If i may refer you to my latest project: Post in thread 'Small+Loud+Light BT Speaker Build' https://www.diyaudio.com/community/threads/small-loud-light-bt-speaker-build.426612/post-8006948
Both the 56 & 36 to lack power. At some point ill integrate multiple different scalable rails onto one board mixing the 23 and 56.
What you basically want is an idle power comsumption of 0. Efficiency for 100W+ is not as crucial. For this, the actual mosfet gate charge and drive level of the internal H-Bridges play a big role. With 3 parallel ICs, its like 5W of idle power draw difference. And that matters alot
Both the 56 & 36 to lack power. At some point ill integrate multiple different scalable rails onto one board mixing the 23 and 56.
What you basically want is an idle power comsumption of 0. Efficiency for 100W+ is not as crucial. For this, the actual mosfet gate charge and drive level of the internal H-Bridges play a big role. With 3 parallel ICs, its like 5W of idle power draw difference. And that matters alot
The first set of measurements includes the FFT spectrum of the amplifier with the power supply disconnected, the FFT of the operating amplifier with the inverting input of the OPA862 shorted to ground, and time-domain waveforms in response to a square wave excitation. For the oscillograms, the measurement point is located upstream of the filter preceding the QA-403 audio analyzer.
FFT measurement of the amplifier with supply voltage removed:
FFT analysis of the operating amplifier with the inverting input of the OPA862 shorted to ground:
Oscillogram of the amplifier's response to a 5 kHz / 50% duty cycle square wave into an 8-ohm resistive load:
Square wave response (5 kHz, 50% duty) with external load disconnected; 2.2 kΩ resistor remains at output:
FFT measurement of the amplifier with supply voltage removed:
FFT analysis of the operating amplifier with the inverting input of the OPA862 shorted to ground:
Oscillogram of the amplifier's response to a 5 kHz / 50% duty cycle square wave into an 8-ohm resistive load:
Square wave response (5 kHz, 50% duty) with external load disconnected; 2.2 kΩ resistor remains at output:
This measurement series was performed with the tube preamplifier stage bypassed, focusing on the TPA3126 amplifier section with the PFF (Post-Filter Feedback) loop and the phase inverter based on the OPA862. An RC low-pass filter (3.9 kΩ + 470 nF) is placed in front of the phase inverter input.
Frequency reponse:
FFT spectrum at an output power of approximately 1 W:
FFT spectrum at signal level -3dB:
Multitone test at an output power of approximately 1 W:
Multitone test at signal level -16dB:
Multitone test at signal level -3dB:
Frequency reponse:
FFT spectrum at an output power of approximately 1 W:
FFT spectrum at signal level -3dB:
Multitone test at an output power of approximately 1 W:
Multitone test at signal level -16dB:
Multitone test at signal level -3dB:
THD measurement series of the TPA3126 stage (PFF loop + OPA862 phase inverter), with the tube preamp bypassed.
Attachments
-
NoTubePreamp-THD_1W_22Hz.png -
NoTubePreamp-THD_1W_125Hz.png -
NoTubePreamp-THD_1W_1kHz.png -
NoTubePreamp-THD_1W_4kHz.png -
NoTubePreamp-THD_1W_5.66kHz.png -
NoTubePreamp-THD_1W_8kHz.png -
NoTubePreamp-THD_-16dB_22Hz.png -
NoTubePreamp-THD_-16dB_125Hz.png -
NoTubePreamp-THD_-16dB_1kHz.png -
NoTubePreamp-THD_-16dB_4kHz.png -
NoTubePreamp-THD_-16dB_5.66kHz.png -
NoTubePreamp-THD_-16dB_8kHz.png -
NoTubePreamp-THD_-3dB_22Hz.png -
NoTubePreamp-THD_-3dB_125Hz.png -
NoTubePreamp-THD_-3dB_1kHz.png -
NoTubePreamp-THD_-3dB_4kHz.png -
NoTubePreamp-THD_-3dB_5.66kHz.png -
NoTubePreamp-THD_-3dB_8kHz.png
The last two test sets evaluate the full amplifier chain with the tube preamp stage, starting with frequency response, FFT, and multitone analysis.
Frequency reponse:
FFT spectrum at an output power of approximately 1 W:
FFT spectrum at signal level -3dB:
Multitone test at an output power of approximately 1 W:
Multitone test at signal level -16dB:
Multitone test at signal level -3dB:
Frequency reponse:
FFT spectrum at an output power of approximately 1 W:
FFT spectrum at signal level -3dB:
Multitone test at an output power of approximately 1 W:
Multitone test at signal level -16dB:
Multitone test at signal level -3dB:
THD measurements of the complete amplifier including the tube preamplifier stage.
Attachments
-
WithTube-THD_1W_22Hz.png -
WithTube-THD_1W_125Hz.png -
WithTube-THD_1W_1kHz.png -
WithTube-THD_1W_4kHz.png -
WithTube-THD_1W_5.66kHz.png -
WithTube-THD_1W_8kHz.png -
WithTube-THD_-16dB_22Hz.png -
WithTube-THD_-16dB_125Hz.png -
WithTube-THD_-16dB_1kHz.png -
WithTube-THD_-16dB_4kHz.png -
WithTube-THD_-16dB_5.66kHz.png -
WithTube-THD_-16dB_8kHz.png -
WithTube-THD_-3dB_22Hz.png -
WithTube-THD_-3dB_125Hz.png -
WithTube-THD_-3dB_1kHz.png -
WithTube-THD_-3dB_4kHz.png -
WithTube-THD_-3dB_5.66kHz.png -
WithTube-THD_-3dB_8kHz.png
In the tube stage, the capacitance of the capacitor between the anode and ground should be reduced. A slight peaking in the frequency response should be observable in the measurement. This will help avoid excessive bandwidth limitation when the load exhibits low resistance and inductance in the range of approximately 30 µH — the worst-case scenario observed in simulations.
The square wave response is, in my opinion, excellent — even when the only load is the 2.2 kΩ resistor mounted on the PCB.
A particularly interesting result is the THD at 1 kHz and approximately 1 W output power, which measures 0.0037% — five times lower (corresponding to the PFF loop depth) than the 0.02% specified in the TPA3126 datasheet.
For signal levels of -16 dB and -3 dB, the measured THD values are 0.021% and 0.022%, respectively. Unfortunately, there is an increase in the third harmonic at -16 dB input level, where I would have expected a lower overall THD — particularly for the -3 dB condition, which I anticipated to yield at least 50% lower distortion.
Well, what remains for me is to further optimize the values of the PFF loop components through simulation to slightly reduce the frequency response dependency for different RL load combinations and to modify the operating point of the tube stage. I might consider using the MAX5440 with an encoder as a replacement for the potentiometer at the amplifier input. As a second option, I'm considering a version with more classic tubes like the EF95, plus a magic eye EM87 as a visual attractor.
The square wave response is, in my opinion, excellent — even when the only load is the 2.2 kΩ resistor mounted on the PCB.
A particularly interesting result is the THD at 1 kHz and approximately 1 W output power, which measures 0.0037% — five times lower (corresponding to the PFF loop depth) than the 0.02% specified in the TPA3126 datasheet.
For signal levels of -16 dB and -3 dB, the measured THD values are 0.021% and 0.022%, respectively. Unfortunately, there is an increase in the third harmonic at -16 dB input level, where I would have expected a lower overall THD — particularly for the -3 dB condition, which I anticipated to yield at least 50% lower distortion.
Well, what remains for me is to further optimize the values of the PFF loop components through simulation to slightly reduce the frequency response dependency for different RL load combinations and to modify the operating point of the tube stage. I might consider using the MAX5440 with an encoder as a replacement for the potentiometer at the amplifier input. As a second option, I'm considering a version with more classic tubes like the EF95, plus a magic eye EM87 as a visual attractor.