Buck switcher thermal

Hi,

I'm simulating the step-down regulator LT3502 in LTSpice and don't know what to think of the thermal part. In the attached screenshot you can see a graph of what the IC itself dissipates. Sure peaks of up to 28 W are high, this would give a junction temperature of 28W * 110°C/W = 3080 °C > 125 °C, the maximum junction temperature (let's say we can neglect the ambient temperature here lol).

When I calculate the average of the graph you see, it is only 0.315 W, so 0.315 W * 110°C/W = 34°C, perfectly acceptable.

How do you decide on what is acceptable and what not?

I think you supply is not stable. 200hz problems.

Hi Ron, you mean because of the peaks every 4-5 ms? That's because I'm simulating GSM current peaks. At this time my supply is drawing 2 A for 0.577 ms, every 4 ms. Could that be what you see? Could you also comment on the thermal question, please?

The IC temperature is determined by the average power, not the peak power.

The LT3502 should not be dissipating anything like 28 W (except for maybe peaks only microseconds long). GSM phones take 2 A at 4 V, so about 8 W. The efficiency of a buck regulator should be about 80% so about 2 W should be dissipated in the power supply, but that is only for 1/8th of the time (1/4 of the time if using GPRS). If all of that is in the LT3502, then the power is 1/4 of 2W, so about 0.5 W

Yes good idea, the 28 watts is for 50nS, the turn on/off time of the switch.

Zoom into one 'period' of that waveform, then do <CTRL> left click on that long plot icon at the top and it should display average power over that time interval. That will give you a better idea of the power dissipated per cycle (hence heat)

Diver300 said:

The LT3502 should not be dissipating anything like 28 W (except for maybe peaks only microseconds long). GSM phones take 2 A at 4 V, so about 8 W. The efficiency of a buck regulator should be about 80% so about 2 W should be dissipated in the power supply, but that is only for 1/8th of the time (1/4 of the time if using GPRS). If all of that is in the LT3502, then the power is 1/4 of 2W, so about 0.5 W

Click to expand...

Sadly it is not 80%. Down from 28 V to 4V gives me about 50% efficiency, I guess it is normal for such a big drop in voltage. So I cannot do GPRS without heatsinking. Normal GSM should be fine: 2A * 4V * 50% * (0.566/4 ms) * 110°C/W = 60°, which is OK.

By the way, is EMI a concern with these switchers?