Heatsink Area and Total Losses Estimation

Posted by: lanaldo on 03/20/2012

Hi! Please,
Could you tell me how PIExpert estimate a custom aluminum heatsink area? I tried to find any explanation/formula at PI application notes, but I didn't find it.

Total Losses = conduction loss + switching loss? How do I can estimate RDS and CXT from datasheet as close as PIExpert?

I want to use PIXIs Designer that does not give losses and heatsink estimation.

Thanks
Arnaldo

Comments

The heatsink area is calculated based on empirical data. This data is based on an assumption of total switching losses inside the TOPswitch. While it is relatively straightforward to estimate conduction losses, the switching loss estimation can be difficult (without actually having hardware built and waveforms to see).
The software uses the typical RDSon value from the datasheet (at Tj = 125 or 100 C and not at 25 C)

Thanks a lot!
I made some designs with PIExpert using the same input specification, but different devices in the same family (TOPSwitch-HX).
Input: 90 to 240Vac, 60Hz, 132kHz Output: 12V @ 2A.
TOP254EN: Ip=1,09A, Irms=0,48A; Total losses=2,3W Heatsink area= 1752mm2
TOP258EN: Ip=1,10A, Irms=0,48A; Total losses=0,9W Heatsink area= 400mm2 (KI=0,30)
Conduction loss regarding RDSon typic:
TOP254EN: RDSon typ= 8,35=> PIR = 1,93W
TOP258EN: RDSon typ= 2,5 => PIR = 0,58W
I supposed the differences are due to switching loss estimation, am I right?
that are around the same for both, 0,3W.
You told that switching loss estimation can be difficult without the hardware and waveforms. Does PIExpert use a typical value to estimate the capacitance at the drain node? In this case, the values seems to be almost the same to TOP254 and TOP258, am I right?

for VDS = Vmax + VOR =339,4+122,6 and f=132kHz => Cxt~21pF

Please, could you help me in how to use the graph "Output Characteristics" to estimate RDSon?
Suppose Ip = 1.10A, Irms = 0,48A using TOP258 and TOP254. How do I use the "Scaling Factors"?
and, please, how can I estimate Coss using "COSS vs. DRAIN Voltage" graph?
VDS = 460V to TOP258 and TOP254
Thanks
Arnaldo

I made some questions in my last reply, but nobody help me with the answers or comments.
Please, could you help me.
Thanks
Arnaldo

The best way to do this is to plot the waveform at the switching transtion and then compute the losses from there.
Theoretically the way to compute the COSS losses is to assume a transtion time (of 50 or 75 ns say) and then use this to estimate the losses. Then you can use the Voltage (as you computed) and the peak current to regenerate the waveform and then perform a theoretical integration over 50 ns transtion time to compute the losses.

Thanks a lot