My friend was using a Burmester 077 Preamplifier paired with a dCS Apex DAC.
Sorry, I didn’t receive all the files, but I asked my friend to redo the test and upload it to YouTube.
Please check the video here:
Unfortunately, the video is in Chinese. Class D record comes first, followed by the Gryphon Apex.
He used a Burmester 077 preamplifier and a dCS DAC for the test.
Thanks!
What was the ADC, and where was it connected, or how was it mic'ed?
Also, is there a copy of the original recording?
Last edited:
Nice, thanks for that. I do not really care which pipeline was used, as long as the pipeline was the same for both scenarios.
However, it would be nice to know what class D amp was used, and if possible, I'd like to see the innards.
I've done some preliminary listening, and I can tell the differences are again quite obvious.
However, it would be nice to know what class D amp was used, and if possible, I'd like to see the innards.
I've done some preliminary listening, and I can tell the differences are again quite obvious.
I'd love to be able to compare them using the HQPlayerPro... but that's a no-go because the HQPlayer won't open (can't decode) m4a files... bummer.
All files seem to be compressed a fair bit.... maybe capture them as wav files and upload to Cloud, then share a download link with us..??
On a quick listen, I can tell that I don't like (at all...) how 5-A sounds compared to 5-B....
All files seem to be compressed a fair bit.... maybe capture them as wav files and upload to Cloud, then share a download link with us..??
On a quick listen, I can tell that I don't like (at all...) how 5-A sounds compared to 5-B....
I listened to the first two files posted. They both sound wrong. There is sort of etched HF edge on the sound of vocals that is unlikely to be on a good recording. Could be an artifact of a tube preamp though. Some high end dacs have that quirk too.
Without knowing what is going into the inputs of the power amps versus what is coming out, I find such comparison recordings useless. Yes, the recordings sound clearly different, but neither one sounds right. So what am I to conclude, is it that there are two problematic amplifiers? Or, maybe its that the ADC and or the rest of the recording setup is bad? There is no way to tell without some more recordings. There should be the original recording, a recording of what is coming out of the preamp, and a recording of what is coming out of each power amp. All recordings need to be uncompressed wav files.
Without knowing what is going into the inputs of the power amps versus what is coming out, I find such comparison recordings useless. Yes, the recordings sound clearly different, but neither one sounds right. So what am I to conclude, is it that there are two problematic amplifiers? Or, maybe its that the ADC and or the rest of the recording setup is bad? There is no way to tell without some more recordings. There should be the original recording, a recording of what is coming out of the preamp, and a recording of what is coming out of each power amp. All recordings need to be uncompressed wav files.
@Extreme_Boky @Markw4 Thank you for your suggestions . This test was purely a subjective evaluation, and I hadn't considered using software for further analysis. My friend conducted a comparison using existing equipment at home, but he didn't record WAV files. If there's another opportunity in the future, I will try my best to do so.
Title: The Output LC Filter Matters
Thanks to @S.Huang’s recording, which gave me great encouragement to continue improving the AMP module for even better performance.
Based on feedback from multiple friends who have purchased the AMP module so far, there are some strong points worth keeping—and some weaker aspects that need improvement.
What’s Good:
In response to this feedback, I conducted a test by adding an external LC filter while bypassing the internal (module) LC filter.
The THD test results are shown in the figure below. The THD+N at 1 kHz full-power output is now 0.023%, compared to the previous ~0.06%, with noticeable improvements in high-frequency performance as well. When an input filter is added to the audio analyzer (not shown here), the THD+N result improves further indicating that aliasing is leaking into the audio band.
100Hz, 0.011% @ 95.8W
1kHz, 0.024% @ 95.8W
2kHz, 0.035% @ 95.8W
3kHz, 0.045% @ 95.8W
Subjectively, the improvement in high frequencies is clearly noticeable after the modification.
In summary, the LC filter plays a crucial role in the performance of a Class D amplifier. Here are some key takeaways from the test:
Based on these test results, I designed a third version of the module. The new version features:
Your Comments are welcome!
Thanks to @S.Huang’s recording, which gave me great encouragement to continue improving the AMP module for even better performance.
Based on feedback from multiple friends who have purchased the AMP module so far, there are some strong points worth keeping—and some weaker aspects that need improvement.
What’s Good:
- Excellent low-frequency control – According to a YouTuber, it is infinitely close to a high-end amplifier valued at around $100,000.
- Outstanding transient response.
- Very high resolution – Every detail is presented clearly, giving the feeling that the listener is right in front of the music player.
- Wide and stable soundstage.
- The high frequencies are not smooth and sweet enough.
- The audio input connector is too small.
The THD test results are shown in the figure below. The THD+N at 1 kHz full-power output is now 0.023%, compared to the previous ~0.06%, with noticeable improvements in high-frequency performance as well. When an input filter is added to the audio analyzer (not shown here), the THD+N result improves further indicating that aliasing is leaking into the audio band.
100Hz, 0.011% @ 95.8W
1kHz, 0.024% @ 95.8W
2kHz, 0.035% @ 95.8W
3kHz, 0.045% @ 95.8W
Subjectively, the improvement in high frequencies is clearly noticeable after the modification.
In summary, the LC filter plays a crucial role in the performance of a Class D amplifier. Here are some key takeaways from the test:
- Choose inductors with 5 to 10 times the expected current margin to ensure linear behavior and avoid saturation.
- Use capacitors with highly stable dielectric materials. For the same capacitance value, a higher voltage rating is preferred, as it typically correlates with better linearity and lower distortion.
Based on these test results, I designed a third version of the module. The new version features:
- A newly designed LC output filter
- A higher-performance modulation chip (which is, of course, more expensive)
- A negative power rail for the modulator, to ensure a more precise output triangle wave
Your Comments are welcome!
Attachments
@flmhhh
Thanks for making the updates so quickly! Honestly, I really like this version — it has strong driving capability, excellent mids and lows, and does a great job reproducing details naturally. The new LC circuit also looks like a clear improvement.
Enlarging the input socket is a good move too!
Looking forward to your next version!
Thanks for making the updates so quickly! Honestly, I really like this version — it has strong driving capability, excellent mids and lows, and does a great job reproducing details naturally. The new LC circuit also looks like a clear improvement.
Enlarging the input socket is a good move too!
Looking forward to your next version!
Yeah, most definitely.
I'm glad to see the continued improvements. Nice!
Further to my post #266, I am happy to continue comparing the sound files if you can get someone to record them.
I do have (at least I tend to believe and have been told by many) that my system is very revealing, have worked in high-end industry for a while, have modified some true high-end gear, and listened to music all my life. That certainly helped train my hearing..
So, post the files and I'll be happy to give them a try. Just make sure the file names reveal nothing about the source, but include the sound recording of a fully analog amp - the fewer gain stages, the better. A single-ended class A output stage with a simplistic signal path is always desirable.
I think you're now very close to a decent class AB amp, with a reasonably complex signal path, which is a great achievement.
As always, the above is my approach to best in analog reproduction... many people do not like that sound (2 gain stages only, straightforward signal path, output working in pure class A single-ended) and prefer class D high power at great cost. So, with this in mind, you may have already achieved the goal.
As a DIYer, I see continuous optimization of the design as part of the core enjoyment. These new optimizations don't mean that the previous version is out of date by any means, the key idea and architecture of the design keeps the same.
I'm very appreciate to all the folks who have bought the module. We've been in ongoing private communication, working together to help ensure their amplifiers and the module operate in optimal condition.
I totally agree with you about the core enjoyment of DIY, we are always tweaking and upgrading projects we build.
I guess I tend to separate building from scratch compared to buying a premade module and therefore expectations are different.
But, you have a created unique product that deserves to be the best it can be!
Bravo 👏🏻
I guess I tend to separate building from scratch compared to buying a premade module and therefore expectations are different.
But, you have a created unique product that deserves to be the best it can be!
Bravo 👏🏻