I put an AMB Mini3 DIY headphone amp through a full suite of measurements using a Prism Sound dScope audio analyzer, and the results show some fairly serious problems--especially into low impedance loads. I also uncovered some other concerns about the design in general.
I have lots of respect for the AMB website and their approach so it was disappointing to find the Mini3 doesn't come close to many of their published specs. I wasn't expecting perfection from what's basically an enhanced Cmoy, but it was worse than I expected. There's clear evidence the 3 channel virtual ground configuration, for example, does far more harm than good.
If anyone has any questions or comments please let me know? You can find the review here:
AMB Mini3 DIY Headphone Amp Review
I have lots of respect for the AMB website and their approach so it was disappointing to find the Mini3 doesn't come close to many of their published specs. I wasn't expecting perfection from what's basically an enhanced Cmoy, but it was worse than I expected. There's clear evidence the 3 channel virtual ground configuration, for example, does far more harm than good.
If anyone has any questions or comments please let me know? You can find the review here:
AMB Mini3 DIY Headphone Amp Review
You solicit comments, and they are here:
AMB Laboratories DIY Audio • View topic - Somebody isnt happy
AMB Laboratories DIY Audio • View topic - Somebody isnt happy
"3-channel" is simply wrongly conceived and technically flawed
to be fair on the subject of current limit I have talked to AD app engineers about AD8397 - they indicated a lack of output current limit - there may be thermal limit but too slow to save the chips in all fault scenarios – it is apparently hard to have speed, rail-to-rail output at high current and bullet proof protection all at the same time
my solution was to put the current sense/limit in the supply rail for my app – costs too much headroom in a battery powered headphone amp
my app was active gnd supply splitting in a strain gage amplifier that amplifies signals by a few 1000x voltage gain and then digitizes the signal with on chip ADC/100MHz processor, spitting data over USB 2.0
I very much use hierarchical star segmented gnd plane and carefully examined and planed power distribution
turns out my conception/practice of active supply splitting and "passive" gnd plane/wiring works at a very high quantified level - I do not use the head-fi popularized "3-channel" principles - Industrial/Scientific Instrument customers don't like bad measurements or failing to meet our published specs
to be fair on the subject of current limit I have talked to AD app engineers about AD8397 - they indicated a lack of output current limit - there may be thermal limit but too slow to save the chips in all fault scenarios – it is apparently hard to have speed, rail-to-rail output at high current and bullet proof protection all at the same time
my solution was to put the current sense/limit in the supply rail for my app – costs too much headroom in a battery powered headphone amp
my app was active gnd supply splitting in a strain gage amplifier that amplifies signals by a few 1000x voltage gain and then digitizes the signal with on chip ADC/100MHz processor, spitting data over USB 2.0
I very much use hierarchical star segmented gnd plane and carefully examined and planed power distribution
turns out my conception/practice of active supply splitting and "passive" gnd plane/wiring works at a very high quantified level - I do not use the head-fi popularized "3-channel" principles - Industrial/Scientific Instrument customers don't like bad measurements or failing to meet our published specs
Last edited:
Thanks JCX for validating my belief 3 channel/virtual ground solutions have serious problems.
I can understand the desire to make a smaller portable amp with just one battery and no output capacitor. But even faced with that constraint, I think a charge pump to generate a negative supply rail would be a much better solution.
The measurements clearly show even a "high performance" virtual ground still has lots of obvious, and potentially audible, problems. The worst being the intermodulation mess I observed when I drove the two channels with different signals--as is the case with stereo audio. This just is NOT good. And what's not shown is how the spectrum was constantly changing due to the phase variations between the channels. The result below shows a 1 Khz 1 volt output with the other channel (not shown) driven with 300 hz:
I can understand the desire to make a smaller portable amp with just one battery and no output capacitor. But even faced with that constraint, I think a charge pump to generate a negative supply rail would be a much better solution.
The measurements clearly show even a "high performance" virtual ground still has lots of obvious, and potentially audible, problems. The worst being the intermodulation mess I observed when I drove the two channels with different signals--as is the case with stereo audio. This just is NOT good. And what's not shown is how the spectrum was constantly changing due to the phase variations between the channels. The result below shows a 1 Khz 1 volt output with the other channel (not shown) driven with 300 hz:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.