[Prev][Next][Index][Thread]
Re: Oscillator circuit for solid-state TC
On Sat, 12 Oct 1996, Tesla List wrote:
> Yes I should have explained that the driver board I've put together uses
> dedicated mosfet driver chips - tc4422cpa. These have an output
I thought you had something like that. You had told it earlier :)
That's why I did emphasize such drivers will be a must with low
output current device like 3524. That is worth pointing out for
all the beginners out there who are eagerly soldering their boards
right now :)
> resitor to each gate and a pair of schottkey diodes to clamp the
> chip output to the 0v, 11v rails. They can be switched from CMOS
Uhum.. 11V rail? I figured all out like this: gate-source capasitance
is charged through the resistor you have between gate and driver.
Current flowing will be then (11V-Vgs)/R. When the cap has been
charged the current will decrease exponentially. I thinks that's
why most people go for a 15V drive. With irf740 Vgs of somewhere
between 6V and 10V will start to have little effect. However,
at high powers you certainly want to squeeze everything out and
have the fet as conducting as possible. And quickly!
> so the output capability of the pwm chip is not an issue. They aren't
> that expensive, 4 quid - thats 6 dollars for those without real money :)
And that's not expensive? Wow.. what if you blow a pair of fets,
say $3-4 each. If blown fets take the driver with them it will
add extra $12 there. That's why I like using low-cost pwm drivers
with driver onboard. It keeps the costs to a mininum is does not
cost that much to be replaced if something goes wrong.
> The plan is to drive 2 or more of these boards from the PWM. Not just
> to switch more power but also to use shorter more intense bursts -
> higher peak power - to move towards spark gap performance :)
:) Multiple boards feeding power syncronously to the same load
is certainly a good choise. Making a few low power boards can be
much easier than making one eg. 4kW board.
> I've redesigned the current limiting around an opto-isolator accross
> a resistor on the high side. I've bought three non-inductive
> resistors (0.22ohm film in a to220 case) - more expensive than
> the driver chips!!
Normally metal film or carbon mold resistors do exactly as well as
extremely low inductance resistors. Those types are low inductance
too. Just avoid *wire wound* resistors!
> It might even work - unless someone knows better.
It seems valid solution. However, in bridge type drivers it is
much easier to use a current transformer for sensing both currents
at the same time. Current transformer at high frequency will not
even need to be too large. Just take some small toroid and wound
something like 25 turns etc. Done. Easy. Cheap and reliable.
> A triple or quad would also be possible - if I get
> ambitious.
However, that does not solve the usual problem. I had lot of trouble
fitting all the turns and insulators on the core winding area. I never
managed to get all the turns I had wanted to. Feeding multiple
transformers with secundaries in series will help. Of couse it helps
to have bigger core becouse smaller number of primary turns will do.
However, heating problems are more easily solved in multiple
transformer solution.
--
We have phone numbers already, why would we need IP-numbers! -unknown person
Harri.Suomalainen-at-hut.fi - PGP key available by fingering haba-at-alpha.hut.fi