Eldarion and Bruce's build of Bob's Energy Converter

[eldarion]

A follow-up to my previous post:
I have now cut a good 6" of lead length off of all the primary coils, and soldered everything up.  This had the added bonus of giving me several 6" pieces of that special silver wire to use for power interconnects.

Attached is a picture of the setup, as well as a pulse out of it.  This particular pulse was 150ns long.  There is no trace of the prior 2MHz resonance, which is a very good thing!

I do not know how to remove that last bit of ringing.  I still think it will mess up the effects, but maybe Bob can confirm or deny this?

Also, I wonder if there is any way to make the falling edge steeper.  I'll scope around and see if the MOSFET just isn't turning on fast enough for some reason (very likely).

Eldarion

[MarkSnoswell]

Ok -- now I can see your setup a few comments:

Put bypass caps on the Mosfets -- right up against the body of them. Use 4.7uf and 0.1uf ceramics on each device. You should also put ferrite beads on the V= and GND lines just before the bypass caps. I would also put them back at the other end as well to supress any HF pucked up by the leads.

You wont speed up the falling edge -- not unless you use Mosfet driven by complimentary signal which will give you a low impedance to pull up the signal fast.

... You can also just drive the coils directly with IXYS 414 or simillar drivers. These have very low output resistance and can deliver 10A at 30V. If you do this you must put TVS's (transient voltage supressors) on the outputs to protect the drivers from BEMF spikes.

You can also clean up the output by putting in a fast diode to supress the BEMF -- this will completly kill the ringing (it does for me).

Oh -- you will get faster times with IRF820 mosfets. Dont put these in untill you have everything working cleanly. To get the fastest switching time drive the gate with the maximum permissable -- 30V. If you do everything right your leading edge transition time can get doen to a few ns or so... thats what I read with a 150MHz scope which is the limiting factor with these fast times.

Mark.

[eldarion]

Hi Mark,

Thanks for the tips!  What I meant when I said falling edge [of the output waveform] was the rising edge of the gate drive, therefore the driven portion of the waveform.

So the MOSFETs' bypass capacitors would be soldered directly across the drain and source of each MOSFET?

Thanks!

Eldarion

EDIT: Just as I was fearing, there is a problem with the MOSFET drive signal.  Top trace is the gate drive, bottom is the switched end of the primary coil.  I have to think of a good way to get rid of this induction...

[MarkSnoswell]

I solder the gate driver back to back with the Mosfet -- there is perhaps only 3 mm lead length between them. If you do this then there is no ringing on the gate drive... if you dont then you are faced with having to put a resistor in the gate drive path to kill the ringing. Unfortunatly this will also slow the switching time.

[Earl]

Eldarion,

you are moving in the right direction.

The MOSFET driver, along with its multiple ceramic SMD capacitors between driver VDD and ground must be physically located at the MOSFET with roughly 3mm or less lead length.  There is simply NO other way to correctly do things.

In the drain lead going to the coil, I would insert a series ferrite RFI bead.  If you don't have a bead, use a 1/4W or 1/2W carbon resistor with a 4-turn coil wrapped around it's body and soldered to the resistor on each end.  Values of around 47 Ohms work well.  The parasitic low-Q choke or ferrite bead reduce the Q of any eventual HF/VHF/UHF resonant circuits to a value that stops HF oscillations as the transistor goes through a high-gain linear region when switching.

Earl