improving zvs driver

[x-rayLamp]

hey everyone, i just wanted to ask a few questions, and just want to know why my circuit doesn't work, and some help to make it work if it is possible in the first place,
as you might know the zvs mazzili driver isn't optimized for high frequencies because of the slow gate charging resistors, so pierna used a gate driver chip to eliminate this problem and it worked, but because i cant find the chip he used in my country's electronic retail stores, i decided to try and replace that chip with a totem pole driver
hoping that it would drive the gates faster,here is my take on this concept,
and if this wasn't achievable,i would like to hear your opinions on other equivalent chips that could replace the one he used in his driver
thanks in advance 
pierna's driver https://www.google.com/url?sa=t&source=web&rct=j&url=https://hackaday.io/project/162935-piernass-fast-zvs-mazzilli-driver&ved=2ahUKEwiDpIbHguP8AhVQQaQEHcY7BOQQFnoECA8QAQ&usg=AOvVaw1esn_c94hpCzYMpvQ3PTlH

[SteveN87]

I think your 100R feedback resistors should be connected to the bases of the driver transistors, not +5V.

[davekni]

Suggest looking at this post and forward in the thread, along with the two links in this post:
https://highvoltageforum.net/index.php?topic=528.msg16880#msg16880

Your circuit as drawn needs a few changes, but may work fine after that.  IGBTs are specified at 15Vge, so I recommend changing +12V to +15V.  No use for +5V.  Remove all four resistors to the left of the emitter-follower pairs.  Connect left side of 100-ohm resistors to bases as Steve recommended.  Add pull-up resistors to +15V on the right side of the 100-ohm resistors (node between diodes and 100-ohm resistors).

Will work best if +15V is applied before main +50V power.  The +15V power can start low-amplitude oscillation powered through pull-up resistors and diodes.  That will make full-power oscillation start more smoothly when +50V is turned on, avoiding the common high current start-up spike and subsequent voltage spike in oscillation amplitude.

[x-rayLamp]

  thanks! I appreciate the help, but i think i got something wrong  :'(

[davekni]

Great to see you using simulation.  That's the key to figuring out analog circuits.

thanks! I appreciate the help, but i think i got something wrong  :'(

Yes, the resistors are missing between diode anodes and +15V supply.

[x-rayLamp]

 i am sorry i am starting to get annoying, i really want this circuit to work
now i think i got everything correct, but the waveforms don't look as nice as they should be
i tried various pull up resistor values (form 100 to 10k)  and settled on 470Ω but that didn't appear to change anything

[davekni]

now i think i got everything correct, but the waveforms don't look as nice as they should be

Yes, your circuit looks good to me.

The default parameters for diodes and BJTs within SPICE tend to be problematic.  I'd recommend picking actual parts from the built-in libraries.  I'd suggest 2N2222 for the NPN and 2N2907 for the PNP.  For the diodes, pick a fast-recovery diode with sufficient voltage rating and low current (to avoid excess capacitance).  I think many LTSpice diode models do not simulate breakdown voltage.  So you could probably use 1N4148 even though it would have marginal voltage capability in real life.

Also might be good to simulate 15V supply ramping on first, then 30V supply, as I'd suggest for real life.

For a balanced circuit such as this, occasionally simulators will stay in the metastable state indefinitely, since they do not simulate natural thermal noise.  An initial conditions can fix that.  For example, start with a small current in L3 and the FET drain nodes at 0V.

[x-rayLamp]

it finally worked! , except that it did the opposite of what it was intended to do,the problem is the arcs became shorter and produced a screeching noise (the core is probably saturating), and mosfets got hot to the touch indicating that the gate waveforms aren't shaped well, unfortunately i dont have an oscilloscope to test the circuit, i even tried to directly drive a tesla coil at about 2.6MHz, the regular driver wasn't able to run it for more that 4seconds and the fets got boiling hot in seconds, and this driver wasn't any better
anyways i think i learned a few things from this project even though it haven't succeeded (that reminded me of how much I suck ad designing circuits), and if anyone wants to build an improved zvs driver I'd recommend them to just use a dedicated mosfet driver IC
and again thanks to everyone who helped me in this circuit 
sorry i couldn't put any videos of it running (16s video is 40mb, and the max allowed size is 5000kb)

[davekni]

it finally worked! , except that it did the opposite of what it was intended to do,the problem is the arcs became shorter and produced a screeching noise (the core is probably saturating), and mosfets got hot to the touch indicating that the gate waveforms aren't shaped well, unfortunately i dont have an oscilloscope to test the circuit

Yes, it is very difficult to figure out issues without a scope.  Could be some parasitic high-frequency oscillations due to wiring inductance.  I've found ZVS oscillators to be sensitive to inductance in the main current loop of resonant cap to FET drains and between FET sources.  If you want to experiment with this more, I'd suggest building on a sheet of copper-clad board (or with copper foil).  For the simplest build to get main path low-inductance, cut a single break in copper on one side (slot down middle).  Use other side for ground (FET sources) and cut side for FET drains.  Here's my example for a half-bridge, bit different than ZVS, but explains the general idea:
https://highvoltageforum.net/index.php?topic=1324.msg9795#msg9795

i even tried to directly drive a tesla coil at about 2.6MHz

2.6MHz is high, so requires lots of gate drive power.  Might be better off with resonant gate drive such as typical class-E circuits, though those tend to be finicky too.

sorry i couldn't put any videos of it running (16s video is 40mb, and the max allowed size is 5000kb)

I think most people post videos to youtube (or other such service) and add a link in their post to this forum.  Works for me.