Re: Solid State - air core driver??

On Sun, 5 May 1996, Tesla List wrote:
> Jim wrote:
> > watch out for voltage spikes!
> >from the resonant rise of the TC (300-400 as measured by malcolm
> >watts) divided only by your secondary/primary turns ratio 1500:18 =83.
> >So if you put in 75 volts, multiply it by 300 (coil Q) and then divide
> >that back down by 83 you will get 271 volts  across the coil (twice
> >that if you use a push-pull).

If you have a bridge design no spikes will appear. I have built those
and I used 350V rail for bridge made of IRF740's with no voltage
spike problems. If that kind of effect did happen there would be
still very easy methods for controlling it. Fortunately that does not
happen according to my experience.

> 540v will kill irf740's - at 150v single side operation has been very iffy.

Yap. However, when you have no leakage inductance spike problems 740's
will be fine for almost 400V rail operation.

> A full bridge I will leave till later. Thanks for the transformer details
> Harri - yes I I've enough information now to design one - that might
> even work!!

Full bridge can also be made when you first see a half bridge low power
prototype working well. Blowing up pair of fet at a time is much
cheaper than blowing up two pairs at a time :)

You should have lot of info for going to low power. At high power
like 1kW+ you will probably see more problems again. Up to around
500W or so it should be a plug-and-play stuff by now :)

>  I haven't got bifular wire but I have
> wound two 1.25mm wires carefully side by side.

Better use more wires with smaller diameter. I saw my design suffered
from using only two paralleled 0.7mm diameter wires. With a bunch of
0.25mm wires it is way better. Skin effect and proximity effect again..

> I'll keep the current limiting though in case it
> saturates.

If you use a bridge it will not saturate if properly designed (ie.
enough primary turns). Even then core heating will be the problem
first probably.

> the FET and ground which must be transfered
> to the gate via the internal FET capacitance.
> Presumably this voltage rise makes it preferable
> to measure the current using a current
> transformer, and the same signal can be used
> to drive a phase locked loop - thingy. OK that
> can go in the next one too!

Current transformer is a must unless you measure  voltage directly across
the resistors. FET's have voltage drops way too much..

Phase locking is another thing. I tried that with about 80%+ duty
cycle pwm and got into lot of trouble. Main problem was that the
device was oscillating between two different states. Much more
easier to take some ideas for eg. zero-current resonant power supplys
and make the cycle end always at zero current crossing. Then it will
be self tracking automatically too!

> Right I shall start to build - will it be serious spark
> or serious smoke wait for the next exciting 
> installment......

Smoke I gues :) I got only around 30cm tops with 1kW with my design.
However, it was not the state of the art design and I think there
is still lot more room for improvements.
"If the troubles are a bit less sever (say, someone merely hit your keyboard
 with an axe .."                  - Linux System Administrators guide V1.0
Harri.Suomalainen-at-hut.fi - PGP key available by fingering haba-at-alpha.hut.fi