I purchase a pair of MX50 amplifier kit from eBay long ago. One channel was assembled but never tested. The output transistors were not even connected.
Recently I connected the output power transistors (2SC4468/A1695) and downloaded its schematic from the internet. I want to try changing it compensation cap value and using other compensation methods. The original circuit seems to be over compensated. The Miller cap is 100 pF. It have input compensation RC across input differential pair collectors. 40 pF in parallel base resistor of output CFP.
I made a simplified circuit in LTspice with only Miller cap (dominant pole). Will the circuit oscillate? Output power transistors are temporary MJL21194/21193 (Cordell spice Models).
Loop-gain (LG) measurement is done by injecting test signal via transformer as outlined in Maxim's app note Injection Transformers for Closed-Loop DC-DC Converter Network Analysis - Application Note - Maxim. Probe loading is modeled by 12pF//10 MOhm. This is my Lecroy x10 passive probes characteristic. I will use HP3577A to take measurement for comparison (simulation vs read data).
LG of Miller cap equal to 100p, 40p and 20p is simulated. The results are shown in the second figure. Using lower cap value, LG at 60 kHz is increased by about 13 dB (100p vs 20p). Higher LG at higher frequencies improves amplifier higher freq. THD. Nevertheless, 13 dB is not that much. Phase margin (PM) for 20 p is ~50 deg.
Third figure shows measured LG for cap value equal to 20p and 40p. Crossover freq. are ~1.5 MHz and ~3.0 MHz. Simulation results and measurement are well consistent.
Fourth figure shows simulation result of two-pole compensation (40p, 3.3k, 40p) vs dominant pole 20p. LG at 60 kHz is increased for 29 dB. PM for the two-pole case is 30 deg. This is too low in practice.
Fifth figure is measured LG for two-pole and dominant pole as simulated. LG increment at 60 kHz is about 16 dB. PM is about 64 deg for two-pole.
My next step is to measure THD for the cases above.
Recently I connected the output power transistors (2SC4468/A1695) and downloaded its schematic from the internet. I want to try changing it compensation cap value and using other compensation methods. The original circuit seems to be over compensated. The Miller cap is 100 pF. It have input compensation RC across input differential pair collectors. 40 pF in parallel base resistor of output CFP.
I made a simplified circuit in LTspice with only Miller cap (dominant pole). Will the circuit oscillate? Output power transistors are temporary MJL21194/21193 (Cordell spice Models).
Loop-gain (LG) measurement is done by injecting test signal via transformer as outlined in Maxim's app note Injection Transformers for Closed-Loop DC-DC Converter Network Analysis - Application Note - Maxim. Probe loading is modeled by 12pF//10 MOhm. This is my Lecroy x10 passive probes characteristic. I will use HP3577A to take measurement for comparison (simulation vs read data).
LG of Miller cap equal to 100p, 40p and 20p is simulated. The results are shown in the second figure. Using lower cap value, LG at 60 kHz is increased by about 13 dB (100p vs 20p). Higher LG at higher frequencies improves amplifier higher freq. THD. Nevertheless, 13 dB is not that much. Phase margin (PM) for 20 p is ~50 deg.
Third figure shows measured LG for cap value equal to 20p and 40p. Crossover freq. are ~1.5 MHz and ~3.0 MHz. Simulation results and measurement are well consistent.
Fourth figure shows simulation result of two-pole compensation (40p, 3.3k, 40p) vs dominant pole 20p. LG at 60 kHz is increased for 29 dB. PM for the two-pole case is 30 deg. This is too low in practice.
Fifth figure is measured LG for two-pole and dominant pole as simulated. LG increment at 60 kHz is about 16 dB. PM is about 64 deg for two-pole.
My next step is to measure THD for the cases above.
Attachments
- Status
- Not open for further replies.