Hackers, let’s commiserate. I’ve been up until 2am reflowing a push-pull board for the third time this week, and I need your take.

Here’s the mess: 100W design, failing EMI at 200MHz (peaking 8dB over limit), and the MOSFET’s junction temp hits 118°C with a heatsink. Calculations said it should work—so where’s the leak?

Let me walk through the chaos, and tell me where I’m screwing up:

• EMI gremlins: The spectrum shows three peaks: 80MHz, 200MHz, 420MHz. Swapped in a MOSFET with lower parasitic caps, and 80MHz dropped 10dB… but 200MHz? It’s like a stubborn hum. Is that transformer leakage inductance oscillating with stray caps? Tried an RCD snubber—helped a little, but not enough.
• Layout wars: Power loop’s tight (3cm²), but splitting power/signal grounds with a 0.1Ω resistor moved the EMI peak to 250MHz instead of fixing it. Guarded the gate traces with ground planes—switching got faster, but EMI only dropped 3dB. What’s the actual trick here?
• Thermal math fail: My napkin math: 8W total loss, heatsink with 6°C/W, ambient 25°C → junction temp 73°C. Reality? Thermal camera says 118°C. Did I undercount switching loss? (It’s a 120kHz design—maybe E_on/E_off are higher than the datasheet claims?) Or is the shield I added (to kill EMI) trapping heat like a tiny oven?
• Failure flashback: Last batch fried a MOSFET—die had a molten spot. Dead time was 100ns, but scope showed a 450V spike when it turned off. Leakage inductance? Snubber C was 1nF—too small? Too big?

This isn’t theory—it’s a board on my bench, and I need your battle scars. What’ve you learned the hard way with push-pull designs?

#PushPullHacks #EMIdebug #ThermalHacks #PowerPrototyping

English Version:

Title: Late-Night Push-Pull Headaches: Why’s My EMI Spiking at 200MHz, and Why Won’t This MOSFET Cool Down?

Hackers, let’s commiserate. I’ve been up until 2am reflowing a push-pull board for the third time this week, and I need your take.

Here’s the mess: 100W design, failing EMI at 200MHz (peaking 8dB over limit), and the MOSFET’s junction temp hits 118°C with a heatsink. Calculations said it should work—so where’s the leak?

Let me walk through the chaos, and tell me where I’m screwing up:

• EMI gremlins: The spectrum shows three peaks: 80MHz, 200MHz, 420MHz. Swapped in a MOSFET with lower parasitic caps, and 80MHz dropped 10dB… but 200MHz? It’s like a stubborn hum. Is that transformer leakage inductance oscillating with stray caps? Tried an RCD snubber—helped a little, but not enough.
• Layout wars: Power loop’s tight (3cm²), but splitting power/signal grounds with a 0.1Ω resistor moved the EMI peak to 250MHz instead of fixing it. Guarded the gate traces with ground planes—switching got faster, but EMI only dropped 3dB. What’s the actual trick here?
• Thermal math fail: My napkin math: 8W total loss, heatsink with 6°C/W, ambient 25°C → junction temp 73°C. Reality? Thermal camera says 118°C. Did I undercount switching loss? (It’s a 120kHz design—maybe E_on/E_off are higher than the datasheet claims?) Or is the shield I added (to kill EMI) trapping heat like a tiny oven?
• Failure flashback: Last batch fried a MOSFET—die had a molten spot. Dead time was 100ns, but scope showed a 450V spike when it turned off. Leakage inductance? Snubber C was 1nF—too small? Too big?

This isn’t theory—it’s a board on my bench, and I need your battle scars. What’ve you learned the hard way with push-pull designs?

#PushPullHacks #EMIdebug #ThermalHacks #PowerPrototyping