the breakout point arcs to our
strike rod, at which point the other streamers mostly disappear).  Sometimes
we even get strike rail hits from other spots on the topload.  This says to
me that something isn't working right.

 Have you done much testing without a secondary coil in place?  This should
> allow for better observation of where the power is going.  I found my 7
> foot
> tall coils (about 8kW max input, 10 foot sparks) happily eat up 1500-2000W
> (real power measured from PFC to DC bus) with no secondary installed.  I
> was
> quite surprised by this result and i think most of that power is primary
> coil resistance.
>
We did no secondary tests long ago just for the sake of testing the driver
circuitry.  Not at high power though, and with a lower Q dummy load.

>
>  In my experiments with quasi-CW spark production, i found that my 360khz
> setup was *far* superior in converting primary amps to spark length than
> was
> a 280khz or 125khz resonator.  And not just some 10-15%, but it was better
> easily by a factor of 2!  So basically, same amount of amp-turns in the
> primary, but huge difference in spark output.  At the same time, these QCW
> sparks appear to have significantly different behavior than "transient
> mode"
> coils.  Some day i need to do similar testing with a regular DRSSTC and see
> if winding low frequency secondaries (to keep the silicon switching slower)
> is not shooting ourselves in the foot in order to make long sparks.  In the
> QCW case it certainly appeared that way.
>
That would be interesting.  Suffice to say, there are things going on beyond
the normal dual resonant circuit model.  It probably comes back to the old
problems of not being able to predict the load presented by the topload
(including streamers, not just its bulk capacitance).  Perhaps higher
frequency coils are just less susceptible to there effects.

>
> Oh, could you give some rough specs on your coil?
>
I'll do my best, but it's been a while since we built it:
Secondary:  ~2000 turns 24AWG with a 12.5" diameter, 324mH.
Topload:  54" diameter ring sphere, estimated 82pF capacitance (combined
with self capacitance of secondary)
Primary:  flat spiral coil made from 7/8" copper pipe, 6 turns total.
Innermost turn radius is 9.25", outermost radius is 16.45".  We're not using
the whole coil, and are tapping it at different places to get our tuning.
We're shooting for about 28.5uH and k=0.12
Primary cap:  0.947uF maxwell 35KV poly cap.

Pictures of coil (old, but pretty similar to what we have now):
http://imgur.com/8Xx86
http://imgur.com/G2A6B

Thanks,
-Mike

On Tue, May 17, 2011 at 5:16 PM, Steve Ward <steve.ward@xxxxxxxxx> wrote:

> Hi Mike,
>
> On Tue, May 17, 2011 at 12:28 PM, Michael Twieg <mdt24@xxxxxxxx> wrote:
>
> > Hello, for the last year or so I've been working with a large group on a
> > pair of large (>20KW) musical DRSSTCs.  So far we've been pretty
> > successful,
> > but I'm starting to think that our tuning is far from optimal, resulting
> in
> > poor spark lengths for our given power draw.  At low DC bus voltages
> (like
> > under 100V), I'm seeing nice quenching of the primary current, but as the
> > voltage increases, the dip in the current envelope starts to disappear
> and
> > peak current rises very quickly.  At around 400V, the quench is
> completely
> > gone, and the current envelope doesn't dip at all (instead it just kind
> of
> > plateaus before increasing again).
>
>
> This behavior is very much like what ive seen in many of my coils,
> particularly the big ones, but at the same time ive also managed to find
> tunings that, while maybe not having a primary notch, did not have the
> primary current blow up at the end (so just plateau for long enough).  So
> the basic phenomenon is that before breakout the secondary back-drives the
> primary (as the current phase shifts from zero to 180*, or is that -90 to
> 90?) and so the current goes back to zero (notch).  Once streamers provide
> an exit for the secondary energy, less and less gets "reflected" back to
> the
> primary after 1/k cycles, and instead you end up pumping the primary even
> harder.  The thing i havent quite come to grips with yet is why the
> performance can often be so pathetic despite the rather huge reactive power
> in the system.
>
> I think a critical measurement to make is secondary base current.  Id
> suggest watching that along with primary current as you play with tunings.
> I should take my own advice and do this next time i get a chance to run
> large DRSSTCs.  I think the base current is going to be a strong indicator
> of spark performance, though i realize the spark growth is not an
> instantaneous thing so talking purely about peak secondary amps is not good
> enough, you have to look at how long the current is sustained to feed the
> spark.  I havent got a good grasp on how this works yet, but at least with
> my small coils ive done some various tuning studies to look at the
> relationship between primary current, secondary current and spark
> performance, and there does seem to be some subtle "sweet spot" that
> maximizes I(s)/I(p), whether or not I(s) is at its maximum seems to be
> another issue.  Basically, the most efficient tuning may not be the one to
> make the biggest sparks, so there can exist less efficient tunings that
> consequently use more power and make bigger sparks.  Im using "efficient"
> loosely as the ratio of I(s)/I(p).
>
>
> >  Also of interest is that as repetition
> > rate goes up, the waveform becomes even worse (presumably because the
> > plasma
> > becomes denser and gives the topload a lower resistance load).  But
> > overall,
> > putting the bus voltage at 450V, and using a tone frequency of about
> 440Hz
> > gives consistent 7ft arcs, but the thing draws over 10KW and the peak Ip
> is
> > over 1500A, which seems like way too much.  Somehow we're not efficiently
> > transferring energy to the topload.
> >
>
> I think what you mean to say is "not efficiently transferring energy to the
> sparks"?
>
> How is the 10kW measured?  Whats the power supply input?  Is power factor
> considered?
>
>
> >
> > I've got several theories on what could be causing the issue:
> >
> >   1. Simply bad tuning of k and Lp.  I don't think this is really the
> case
> >   because we've played with the tuning a ton and nothing seems to improve
> > the
> >   Ip waveform or the power draw.  Even if the topload capacitance is
> > changing
> >   with arc size, we should have been able to compensate with tuning
> > changes.
> >
>
> You have changed k?
>
>
> >   2. Losses due to coil resistance, eddy currents, etc.
> >
>
> Ive wondered about this, i think this can eat up >10% of the total power,
> id
> estimate up to 20% on some of my systems is lost in components (including
> the silicon).
>
>
> >   3. Bad topload design.  Right now we have a shperical topload
> >   approximated with five aluminum rings (about 54" diameter).  We seem to
> > get
> >   a lot of corona at low power levels, but once we draw a solid arc most
> of
> >   the corona goes away.
> >
>
> Seems unlikely to be the problem.
>
>
> >   4. Bad breakout point.  We're using a 12" long breakout point on the
> side
> >   of the topload.  Maybe it's breaking out too early?
> >
>
> Also seems unlikely, though i do find that i always prefer to use the
> shortest breakout point that reliably does what its supposed to.  Im not
> sure that the performance difference is due to lowering breakout voltage,
> because the hot streamer channel seems to persist between firings and
> re-ignite fairly early anyway.  I do have some suspicion that the spark
> starting in a higher electric field (closer to the toroid) might help, but
> this is just a guess.
>
> The main issue I can't figure out is where most of the power is going.  Our
> > silicon barely heats up, the primary coil and capacitor get warm but
> > probably dissipate no more than 1KW total.  I have trouble believing that
> > we're actually putting ~10,000W into arcs, otherwise we should be getting
> > at
> > least ten feet.
> >
>
> Have you done much testing without a secondary coil in place?  This should
> allow for better observation of where the power is going.  I found my 7
> foot
> tall coils (about 8kW max input, 10 foot sparks) happily eat up 1500-2000W
> (real power measured from PFC to DC bus) with no secondary installed.  I
> was
> quite surprised by this result and i think most of that power is primary
> coil resistance.
>
>
> >
> > Any ideas?  Or maybe I'm expecting too much?
> >
>
> Well, i'll at least say that I can understand where you are coming from...
> my "Gigantor" coil appeared to be using 6000A in the primary and like 55kVA
> from the line just to make ~20 foot sparks, which is greatly out of line
> from what I hoped.  The thing that gets me is that some DRSSTCs seem to
> claim extraordinary efficiencies, and others just gobble up power and work
> so-so.  I suspect this is not largely an issue of losses in copper or
> silicon, but rather an issue of proper drive to the spark.  When we measure
> sparks only by their length, we might be making some un-fair comparisons,
> but other characteristics are hard to measure.  In any case, one
> optimization is certainly to make the longest sparks for a given amount of
> power consumed, and i think there is probably some issues to look at such
> as
> primary and secondary impedance and operating frequency, and definitely how
> quickly the spark energy is delivered.
>
> In my experiments with quasi-CW spark production, i found that my 360khz
> setup was *far* superior in converting primary amps to spark length than
> was
> a 280khz or 125khz resonator.  And not just some 10-15%, but it was better
> easily by a factor of 2!  So basically, same amount of amp-turns in the
> primary, but huge difference in spark output.  At the same time, these QCW
> sparks appear to have significantly different behavior than "transient
> mode"
> coils.  Some day i need to do similar testing with a regular DRSSTC and see
> if winding low frequency secondaries (to keep the silicon switching slower)
> is not shooting ourselves in the foot in order to make long sparks.  In the
> QCW case it certainly appeared that way.
>
> Oh, could you give some rough specs on your coil?
>
> Steve
>
>
> >
> > -Mike
> > _______________________________________________
> > Tesla mailing list
> > Tesla@xxxxxxxxxx
> > http://www.pupman.com/mailman/listinfo/tesla
> >
> _______________________________________________
> Tesla mailing list
> Tesla@xxxxxxxxxx
> http://www.pupman.com/mailman/listinfo/tesla
>
_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla