MOSFET Upgrade Nightmare in Push-Pull LLC: 2× Voltage Spike! Begging TI Masters to Kill Parasitic Capacitance

vbsemi

Other Parts Discussed in Thread: UCC27517, TPS28225, ISO5852S

1. Design Context
Developing 300kHz/150W LLC converter (48Vin). Replaced BJT push-pull with MOSFET for efficiency:

2. Disaster Scene (Lab Data)
Efficiency improved 91%→94%, but MOSFETs exploded:

Symptoms:
→ 96V spikes on Vds at full load (200% Vin!)
→ 105MHz ringing on Vgs (Scope shot Fig1)
Critical Parameters:

Parameter Value

Cgd (VBGQA1602) 180pF

Loop Inductance 35nH

Switching Speed 18ns

Spike Duration 42ns

3. Failed Rescue Attempts
① Enhanced Driving:

Added totem-pole with TI TPS28225 → Drive current 2A
→ Outcome: Spikes reduced to 82V, but ringing worsened (Fig2)
② RC Snubber:
DS snubber (2.2Ω+470pF) → Spikes 68V, but 1.8W added loss
③ MOSFET Swap:
Used TI CSD19536 (Cgd=65pF) → Spikes eliminated, but 300% cost overrun

4. Desperate Plea to TI Community
Seeking wisdom from TI experts:
① Drive Design:

Is isolated driver (e.g. TI ISO5852S) mandatory? Any cost-sensitive alternatives?
Is Rg=√(L_loop/Ciss) valid for such scenarios?
② Parasitic Taming:
How to quantify PCB stray inductance vs Cgd resonance? TI AN-2014's F_ring=1/(2π√(L_stray×Cgd)) shows >40% error!
③ TI Device Trade-off:
Any cost-performance TI MOSFET between CSD19536 ($4.2) and local MOS ($0.9)? Specs:
text
```
Vds≥60V, Rds(on)≤5mΩ, Cgd<100pF, DFN5x6  
```