urgent help about pulse transformer gate driver......

[olemp]

hi. i am trying to build driver for phase shifted dc-dc converter topology(figure 2).
to test driver i have built figure1. but i could't obtain any successful result.
frequency of the pwm pulses are 85khz. pulse transformer will drive a mosfet which has 8nF input capacitance.
simulate it i have put 10nf to output.
centre area of the pulse transformer(EE core) approximately 0.8cm^2
i have tryed different wind nubbers.(11-15-20-28 number for each windind). outputs of the pulse transformers
are not in desired sheap. they are like in figure 3.
also i am burning my buffers...
i need urgent help.
what do you suggest?..........

 

[sefton]

If you give the values of voltages in fig 3 and the ratio of the transformer, we can determine if the transformer is saturating or not. Perhaps you should try running it without the capacitors first and see if the output voltage is correct according to the wind ratio. If it is low, then the transformer is saturating and then you will need more turns on the primary. If it is saturating, then that would account for your buffers burning.
However too many turns will increase the inductance and there will be higher switching losses. According to fig3 you have some dead time, perhaps it should be increased a bit.

 

[olemp]

thank you sefton.
turn ratio of my transformer is 1:1.
and as i said output of the buffers are 16v.
how can i calculate whether my transformer saturates or not.
is there any link.......
or could you explain this in a basic example...

 

[sefton]

You need to calculate the max flux density. You can use the following formula:

N = V/4*f*a*B

where B = Flux Density (Tesla)
V = voltage (V)
f = frequency (Hz)
a = cross sectional area of core(square metres)
N = no of turns

I haven't done much ferrite design but if I remember correctly a flux density of about 0.05 Tesla should do it. I would use this as a starting point and experiment.
For 0.8cm^2 centre core area and 16V and 85kHz, I get about 12 turns.

I would start with 12 turns and make sure that the peak voltage on the secondary (V2 ) is 16V. If this voltage is less than 16V then the transformer is saturating, and you should increase the turns.
If there is no saturation then reduce the turns by 2 and do the same check again. Continue reducing the turns by 2 until you see the peak voltage dropping, at this point you know that it is saturating and you can then increase the turns by about 40% of the last result. When you reduce the turns you should see the output becomming more square, provided the core is not saturating.

Also at 85Khz, remember the skin effect on the wire, although with a pulse transformer this should not be a problem.

Some ferrite cores have an air gap, I'm assuming that your's does not.

 

[VVV]

I only have three things to say:
1. although the circuit is symmetrical, there may be flux imbalance in the transformer, since the two drivers may not be identical. I suggest the use aof a capacitor (0.1uF) in series with the primary.
2. The MOSFET input capacitance is variable, so it cannot be accurately modeled by the lumped capacitors. I would use the actual MOSFETs in the actual circuit. Just short power-ups, maybe even at a lower frequency initially. If the gate drive waveforms still look bad, then we will have to look at the actual transformer.
3. I think the resistors in series with the primary are too low. I don't know what MOSFETs you are driving, but I thought 10Ω total should do it.