Quote:
|
Originally Posted by Richard
As I said above, thermal analysis of FET installations can get complicated. To see how, let's use the Victor as an example:
As an extreme case, let's consider operating the FETs in free air with no convection. For this case, the data sheet provides a figure of 62 Kelvins (Celsius degrees) junction temperature rise (over ambient) for each Watt dissipated in the FET junction. A reasonable maximum for usable junction temperature might be 150 C, and ambient temperature on the FRC field might be 30 C, so the rise would be 120 Kelvins and the maximum dissipation would be 120/62 = 1.94 Watt. At 150 C junction temperature the FET's on-state resistance is Rds(on) = 0.02 Ohm [this was obtained by multiplying the nominal value 0.012 Ohm by 1.7, taken from Fig. 4 on the data sheet] so the maximum current in the FET is sqrt(1.94/0.02) = 9.7 Amperes. There are three FETs per leg, so the maximum leg current is about 29 Amperes. We can improve on this by running the cooling fan, which will reduce the thermal resistance from junction to ambient by a considerable factor, probably at least two. Using 31 Kelvins per Watt of dissipation and repeating the calculations above gives 3.87 Watts maximum dissipation for each FET and therefore 13.9 Amperes per FET and 41.7 Amperes per leg.
Of course this analysis predicts far less than the maximum 120 Amperes arrived at in the earlier posts; to get there, we would need to improve the thermal resistance from junction to ambient still further. A practical lower limit might be about three times the thermal resistance from junction to case (given as 1.6 degrees K/W by the data sheet) or 4.8 Kelvins per Watt. Using this figure, the maximum dissipation per FET is 25 Watts, the maximum FET current is 35 Amperes, and the maximum leg current is 105 Amperes. Based on the Victor's 40 Ampere rating, I'd guess that this level of FET cooling is probably beyond the capability of the Victor design. For reason's I gave earlier, it is probably not a good idea to try to retrofit the Victor with home-made heatsinks in an attempt to improve its thermal performance.
|
I am not arguing that the heat analysis was simple, I am just reminding folks that there were 3 fets per leg.
As to the heat analysis, I would suppose that 31 Deg K per Watt is too high for forced air convection.
This relatively small, clip on heatsink has rise of less than that based on natural (unforced) convection. A little bit of wind goes a long long way with respect to cooling.
Unless something happened that I am not aware of, Victors were not dropping left and right due to fet failure. Based on the years of service we have had with the Victor, I would guess that the 40A breaker is not allowing the fets to get anywhere near their temp limit. Based on that and assuming the math above is correct, I would estimate the effectiveness of the cooling to be higher than the 31 Deg K per Watt.
For Watt it's worth... ...;-)
Joe J.