FAQ
TL;DR: At 50 kHz, an ETD49 3C90 transformer driven at 500 W reaches 140 °C after 10 min—"the primary winding is too large" [Elektroda, hajy, post #16096713]—reduce turns, avoid air gaps, and add an output choke to cut copper and core losses [Elektroda, dannnek, post #16082003]
Why it matters: Small winding tweaks slash >60 % thermal stress, protecting MOSFETs and insulation.
Quick Facts
• Safe flux density for 3C90 at 50 kHz: 0.20–0.25 T [Ferroxcube, 2015].
• Skin depth in copper at 50 kHz: ≈0.3 mm [IEEE Std 519].
• ETD49 3C90 AL (ungapped): 4200 nH/N² [Ferroxcube, 2015].
• Recommended litz current density: ≤3 A/mm² [Elektroda, dannnek, post #16082689]
• Primary copper fill factor target: <40 % [Ridley, 2014].
1. Why did my ETD49 primary winding overheat to 140 °C?
Too many primary turns (39) created high AC resistance and proximity losses. Combined with 500 W load and no output choke, copper losses pushed temperature to 140 °C in 10 min [Elektroda, dannnek, post #16082003] Reducing turns by ~50 % cut winding temperature to 50 °C [Elektroda, dannnek, post #16098769]
2. How many primary turns suit 325 V DC, 50 kHz on ETD49?
Keep flux below 0.22 T. Use N = (V·D)/(4·B·A_e·f). For 162 V peak (half-bridge), 0.22 T, A_e = 2.55 cm², and 0.45 duty, N≈21 turns. This matches empirical 20–22 turns from autotransformer testing [Elektroda, lazor, #16124952].
3. Should I add an air gap to a half-bridge transformer?
No. A transformer should store minimal energy; gapping raises magnetising current and losses [Elektroda, jarek_lnx, post #16111779] Use gaps only in flyback or chokes.
4. What current waveform is normal in the primary?
With an output choke, current ramps linearly (triangle). Without the choke or with a capacitive doubler, the current flattens toward a rectangle [Elektroda, hajy, post #16093039]
5. How do I measure volts-per-turn safely?
Follow this 3-step test:
- Wind 10 turns test coil on the core.
- Drive the half-bridge from a variac and monitor RMS current.
- Increase voltage until current rises sharply; divide applied volts by turns for safe V/turn [Elektroda, lazor, #16124952].
6. Why does a voltage doubler make the transformer hotter?
The doubler’s capacitors draw large reactive current; apparent power quadruples, so core and copper losses rise sharply [Elektroda, hajy, post #16095289] Energy stored equals ½ C·V²; at 1 µF and 1 kV that is 500 mJ each cycle, stressing MOSFETs.
7. What current density should I use for 0.1 mm litz at 50 kHz?
Stay below 3 A/mm² to limit joule heating; the thread’s 0.94 mm² litz safely carries ~2.8 A RMS at this frequency [Elektroda, dannnek, post #16082689]
8. How do I size the output choke to keep it cool?
Set peak-to-peak ripple ≤20 % of output current. L = (V_out·(1-D))/(f·ΔI). For 120 V, 0.5 A ripple, 50 kHz, duty 0.45, you need ~5 mH. Use gapped ferrite to keep flux under 0.25 T. Oversize wire to keep ≤80 °C rise [Texas Instruments, 2021].
9. Can an LLC converter provide 0–500 V DC regulation?
Only across a narrow range. Frequency control gives about 3:1 voltage span before ZVS/ZCS is lost [Elektroda, lazor, #16184516]. For wider range, add preregulation or phase-shift control.
10. What edge-case causes instant MOSFET failure here?
Core saturation plus capacitive load forms a dead short; MOSFETs see uncontrolled current and short within 60 s [Elektroda, dannnek, post #16125434]
11. Could primary burn-out be due to mains over-voltage?
Yes, but unlikely if other devices on the circuit survived. Consistent over 260 V AC would be required to double core flux [Elektroda, brofran, post #18045361]
12. Which software helps calculate SMPS transformers?
SFDT2010 estimates 740 W capability of ETD49 at 50 kHz [Elektroda, dannnek, post #16136148] Eastern-bloc program ‘Магнит’ provides detailed winding loss data [Elektroda, hajy, post #16130759]
Because there is a difference between 10kV and 10mA and 100mA 
