You can use either 494 or 3525, botch chips are fine. Correctly wound gate driving transformer will have negligible leakage inductance and will provide pretty hard turn of currents by itself. When driving the transformer with an H bridge, as you have likely thought of in the image above, the Q1 and Q2 transistors are not really required.
Note also the driving transformer inductance needs to be calculated so that the transformer driver can cope with the magnetization current. Poor-man's solution to that is using just three PE insulated solid core wires, but I would rather go for teflon, to be a bit more chilled. TEX-F wire would be the best option, however sourcing that is not easy and is quite expensive. (bare enamel (magnet) wire is NOT enough). Make sure you use a suitable high permeability ferrite core, wound tri-filar (three twisted wires at once), and use wires with proper amount of insulation strength. Transformer based gate driving is pretty good technique and very robust one too. If you are designing a power supply, doing it properly from the beginning can help a lot. Ditching the IR2153 is definitely a smart move. And it is the last thing I would go after here. Not ideal, but on a clean conformally coated board, is actually quite good. Proper trace routing to minimize trace inductance is also very important. Then, I use an ES1D diode, also to ground, to prevent the midpoint from going below -1 V or so. I use something like 4700 pF C0G capacitor (C0G to get LOW ESR and therefore reduce self-heating) and a 10 Ohm resistor in series, to ground. How did I fix it? Well, first, I put in an RC snubber to slow down the voltage swing at the midpoint. I was able to capture something like -7 V across the body diode for several microseconds, without ANY SIGN of the body diode starting to turn on. Now, you say, "what about the body diode?" Well, I did measurements, and the FET body diodes are glacially slow to turn on. This current is way too large, and destroys the driver chip, in the process, turning both transistors on at the same time and blowing everything out. Current is flowing out through the inductance, and without the low-side transistor on, the current ends up flowing out through pin VS of the driver. Then, the high-side transistor is shut off, but the low-side has not turned on, yet. So, here's the scenario : The high-side transistor is on, sending current out from the midpoint of the two transistors to the inductive load. One of the things I found out was that the inductance of the load causes currents to flow that the driver cannot handle.
I make some PWM servo amps that have similar half-bridge stages, and use IR gate drivers in them. With just one cap, the design is cheaper (in case you need to buy those large caps) and the loop inductance and ESR is lower. Use only one rated at 400V and tie the other end of the transformer to a capacitive divider. EDIT: Also, if you do not need dual input voltage range, there is no point in using two of those electrolytic caps in series. There is no way around it with this simple IC. (something in the ballpark of 1uF 630V here) Not sure why wraper is hating your bias resistor - some is needed for sure, to start operation of the converter.
And if they can't, separate decoupling foil type capacitor has to be used directly across the mosfet pair. The decoupling capacitors must be in close proximity of the mosfets.
I see you copied probably from PC ATX supplies, nothing wrong with that, except PC supplies used NPN transistor, which are rather slow switches, unlike mosfets. Also, the whole control circuit resides directly inside the high current switching loop. Separate those two and join them exactly at the source terminal of the lower mosfet. Your GND of the control circuit is shared with a power ground. The layout does not look that bad, but it is quite bad.
What's wrong with my circuit? How to save those FETs? Please help. I've also tried with the gate-source pull-down resistor and gate to chip reverse diode.
Here you can see, there is only the gate series resistor. Last two days I've damaged total 9 FETs and 5 chips. Sometimes both mosfets burns, sometimes only the high side one. There is no special situations when FETs are burning, sometimes with heavy load, sometimes with light load, and sometimes without any load. But within a few moment, MOSFETs are burning randomly, and sometimes the chip also.
I've then connected it to 220V and checked the pulses again, still OK. Before connecting to 220VAC, I've powered it with 24VDC bench power supply and measured the gate pulses. Hello, I'm trying to make a half bridge converter based on the self oscillated chip IR2153 with an input of 220VAC and output of 12VDC.