[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: Terry's DRSSTC -"different" H-drive functions...

Original poster: Terry Fritz <teslalist@xxxxxxxxxxxxxxxxxxxxxxx>

Hi Steve,

At 03:51 AM 1/13/2005, you wrote:
>If all the IGBTs are
>not conducting, then the primary Q is very low and the energy stored in the
>primary is very quickly being dissipated in the free-wheeling diodes in the

This is not quite right. If you turn all the IGBTs off, then the free-wheel
diodes act like a bridge rectifier and transfer the primary energy back into
the DC bus. It's not wasted, just recycled. It seems that about half the
energy in the primary at the time of gate drive stopping goes back to the DC
bus and the other half carries on to the secondary.

Yes indeed!! I was wrong there. The diodes are simply forward biased back to the buss rails... Models show I am only heating the diodes at 0.164 watts each. A little more really with switching losses, but trivial...

Of course it would be nice to make sure all this energy goes forwards
instead of back. I remember Jimmy Hynes originally planned to use the idea
of shorting two IGBTs on to make sure the energy all goes into the
secondary. But it's not possible if you use transformer drive to the gates.
Jason Judd tried the method with success. He uses optoisolated gate drive
ICs from Agilent, controlled by a PIC or suchlike, so he can have any
combination of IGBTs on at any time.

Cool!!! I will look into his work!! Opto gate drives like the TLP250 are considerably faster than Agilent's, but that might not be a big deal...

I am all for recycling. However, I have gone to a lot of trouble getting energy into the coil and I really don't want it back ;-)) It looks like 154 watts of power is being returned to the buss caps ( @200BPS ). I think shorting the two lower FETs is a good idea to keep the power in the coil. Pretty simple logic should do it...

I decided not to do it because I'm determined to use transformer coupled
drive (the reason being that it makes arm shoot-through incidents almost
impossible- a transformer can't turn two devices on at the same time) and
I'll just live with the sub-optimal use of energy.

Ok, I just slightly redesigned my gate drive to allow longer hang-on times, higher gate voltage, and allow full 300 volts on it for a longer time too without it getting the resistors too hot.


I have a new idea about the gate controller that will very easily dis-allow cross conduction and add dead times. Adding the freewheeling logic should be easy too. Not written up yet though... If it works well, maybe we can draw you away from GDTs ;-))

BTW - ALL my electronics can run from just +5 volts now!! I will have to get a 24VDC fan just to say I used the +-12 volts from the fancy power supply for something ;O)))

I reckon it's not such a big deal. If you use the "turn all IGBTs off"
method they have to deliver more energy over and above your desired bang
energy, to make up for the energy that rectifies back. So they will
dissipate more heat. If you short two on, then you don't need to deliver
this extra energy, but the two "on" IGBTs will still dissipate more heat
because of the extra current flowing in them. The freewheel diodes work
harder in the former case (they don't do anything at all in the latter apart
from catching random spikes) but I still think they will hardly heat
compared to the IGBTs.

Apparently, conducting the tail current in the diodes or on IGBTs presents less than a watt of heating... Not an issue...

The Antonio/Hynes style tunings can arrange it so there is practically no
energy in the primary when the gate drive ends, which avoids the problem
altogether. But probably at the cost of needing a bigger resonator for a
given spark length.

I will look into that next ;-)) But I want to be able to run just about any existing lower Fo coil off the box too.

In other news...  I updated the H-Bridge and Gate drive schematics:



Looks like the cost of my thing is coming in at $1328. If one takes out the "fancy stuff" like meters, box, computer controller, and other decadent items, then the cost of the real basic electronics is $275!! Buying extra parts and tossing a few "bad ideas" in the trash can have the cost for the whole project at about $1700.

Getting there...



Steve C.