TVS Diode and Flyback Converter

[strape81]

Hi friends,

I have a serious overvoltage problem in a flyback converter and I need a piece of advice.

As attachments you can find the Drain-Source Voltage of the MOSFET IPB60R385CP during switch-off time with a snubber circuit (TVS diode : SM15T150CA) and without it.View attachment without snubber.bmpView attachment with snubber.bmpView attachment IPB60R385CP_rev2.0.pdfView attachment CD00000724.pdf

The power dissipated on the TVS Diode is calculated according to the following formula:

P_tvs = 0,5 x F_sw x L_leak x I_pk^2 x (V_z / (V_z-(V_out+V_f)/N) = 9,055W (Source: Christopher Basso- Switch Mode-Power Supplies, page 614)

with F_sw = 65kHz - switching frequency
L_leak = 23uH - leakage inductance of flyback transformer
I_pk = 2A - Peak current through mosfet
V_z = 150V - TVS breakdown voltage
V_out = 19V - Output voltage
V_f = 0,7V - Forward voltage of the secondary diode
N = 10/51 - turns ratio

Using this calculation and the junction-ambient thermal impedance of 90 °C/W I come up to a huge junction temperature of Tj=Ta+Rth(j-a) x P_tvs = 25°C + 90°C/W x 9,055W = 840°C.
I am sure there is a mistake in this calculation. Can somebody tell me how can I correctly compute this???

Also how can I decide which snubber circuit is the correct for a flyback converter? Which parameters are playing the most significant role??

Thanks a lot for your time!!!

 

[dselec]

Where did u get this calculation ?
and some thing i dont see in eqation is the pulse factor , u are using pulse and not plain dc and 90 ©deg /w is for dc.
check it out

 

[strape81]

Thanks dselec for your trigger-answer!

If I understood correctly, the correct way to compute the junction temperature is to compute the Z_th(j-a), right? This can be done by extrapolating for a pulse width of 300ns.

In this case, I have done the following calculation:

Pulse duration Zth(j-a)/Rth(j-a)
1ms 0,025
300ns x=7,5x10e-6

and I have calculated the Zth(j-a) = Rth(j-a)x7,5x10e-6

The junction temperature is computed then: Tj=P_inst x Zth(j-a) + Ta

Please could you verify if my thoughts are on the right way???

Thanks a lot!!!

 

[dselec]

Seems u got it right this time.
i have not enough experience in this matter but i know computing heat differs ac from dc.
good luck