Re: FET coil questions

["Tesla list" <tesla-at-pupman-dot-com>]

Original poster: "Terry Fritz" <twftesla-at-qwest-dot-net>

Hi Carl,

>At 05:12 PM 11/5/2001 -0500, you wrote: 
>
>This is a set of questions for the SSTC folks.  
>
>Background: I just lashed up a small coil that is driven from a 555 timer, 
>gate driver chip, and IRF740 FET with the body diode isolated with the 
>standard schottky and fast diode.  This coil draws about 100 watts and puts 
>out 6"+ sparks, with another 555 timer interrupting the main oscillator in a 
>staccato fashion to keep duty cycle low.  

WOW!!  That looks cool!

>
>Question #1:  Can I use an oscilloscope to make sure that the FET switching 
>transitions look good?  There does seem to be some high-frequency ringing 
>when I look at the signal recieved with the 'scope probe dangling near the 
>FET gate.  Does anyone know what a good (i. e. the FET is not in danger of 
>failing) signal should look like?

They should look basically the way you "want" them to look.  The tops and
bottoms may be a little wavy and all but the switching transitions should
be sharp and give a solid fast on/off.  You don't want it to be half/on or
anything like that.  You also have to watch out for transient spikes that
may blow the FET.  A dangling wire has a pretty messy impedance so it will
have it's own ringing and all that.  You really need a high-voltage and
high-bandwidth probe like the TEK5100.  The TEK P6015s are all over the
places like ebay.  They used to use freon so nobody wants them now.
However, they rarely leak and they will work fine empty too but at lower
voltage.

>
>Question #2:  What is the best method for measuring peak primary current, 
>and how much peak current should the IRF740 be subject to?  

There is only one way to measure high-frequency high-currents accurately
and simply.  Current monitors like, Pearson makes, have very wide bandwidths.  

http://www.pearsonelectronics-dot-com/Pages/Choosing_currrent_monitors.html

For these measurements, you need high-bandwidth, high-voltage, and
high-current probes.  The spikes and such need to be seen and measured
accurately.  Unfortunately, the simple homemade resistor dividers, current
shunts, etc. just won't work very well at all.  Of course the Tek probe is
$250 and the monitors start at $450 :-((  Used P6015 are common and the
current monitors show up on ebay now and then but they go for a hundred
bucks there too (cause "I" drive up the price trying to get them for "me"
:-)).  I see you have an .edu mail so maybe your school has this stuff.
With such toys measurements like this are easy!

http://hot-streamer-dot-com/temp/2001-04/Power.jpg
 
>I know the specs 
>say it will take 48 A or something on that order, but is it permissible to 
>run at 48 A peak on every pulse in this sort of application?  

Sure!  No problem.  If it were a life support or nuclear reactor situation
then you would want to worry but for a Tesla coil you can run them right up
to the limit with no problem.

>I don't know 
>the mechanism by which the FET would fail if the peak current is exceeded.  

The wire bonds melt and fatigue in pulsed applications but they will last
long enough for this purpose.

>Maybe it just burns up because the localized heat cannot be dealt with, 
>although the FET may be staying quite cool in general.

If the FETs are not getting too hot, your are 95% of the way there :-)

You may also want to look at IGBTs.  They are much like FETs but may be
cleaner and have even more power (I use 1200V 180A).  You may want to check
out the files at:

http://hot-streamer-dot-com/TeslaCoils/MyCoils/SolidStateCoil/

Other projects have put my SSgap project on hold till I get more time.

>
>Question #3:  I'd like to isolate the gate with a pulse transformer, but all 
>the ones I have made so far seem to decrease coil performance (so I am still 
>using direct gate drive from the driver chip).  True, I have no clue what 
>type of ferrite I wound the test transformers on.  In the future, should I 
>get a bead that is recommended for the HF range? MF range? I may have also 
>read something about pulse transformers not being able to handle pulse duty 
>cycles different from 50% very well.  Is there truth to this claim?  What 
>would be the reason?

You may want to use an IGBT driver chip to drive them with nice clean
signals too like the IR2118.  Chips like this usually need a resistor
between the output and the FET since the drive signals are so solid they
may otherwise damage the IGBTs!  They are made to drive big gate
capacitances on and off instantly.

Cheers,

	Terry


>
>Thanks for your help!
>
>Carl
>
>