DRSSTC Discussion
Edited/Updated: October 24, 2004
Date : Sat, 16 Oct 2004 17:33:26 -0600. Subject : drsstc
Original poster: "Vladimiro Mazzilli"
<mazvla@iol.it>
Hi all in 1997 I developed my first sstc and knowing that was impossible to have
on the primary a voltage higher than the dc link in full bridge configuration, I
have built a self resonant primary oscillator, so I thought "I will have some kV
on the primary winding. But in that year I did not knew very well the tesla Coil
working principle. Was sufficient to tune the two resonant circuit, in fact, as I
was changing secondary with different dimension and form I obtained in output
the same 10" spark.
Reading in DRSSTC article I saw several bridge with IGBT brick up to 600A.
I work daily with these brick and I noted that the type used or in plan to use
is not the better solution. In1993 we developed 30kHz resonant converter from 10
to30kW using the current generation of that year.
In this moment in europe Semikron company has developed a new generation of
short tail IGBT. the marking is for example skm400gb125 where the last
number identify the family. The family 5 is suited to work in square wave >20kHz
and in resonant mode >100kHz, skm400gb125 is a brick 400A at 80° half bridge
1200V fast mode IGBT.
Regards, V.Mazzilli.
Date : Mon, 18 Oct 2004 08:00:42 -0600. Subject : RE: drsstc
Original poster: "Steve Conner"
>I work daily with these brick and I noted that the type used or in plan to
use is not the better solution.
I know this. But builders of big coils tend to choose their IGBTs based on what
they can get cheap on the surplus market. They are just too expensive to buy
brand new. So we make do with the older, slower bricks, and design
the coils to have the lowest resonant frequency possible. Steve Ward's ISSTC II
ran at 40kHz (iirc?) for example.
It helps that we are running in pulsed mode with a low duty cycle, as the
devices can tolerate much higher switching losses than they could in continuous
running.
For smaller coils, where the cost of new silicon isn't prohibitive, and it's
hard to get a low resonant frequency, we use the latest fastest IGBTs like the
Fairchild SMPS I/II series.
>In this moment in europe Semikron company has developed a new generation of
short tail IGBT.
I've been looking out for the Semikron '5' devices but they probably won't be
turning up on eBay for another few years yet :( I picked up some of the '3'
series ones a couple of months ago. They seem fairly fast, about equal to
Powerex 'H' series. Steve C.
Date : Mon, 18 Oct 2004 11:09:55 -0600. Subject : R: drsstc
Original poster: "Vladimiro Mazzilli"
<V.Mazzilli@aros.it>
Now I have as sample (for UPS, the field where I actually work, they have a vce
sat too high and I've to discard them) four half bridge brick 400A @ 80°C, from
Semikron series '8' soft punch through technology, that are not
fast like '5' but not bad and I think to start with these devices. I've selected
also 4 CM600HU24F (mitsubishi-powerex) for a full bridge tentative but they have
a lot of gate capacitance 230nF max without Miller effect, 6600nC total gate
charge typical, and I've also the possibility to test two CM1400DU24NF but in
this case too there are problem with the total gate charge.
>I work daily with these brick and I noted that the type used or in plan to
use is not the better solution.
I know this. But builders of big coils tend to choose their IGBTs based on what
they can get cheap on the surplus market. They are just too expensive to buy
brand new. So we make do with the older, slower bricks, and design
the coils to have the lowest resonant frequency possible. Steve Ward's ISSTC II
ran at 40kHz (iirc?) for example.
It helps that we are running in pulsed mode with a low duty cycle, as the
devices can tolerate much higher switching losses than they could in continuous
running.
For smaller coils, where the cost of new silicon isn't prohibitive, and it's
hard to get a low resonant frequency, we use the latest fastest IGBTs like the
Fairchild SMPS I/II series.
>In this moment in europe Semikron company has developed a new generation of
short tail IGBT.
I've been looking out for the Semikron '5' devices but they probably won't be
turning up on eBay for another few years yet :( I picked up some of the '3'
series ones a couple of months ago. They seem fairly fast, about equal to
Powerex 'H' series. Steve C.
Date : Wed, 20 Oct 2004 21:47:32 -0600. Subject : RE: drsstc
Original poster: "Rajesh Seenivasan"
<rajeshkvs@hotmail.com>
Dear Mr.Steve,
I am trying to design tesla coil and induction heater at home.
I am trying to use that in my design. I have a doubt in my design. I
designed a CW
H-Bridge inverter using IGBT bricks. Now, I want to connect a inductance
(tesla or induction heating transformer) and a capacitance (capacitor bank)
across
the output of two legs of H Bridge inverter. L and C are in series.
So, the resonant freq is given by,
f = 1/( 2*pi*sqrt(LC) )
Now, my doubt is:
Can I run this setup at exactly resonant frequency ? I ask this question
because i studied that
at resonant frequency, the series LC circuit's impedence becomes ZERO. So,
if L and C are
connected across the outputs of two legs of HBridge, will it short both the
output together
because of Zero impedance?
In other words, how do I calculate the current through the series LC
circuit at resonant frequency ?
Another doubt:
In one of the message, Mr.Vladimiro said:
"The family (Semikron) 5 is suited to work in square wave >20kHz and in resonant
mode >100kHz". Lets consider the resonant frequency is 60kHz. Can I run
the setup at a frequency more that resonant frequency, say 75kHz to make sure
that current drawn is not huge (say, 30A) ? Will the IGBT be damaged in
this case?
Please clarify my doubt. Thanks in advance. Rajesh.
Date : Thu, 21 Oct 2004 17:17:04 -0600. Subject : Re: drsstc
Original poster: "Bob (R.A.) Jones"
Subject: RE: drsstc
> Original poster: "Jan Wagner"
<jwagner@cc.hut.fi>
>
> Here's a paper from IEEE that has some calculations on power and current, that you could also use as a starting
>point: http://www.hut.fi/~jwagner/electr/tmp/01296120.pdf
I don't understand this paper. A 1.5kV about 100uF cap that dissipates almost most 0.5MW at 5kHz at 1.5kVpk
that's 6,000A.
What are they using, salt water bottle caps with the sea water pumped thru to
cool them!! LOL
Perhaps they don't know about the Panasonic caps or I missed something.
Just to stay on topic it would make a large SSTC with 1MW cont. output. A
curious configuration too parallel tuned circuit with an inductor in the DC
power line.
I checked the paper again. It does have the right order for the R of the C but
the dissipation in the cap looks orders too high. Bob.
Date : Thu, 21 Oct 2004 07:37:31 -0600. Subject : RE: drsstc
Original poster: "Jan Wagner"
Hi, just thought I'd comment as well...
On Wed, 20 Oct 2004, Tesla list wrote:
>Original poster: "Rajesh Seenivasan"
<rajeshkvs@hotmail.com> I am trying to design tesla coil and induction heater
>at home. I am trying to use that in my design. I have a doubt in my design. I designed a CW H-Bridge inverter using
>IGBT bricks. Now, I want to connect a inductance (tesla or induction heating transformer) and a capacitance
>(capacitor bank) across the output of two legs of H Bridge inverter. L and C are in series. So, the resonant freq is
>given by, f = 1/( 2*pi*sqrt(LC) )
I've an old Excel design sheet that could give you some ideas on some of the
calculations, even though the sheet is for a slightly different subject:
http://www.hut.fi/~jwagner/ele-tmp/LLC-resonant-welder-designsheet.xls
(but I haven't checked yet if it's up to date, and there _might_ be some errors
there...)
For induction heating you might want to use a parallel-resonant LC circuit to
get the current (and thus magnetic field) increased, instead of the voltage
(electric field). Inductive heating works with induction, i.e., magnetic field.
Here's a paper from IEEE that has some calculations on power and current, that
you could also use as a starting point:
http://www.hut.fi/~jwagner/electr/tmp/01296120.pdf
>Now, my doubt is: Can I run this setup at exactly resonant frequency ?
Yes.
How well it runs, or when it explodes, depends on the quality factor of your
series resonant circuit and how much the IGBT modules can handle. The circuit
has enough resistive and radiative losses that you don't actually get zero
impedance. Of course even this impedance can still be too low for the IGBT's to
survive long runs...
What you can do though instead of boring maths ;-) is to just plug in the LC
circuit and a load an run at the resonant freq, but start with a really small
input voltage to your IGBT system. Slowly increase the voltage, and monitor the
input current. When the current seems "hmm, this might be too much", then
_don't_ increase the input voltage further, okay? ;-))
>I ask this question because i studied that at resonant frequency, the series LC circuit's impedance becomes ZERO. So,
>if L and C are connected across the outputs of two legs of HBridge, will it short both the output together because of
>Zero impedence? In other words, how do I calculate the current through the series LC circuit at resonant frequency ?
Can't remember right now how one would calculate the max tank current in the
steady state, but it was via the Q (quality factor, Q=2pi*f*L/R) and then input
power considerations. Something like "P=Q^2/2R=UI" or there you
go ;) Maybe Google will help, "lcr tank current", "rlc tank current", ..., etc.
Or that electronics-basics book that I don't have at hand right now. ;o)
>Another doubt:
>In one of the message, Mr.Vladimiro said: "The family (Semikron) 5 is suited to work in square wave >20kHz and in
>resonant mode >100kHz". Lets consider the resonant frequency is 60kHz. Can I run the setup at a frequency more that
>resonant frequency, say 75kHz to make sure that current drawn is not huge (say, 30A) ?
Yes, though you could also go _lower_ in freq. The circuit will look a bit more
capacitive than resistive then (vs inductive if you go higher freq), but, who
cares ;-)) Power draw will drop, in either scenario.
By the way, if the IGBT module doesn't already contain internal
fast&soft-recovery freewheeling diodes then you'll have to add some external
ones. cheers, Jan.
Date : Fri, 22 Oct 2004 10:11:01 -0600. Subject : Re: drsstc
Original poster: "robert heidlebaugh"
Just how large is that 100 ufd 1.5 Kv capacitor ? My o.5 ufd 30 Kv
capacitors are 3 cu ft in size. Robert H.
Date : Fri, 22 Oct 2004 17:28:21 -0600. Subject : Re: drsstc
Original poster: "Bob (R.A.) Jones"
The size of caps is approximately proportional to their energy capacity for the
same type of cap.
So as a comparison, your cap is 225J the other cap is 112.5J. So it would
be approximately half the size of yours.
That assumes yours is a rated at 30kVpk AC @ 5kHz.
Actually I think your cap is large for its rating. Is it a high rep pulse cap?
I have a 1.5cu ft 27uf 5kV ie 337J but its low rep low peak current. Bob.
Date : Fri, 22 Oct 2004 21:15:37 -0600. Subject : Re: drsstc
Original poster: "David Rieben"
<drieben@midsouth.rr.com>
Bob(s):^),
The latest capacitor technology has caps of much smaller volume/energy ratios
than either of the caps that you mention. Defibrillators have a capacitor
of approx. 5.2 kV at around 35 to 40 uF rating and they are not much larger than
say two combined microwave oven filter caps. I suppose their peak current rating
doesn't have to be that great since they're designed to discharge across a
relatively high resistance - human chest :^O Even with the conductive gel
that they apply to the skin before shocking, the resistance would still probably
be in the dozens, if not hundreds of
Ohms. David Rieben.
Date : Sat, 23 Oct 2004 17:38:41 -0600. Subject : Re: drsstc
Original poster: "Bob (R.A.) Jones"
Subject: Re: drsstc
Hi David,
That's very very impressive 540J in the volume of 2 MO caps WOW. That's
comparable to an electrolytic cap for energy density. It would require more than
1500 Panasonic 0.15uF/2kV to do that. But as its has been said before its one
thing for a cap to have to charge for a few seconds a few times a day. and
its a completely different duty to operate continuously.
Seriously off topic (sorry Terry) but what the h$## are they trying to do to the
patient 540J into his chest!!
Does it double as an elephant defibrillator too!! Bob.
Date : Sat, 23 Oct 2004 17:40:46 -0600. Subject : Re: drsstc
Original poster: "robert heidlebaugh"
Bob: My capacitors were designed for high current and rep rate like an induction
furnace. The contact electrodes are 6 inch across and have a 5/8 bolt mount on
each plate. My two larger caps are 60 Kv energy discharge capacitors 5.99 uF and
weigh about 70 lbs each. Robert H.