Re: [TCML] A Question about Solid State Drivers

[Steve Ward <steve.ward@xxxxxxxxx>]

For fast modulation techniques, you are better off using "high-side" gate
driving techniques rather than a GDT.  For a high-side gate driver you need
1) an isolated power supply for the gate driver and 2) a method to provide
an isolated signal to the driver.  For #1 i often use small transformers and
a simple rectifier/capacitor and a linear regulator to derive 15V and 5V
power, for example.  Each high-side switch needs its own isolated
power/signal.  For #2 i found a chip by TI, part number ISO721 or ISO722 (i
cant recall which one).  Its a 5V level ultra-fast digital isolator.

The problem with using transformers to couple PWM'd signals is that they
like to ring a little bit during the dead band.  Also, depending on the PWM
technique used, you might lose the DC bias you were relying on for proper
gate drive.  What i mean is, the drive wave-form must be symmetric in order
to maintain the volt-seconds at 0 per cycle.  If the average volt-seconds
are not zero, then the waveform will first begin to climb up or down (to
balance the V-s) and then likely saturate as the condition persists.  If you
use a complementary PWM drive scheme to the transformer, it will be fine.

Anyway, since you are switching at 200khz, you should use more turns on the
transformer, like 10-15 (generally speaking, without really knowing what
core you have).

Either approach (high-side drive, or GDT) should work for you, its a design
decision you will have to make.  The GDT is generally simpler.

I dont know that i would use a PIC so intimately integrated with a SSTC.
Ive used PICs and AVRs in tesla projects, but usually safely shielded in a
metal box.  You may want to use some high speed isolators (like the ISO ones
i mentioned) or even better, high speed opto-couplers, between the PIC and
the rest of the SSTC electronics.  Use separate power supplies for the PIC
so you can keep it clean.  An aluminum box would be a good idea too.

Steve

On Mon, Dec 7, 2009 at 8:23 AM, Greg Morris <gbmorris@xxxxxxxxx> wrote:

> Hi Steve,
>
> Thanks for the advice!
>
> The reason I mentioned such a high "bandwidth" is because I intend to music
> modulate the coil. The coil will be nominally operating at around 200kHz,
> but I want to be able to modulate it down to a 10% duty cycle, and up to a
> 90% duty cycle (as well as 0% and 100%, of course, but not 95%, for
> example). And so for a 10% duty cycle (or 90%) at 200kHz, the system needs
> to be able to operate at a switching frequency of 10*200kHz = 2MHz.
>
> FYI: For the last stage of the digital controller, I plan to use the
> digital-out lines on a 40MHz PIC, which I think should work, as long as the
> code is extremely streamlined. For the first stage, however, I have some
> ideas about using an AC'97 codec, but I'm not clear on the details. Has
> anyone tried using anything like that? Any words of wisdom you would like
> to
> pass on?
>
> The intermediate stage of the digital controller is a both more involved,
> and will involve interfacing with the user, etc.
>
> Cheers,
> Greg
>
> 2009/12/6 Steve Ward <steve.ward@xxxxxxxxx>
>
> > Greg,
> >
> > The transformer is a "gate drive transformer".  Its construction is such
> > that it has exceptionally good high frequency response (i wont say high
> > bandwidth because it only works down to 10's of kHz generally, this is
> > design dependent of course).  Anyway, the combination of a ferrite core
> and
> > inter-twining of the windings gives very low leakage inductance, which is
> > what limits response time (self C does too, but its small compared to the
> > load C).  Make sure you fully understand the distinction between a
> > transformer's magnetizing inductance and leakage inductance.  Despite the
> > fact that the magnetizing inductance is large, it does NOT impact the
> high
> > frequency response of the signal being transmitted (it does slow down the
> > magnetizing current, which is un-related).
> >
> > By 2MHZ are you intending to switch at 2MHZ, or do you really only want
> > 2MHZ
> > for your rising/falling edges?  Careful about the relationship between a
> > "square" wave and its frequencies...
> >
> > If you want a square wave with an Fo of 2MHz and you want it to look
> > "pretty
> > square" you probably need something thats responsive to 20MHz or better.
> >  So
> > yes, this should be constructed on a small ferrite ring, something with
> > permeability in the 2000 to 5000 range and probably no more than 1" OD.
> >  Use
> > few turns, maybe only 3 or 4 turns.  This is a trade off between high
> > magnetizing current, but gains you low leakage inductance.  The external
> > wiring between driver, transformer and switch gates needs to be kept very
> > short.  Use twisted pair, or even coaxial cable if the wires *must* be
> > long.  If this is your first SSTC, id suggest aiming for a lower
> frequency
> > first because a lot of tricky problems come up with very high frequency
> > solid state drives.
> >
> > Steve
> >
> > On Sun, Dec 6, 2009 at 7:03 PM, Greg Morris <gbmorris@xxxxxxxxx> wrote:
> >
> > > I've been looking into interfacing a low-voltage, low-current control
> > > circuit with the relatively high-current power circuit of an SSTC. A
> lot
> > of
> > > schematics I've found:
> > > http://www.angelfire.com/electronic/cwillis/fetcoil2.html
> > > http://www.richieburnett.co.uk/sstc02.gif
> > > http://www.alansharp.co.uk/page6.htm
> > > etc.
> > >
> > > seem to use signal transformers between the controller and the power
> > > circuit, but it seems like you would need an incredibly small
> inductance
> > to
> > > be able to get a decent signal at Tesla Coil frequencies, especially
> when
> > > you consider that the secondary of the signal transformer is in series
> > with
> > > the gate of a power MOSFET or IGBT, which together act as a pretty
> > > effective
> > > low-pass filter. Am I missing something? What do you folks use? I am
> > trying
> > > to find a solution with a bandwidth in the area of 2MHz. If you have
> any
> > > suggestions, I would love to hear them.
> > >
> > > Thanks!
> > > Greg
> > >
> > > --
> > > Greg Weyrich Morris
> > > VP External - Engineering Undergraduate Society
> > > B.Sc. Electrical Engineering (Final Year)
> > > University of New Brunswick
> > > G.B.Morris@xxxxxx
> > > www.unb.ca/robotics
> > > _______________________________________________
> > > Tesla mailing list
> > > Tesla@xxxxxxxxxx
> > > http://www.pupman.com/mailman/listinfo/tesla
> > >
> > _______________________________________________
> > Tesla mailing list
> > Tesla@xxxxxxxxxx
> > http://www.pupman.com/mailman/listinfo/tesla
> >
>
>
>
> --
> Greg Weyrich Morris
> VP External - Engineering Undergraduate Society
> B.Sc. Electrical Engineering (Final Year)
> University of New Brunswick
> G.B.Morris@xxxxxx
> www.unb.ca/robotics
> _______________________________________________
> Tesla mailing list
> Tesla@xxxxxxxxxx
> http://www.pupman.com/mailman/listinfo/tesla
>
_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla