Geshelli Archel2 Headphone AMP Review: another champ?

[solderdude]

To put the rise in perspective I overlayed the Phon curves.

Even when 0dB (which is -20dB) would be 110 Phon you would still hear total silence. 110 Phon is very loud. One could have 130dB peaks and still be below the hearing threshold for the amplifier noise.

 

[LLL]

Because real engineers don't do audio products.

What? @tomchr would have something to say about that; and probably all the Topping engineers, too.

 

[JohnYang1997]

What? @tomchr would have something to say about that; and probably all the Topping engineers, too.

Very few I should say. Because if you really think of it. Audio is already a dead horse which has no meaning of improving and technology/knowledge was already proven and set for decades. Surely designing ICs are more extensive. Still it's just something someone loves audio. Like johnc who designed opa1622 opamp.
I will be doing my thesis soon, and including my friends no one is doing audio related topic. The most related is probably neural signal amplifier. Also designing opamp using 32nm process etc. Designing audio is easy. I didn't have the scope for years until I finally realize that audio is just something in your heart. I'll be doing 32nm 1V opamp or 0.2nV/sqrthz 200MHz BW 100dB gain amplifier or something like that. And some others would to tele, control, DSP etc.

And when I say topping I didn't mean in such way. I was talking about making it stable is easy. And making this level of performance is nor hard at all. I didn't mean that topping has all the great engineers or anything.

 

[KSTR]

It is relatively easy to get good 1kHz THD and noise in unity gain with the pot wiper full up. It would be interesting to see how it fares at worst-case wiper (0.5 == 10...15% rotation for a log pot) where the source impedances at the opamp are mismatched. In a noninverting config the OPA1642 still is very good because of its internally bootstrapped input capacitance. But the 1642 is not a loop gain monster so at 10kHz the distortion would be signifcantly higher... pure guesswork of course as we don't know if the 1642 is the opamp used here....

 

[LuckyLuke575]

Very few I should say. Because if you really think of it. Audio is already a dead horse which has no meaning of improving and technology/knowledge was already proven and set for decades. Surely designing ICs are more extensive. Still it's just something someone loves audio. Like johnc who designed opa1622 opamp.
I will be doing my thesis soon, and including my friends no one is doing audio related topic. The most related is probably neural signal amplifier. Also designing opamp using 32nm process etc. Designing audio is easy. I didn't have the scope for years until I finally realize that audio is just something in your heart. I'll be doing 32nm 1V opamp or 0.2nV/sqrthz 200MHz BW 100dB gain amplifier or something like that. And some others would to tele, control, DSP etc.

And when I say topping I didn't mean in such way. I was talking about making it stable is easy. And making this level of performance is nor hard at all. I didn't mean that topping has all the great engineers or anything.

I really hate what they've done in the modern academy; it's all fandangled and outlandish bullshit research topics that they want to publish in papers for other academics to arbitrate and read. It's a total sham. Audio is a huge industry with legitimate design and R&D going on. I can assure you that there are real engineers and designers hard at work on the next generation of audio equipment; this Archel 2 is one great example of that.

 

[JohnYang1997]

It is realitvely easy to get good 1kHz THD and noise in unity gain with the pot wiper full up. It would be interesting to see how it fares at worst-case wiper (0.5 == 10% rotation for a log pot) where the source impedances at the opamp are mismatched. In a noninverting config the OPA1642 still is very good because of its internally bootstrapped input capacitance. But the 1642 is not a loop gain monster so at 10kHz the distortion would be signifcantly higher... pure guesswork of course as we don't know if the 1642 is the opamp used here....

My solution is just to use 1K pot. Also 10k pot(2.5k maximum equivalent resistance) shouldn't cause distortion either.
Opa1642 opa827 are the best ones in this regard nonetheless.

 

[JohnYang1997]

I really hate what they've done in the modern academy; it's all fandangled and outlandish bullshit research topics that they want to publish in papers for other academics to arbitrate and read. It's a total sham. Audio is a huge industry with legitimate design and R&D going on. I can assure you that there are real engineers and designers hard at work on the next generation of audio equipment; this Archel 2 is one great example of that.

What I think of it now is the solving problem. If there is a problem, then we can solve it. If there isn't a problem or has been solved decades ago, then there is no need for that. Audio industry is full of BS. If the atmosphere is clear and everyone is making highest measurable performance products(it's definitely better recent years). These kind of performance won't be hard to see. It should be more like 90% of products are at THX performance, and 9% would be ATOM and 1% would be O2 or worse.

 

[KSTR]

OPA828 is my personal favorite, especially for composite. Ideal transfer function with no "phase kink", low and stable open-loop output impedance, low noise.

 

[JohnYang1997]

OPA828 is my personal favorite, especially for composite. Ideal transfer function with no "phase kink", low and stable open-loop output impedance

Opa828 is great. opa827 was my favorite and 828 is the next gen opamp to 627 and 827. However there is a graph showing the distortion harmonics. The 2nd harmonic is a bit too high comparing to other existing products (won't be an issue of course). So I adopted opa2156(before opa1656 release) , and opa1656 is one of my favorite now.
828

1656

One of the thing about phase kink. Actually it doesn't matter. In fact opa1612 has a lot of bandwidth that we can extract a lot more phase margin. lm4562 has even better phase characteristics. One can consider making discrete ultra low noise input stage with lm4562 in the middle and an output buffer all in one loop.

 

[KSTR]

Yeah, this new 1656 is a giant killer ;-)
Currently building an OPA828 + THS4631 composite to see if the "high" 2nd HD of the 828 is coming from the input stage...

 

[JohnYang1997]

Yeah, this new 1656 is a giant killer ;-)
Currently building an OPA828 + THS4631 composite to see if the "high" 2nd HD of the 828 is coming from the input stage...

It's -140dB non the less. With a load the number would be irrelevant.

 

[scott wurcer]

I think their worry is other companies, not DIY.

The reality of our world today is that there is no stopping that. I guess it would be an interesting conversation around the state of IP and having all these small players not having any actionable IP with the inevitable cloning benefiting the consumer in the end. Not trying to be a PITA, in fact I would buy one of these in a minute and probably will.

 

[JohnYang1997]

The reality of our world today is that there is no stopping that. I guess it would be an interesting conversation around the state of IP and having all these small players not having any actionable IP with the inevitable cloning benefiting the consumer in the end. Not trying to be a PITA, in fact I would buy one of these in a minute and probably will.

They can copy the circuit or learn the circuit. But they can't copy the layout. There is saying that layout is the circuit. It's 100% true. More over with the same circuit different layout. Performance can differ from -110dB to -127db with 8ohm load, from -120dB to -130dB with 32ohm load. Also there isn't much about new topology or novel designs. Most if not all are documented decades ago. I really admire those old pioneers doing the research. It's really really hard to do research when nobody else have gone into it.

 

[scott wurcer]

They can copy the circuit or learn the circuit. But they can't copy the layout.

It would be fatuous to say the exact layout to the mm determines all the performance, the only cases of of copyright or maskwork that I have seen go forward are literally stolen artwork.

 

[JohnYang1997]

It would be fatuous to say the exact layout to the mm determines all the performance, the only cases of of copyright or maskwork that I have seen go forward are literally stolen artwork.

Ground layout is most important, feedback sensing, power rail distribution, compensation components, even trace width trace shape can lead to big difference.

 

[JohnYang1997]

Fiio k5 pro also has opa1642 around tpa6120a2. What's the performance of that thing?

 

[scott wurcer]

Ground layout is most important, feedback sensing, power rail distribution, compensation components, even trace width trace shape can lead to big difference.

There's no one unique solution. By definition if you discover a single unique way of doing something and there is no other possibility you patent it. It's a frequent misconception copyright covers an exact copy it covers no ideas embodied. That is to say routing power supply lines a certain way under copyright means as drawn, moving them even a little another way for the same result has no protection.

 

[JohnYang1997]

There's no one unique solution. By definition if you discover a single unique way of doing something and there is no other possibility you patent it. It's a frequent misconception copyright covers an exact copy it covers no ideas embodied. That is to say routing power supply lines a certain way under copyright means as drawn, moving them even a little another way for the same result has no protection.

I wasn't talking from copyright standpoint. I was talking about how much copying affects performance hence the company who designed it. Sure it's bad that someone copies. I mean for people who bother to pay attention to ground wiring, ground impedance and applying feedback to it, they can already design something that good. If someone "needs" to copy they will certainly omit the routing.

 

[armani006]

Is it possible to use 2 Pcs. of this amps + balanced DAC and get true balanced amplified output? I mean connect Left channel of Headphone to one amp and Right to another..

 

[JohnYang1997]

Is it possible to use 2 Pcs. of this amps + balanced DAC and get true balanced amplified output? I mean connect Left channel of Headphone to one amp and Right to another..

Theoretically yes you can. Practically, it's very difficult to adjust two volume knobs. If you use digital volume control then yes by all means.