No, the schematic is same as what for headphone amp beside the more powerful power stage (GaN).
If feedback after LC filter, the AMP won't work or work with low modulation frequency, depending on the cut-off frequency of the LC.
The circuitry can be divided into two parts, the part before the LC filter is a high resolution one-bit ADC, LC filter itself compose a signal reconstructor (DAC). the only red line you shown broken the circuitry of ADC, and it does not work/work well.
of course, keeping these 2-feedback path, it still works. the red path functions as typical negative feedback (inductor and capacitor acting as memory elements) (one-bit ADC is no memory), helping to minimize the distortion caused by the LC filter.
If feedback after LC filter, the AMP won't work or work with low modulation frequency, depending on the cut-off frequency of the LC.
The circuitry can be divided into two parts, the part before the LC filter is a high resolution one-bit ADC, LC filter itself compose a signal reconstructor (DAC). the only red line you shown broken the circuitry of ADC, and it does not work/work well.
of course, keeping these 2-feedback path, it still works. the red path functions as typical negative feedback (inductor and capacitor acting as memory elements) (one-bit ADC is no memory), helping to minimize the distortion caused by the LC filter.
The new housing design including the 2 cascade LC filter is finished, the total size now is 90.6(L) x 53(W) x 11.5(H). I am quite confident with the integrated LC filter now, because the new inductor selected is satisfied with the requirement (fully shield, high SRF frequency, 14A rating DC current and 4mOhm DC resistance, that is great). the EMC is also handled inside the module by 2 high frequency sucking network.
the module is designed to support differential and single end audio input, the gain is 39dB, frequency responds difference of 5Hz - 20kHz is less than 0.8dB, 5Hz-30kHz is less than 1.5dB with the load range from 4ohm to 8ohm.
the module can be driven directly by differential output of DAC through AC coupling, no need pre-amp. if needed, it can also connect via DC coupling, provided DAC accept the bias of module (not for this version).
Need to highlight, a sweet part of the sigma-delta Class D AMP is it can provide higher gain without loss the performance.
Any suggestion still welcome!!! I will wait couple of weeks before the design is sent for fab, until all the suggestions and comments (online and offline) are clear. Please let me know your comments if you want your throughs is implemented in this module.
the module is designed to support differential and single end audio input, the gain is 39dB, frequency responds difference of 5Hz - 20kHz is less than 0.8dB, 5Hz-30kHz is less than 1.5dB with the load range from 4ohm to 8ohm.
the module can be driven directly by differential output of DAC through AC coupling, no need pre-amp. if needed, it can also connect via DC coupling, provided DAC accept the bias of module (not for this version).
Need to highlight, a sweet part of the sigma-delta Class D AMP is it can provide higher gain without loss the performance.
Any suggestion still welcome!!! I will wait couple of weeks before the design is sent for fab, until all the suggestions and comments (online and offline) are clear. Please let me know your comments if you want your throughs is implemented in this module.
Last edited:
What about hysteresis distortion in the output filter?
https://ptt.purifi-audio.com/blog/tech-notes-1/this-thing-we-have-about-hysteresis-distortion-3
https://ptt.purifi-audio.com/blog/tech-notes-1/combating-hysteresis-distortion-part-1-amplifiers-2
https://ptt.purifi-audio.com/blog/tech-notes-1/this-thing-we-have-about-hysteresis-distortion-3
https://ptt.purifi-audio.com/blog/tech-notes-1/combating-hysteresis-distortion-part-1-amplifiers-2
What SNR do you predict. The gain is indeed high and e.g. horn owners might want to trade gain for less noise. An other consequence is that for users that utilise a digital volume control and no preamp will end up with a high attenuation meaning that they will "lose" bits in their DAC. As you offer now about 12 dB higher gain than usual, 2 bits is will be lost for many systems... Perhaps you could consider a setting with two different gains e.g. 26 and 39dB? If it is easily fixed and there indeed where to be a, even if slight, advantage for lesser gain, I would suggest: 12, 24 and 36 dB.
//
//
The expected SNR is larger than 120dB. I did add a switch to change the gain in the Class D headphone AMP. but not for this AMP module, since it is very easy to change the gain with different resistors. What do you think is the right gain setting considering no need pre-amp?
Thanks for your point. but I think they exaggerated the non-linear effect caused by hysteresis feature of magnet core. Please see my test results of the Class D headphone AMP project, it's THD reach 0.001%.
The easy way to demonstrate the non-linear effect of the inductor is measure the THD before and after the inductor. If it is significant, use a coreless inductor might be a good choice, there is no hysteresis effect at all.
To minimize the distortion caused by inductor, select an inductor with much more current margin is a good way to go. that is how I select the inductor in the project.
And absolute voltage RMS noise on output in 8 ohm? Here a class leader amp specs:
| THD+N | <0.00035% @ ?W, 4Ω, 1kHz |
| Dynamic Range | ~129dB(A) |
| Output Noise | ~14µV(A) |
| Gain | 13dB |
| Output Impedance | <13µΩ @ 1kHz |
| Efficiency | >94% |
| Idle Losses | ~2.8W |
How many watt in 4 ohm? 300W?
It depends of course a lot on the speaker sensitivity. Most commercial amps seem to sit around 25 dB +/- 2 dB... I find that this gain in combination with a 90 dB 1m 1W speaker and direct connected angle end DAC gives a resonable gain stage. But as I think about it some more it "varies".... 🙂
There settings would be good I think; 12, 26 and 39.
//
I believe the RMS noise level is quite low. I can't hear any noise using the in-ear earphones with the headphone amp I built. The estimated RMS noise should be around a few microvolts. The amplifier module runs on a 48V supply, which is four times the typical headphone supply voltage, meaning the RMS noise will also be approximately four times higher. Therefore, the RMS noise is expected to be around 20 to 30 µV.
the peak wattage in 4 Ohm is above 200W, will see what happens.
I believe the 26dB is reasonable gain setting, not easy to add a switch to change the gain with minor modification so far.
the peak wattage in 4 Ohm is above 200W, will see what happens.
I believe the 26dB is reasonable gain setting, not easy to add a switch to change the gain with minor modification so far.
Last edited:
Add a DPDT slide switch, the gain can be switched between 13dB and 26dB. the dimension now is 83(L) x 53(W) x 12(H) mm.
I will get the PCA back in 10 days after I place the order, I am still finalizing the design according to the feedback of my friends. probably I plan to receive the playing unit in 2 or 3 weeks.
I will share my debugging process and show the measurements whatever it is good or bad.
By the way, I’d appreciate any ideas for estimating the effective output impedance of a Class D amplifier without Audio signal feedback from LC output.
Received 5 new headphone amp PCAs for my friends, who couldn't resist after their first listen. thanks！
What an interesting project! Very good to see this!
In my experience of modding the chip D amps, such as the TPA3116D2 chinese boards, I rather liked the sound when I removed the ferrite cores and instead wound 10uH air cores, and used 220n poly caps.
If you are aiming for the best sound, in my view this must be air cores, but carefully positioned so the fields do not affect other components.
I felt the air cores gave a very natural clear sound to the amplifiers. If you try them, I'd be interested in how you think they sound.
A metal powder shielded inductor with high SRF is chosen for the headphone amplifier to ensure lower radiation as possible, as the headphones are worn directly on the head, making reduced emissions essential. I may experiment with air core inductors for the louder speaker amplifier later.