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

Re: DRSSTC-3 VCO drive test (long)

Original poster: Steve Ward <steve.ward@xxxxxxxxx>

Hi Colin,

> > you state you are using 10ohms on the gates! what are the more usual > figures then please?

I usually used about 5 ohms.  Right now i have no resistance and it
seems to be working fine.  You just have to make sure there is no
cross-conduction.  Ditching the resistor there gave me an extra 300nS
or so of less delay.

 im using 5ohms but must admit i havent scoped it all
> properly yet.
> perhaps i should check and see whats going on then.
> wow im not clamping my output of the gate drive at all just running 33v
> zener setup on the gates.
> what is the reason for clamping them? is it back emf from the gdt?

Exactly, the GDT can feed back into the driver chips.  Im sure the
drivers have diodes in them already, but it seems like a good idea to
add some robust diodes as well, just to be sure ;-).

> cheers
> colin heath

Anyway, i did some more testing today. Firstly, i did all i could to help the gate driver situation. Im stuck with about 300nS turn on and 200nS turn off delays. Oh well... But things are looking much better. I found that if my CT had much self capacitance at all that i could really screw up the phase relationship. So i tried making a new CT with far less turns, and only a single layer of winding... this seems to have helped! I also found that the switching eventually does get super close to ZCS when i crank up the power. I can tell because the switching noise on the current waveform starts *right* at zero :-). I was running primary feedback today without any problems at all :-) Had it running 400A in the primary and about 25" sparks to my target. Didnt push it too hard cause i didnt have my metal shielding to protect the lower level from the sparks... it would be pretty annoying to have another failure because of that again.

I noticed some interesting things about the primary feedback.  It
seems that with my current tuning, the primary current bounces back
and forth from the upper to lower modes, about 2-3 cycles in each
mode.  It produces some interesting "notch-like" steps in the primary
current.  I noticed this in pspice as well.  I guess Terry is running
his setup sorta like this too.  I also notice that when i get near my
maximum power, the primary current appears to stick with one pole
about half way through the 100uS burst.  I figure let it do whatever
it wants so long as its making sparks and the IGBTs arent exploding

My overcurrent detector is working like a charm still... but i wish i
would have used a 10T trimmer rather than a single turn... its a bit
hard to get it *just* right.  But im already designing a new PCB to
replace the old one, and the new PCB will have the overcurrent stuff
built in so i dont have 2 seperate boards on the coil.  I had to use a
little trick (Terry might want to listen up) to make the overcurrent
detect work properly.  Im using an LM311 comparator, and never thought
about the super high input impedance.  Noise was triggering the
detector at first.  Now i just put a 100 ohm resistor across the CTs
input to the "-" input.  This resistor acts as part of the CTs burden
resistor (10 ohms).  This is needed because the burden resistor is on
the OTHER side of the full wave bridge im using to get all + peaks.
Also, its a good idea to filter the heck out of the other input, i put
a 47uF tantalum right on the pins.  Seems quite stable now :-).

More work to be done... so little time!

Steve W