[carlmart]

I'm talking more in the "3 channel active ground" kind of implementation (like the Pimeta https://tangentsoft.net/audio/pimeta2 ) rather than just using an opamp or similar just to set a virtual ground reference though there are a lot of drawbacks especially in terms of noise to that approach but the AD8066 seems to work really well in terms of producing something that is pleasing to listen to even if from an engineering perspective there are a few negatives to the amp design itself.

The Pimeta does implement a virtual ground with an opamp, the TLE2426, only more compactly packed. For headphone drivers it's a simpler powering system, certainly better than using two batteries in portables, using the mid junction as ground. Those batteries may discharge differently and cause power unbalances. The other way is to use a DC-DC supply to generate the +/- voltages. Using the TLE is a simpler way to do that that may work on such circumstances.

I built an amp (somewhat along the lines of the PBA https://www.head-fi.org/threads/pseudo-balanced-amplifier-pba.553236 ) that uses one AD8066 per audio channel, with a fully isolated ground per channel, with the headphone drivers sitting across the outputs of the two amps in the package and it takes the AD8066 to another level again.

The additional circuit for the ground amplifiers allows more current to be used, which is the limitation of the TLE and essential for hungry headphones.

For a very good representation of what active ground means and why would this be better than regular 2-channels headamps, Dr. Meier Corda has a dedicated webpage here: http://www.meier-audio.homepage.t-online.de/grounds.htm.

Balancing a signal can be interesting, particularly if the grounding can't be kept absolutely zero. You could use a balancing chip to feed the separate drivers with more isolation, or a transformer. But of course, you would need to recable the headphones. But most quality headphones that I know of already have two-wires for each driver, that only get together on the connector. There wouldn't be any-rewiring.

 

[raoultrifan]

I've measured the ground-wire for most of my cans and I found a higher resistance that the hot-wires, which kinda sucks to me. I immediately understood that re-wiring is not snake-oil for some cans (AKG K701, Beyerdynamic DT880 a.s.o.).

On the 3-channel amplifiers there's no need to do any rewiring, as long as a good cable is used.

 

[carlmart]

I think it's a worth a try and listening, which in the end seems to me as the best possible test. Many electronics engineers will probably disagree.

 

[raoultrifan]

I'm sure all EE's will agree that active ground might be better on most circumstances, I'm confident on PIMETA's implementation and also on what Dr. Merier Corda has to say. In my above post I said that ground-wire on some of my cans are actually having a higher resistance (about 1 Ohm) than the hot-wires (right/left channels); this might translate into possible crosstalk issue on >10KHz frequencies or even EMI/RFI/radio-waves getting injected into the headphones, especially when speaking about high impedance cans and more than 3 meters long cables. I really think that going balanced or active-ground driving might help under this scenario, in compare with the single-ended.

 

[Rroff]

The Pimeta does implement a virtual ground with an opamp, the TLE2426, only more compactly packed. For headphone drivers it's a simpler powering system, certainly better than using two batteries in portables, using the mid junction as ground. Those batteries may discharge differently and cause power unbalances. The other way is to use a DC-DC supply to generate the +/- voltages. Using the TLE is a simpler way to do that that may work on such circumstances.




The additional circuit for the ground amplifiers allows more current to be used, which is the limitation of the TLE and essential for hungry headphones.



Balancing a signal can be interesting, particularly if the grounding can't be kept absolutely zero. You could use a balancing chip to feed the separate drivers with more isolation, or a transformer. But of course, you would need to recable the headphones. But most quality headphones that I know of already have two-wires for each driver, that only get together on the connector. There wouldn't be any-rewiring.

What I'm talking about isn't so much the virtual ground part of the equation itself though - but the full 3-4 channel active topologies - you can also drive these setups using a charge pump or other DC-DC POL with dual outputs or even from an AC transformer to get dual supplies it isn't the relevant part of what I'm talking about the circumstances where the AD8066 really seems to take off - for some reason where you have that setup with an AD8066 output one positive one negative on either side of the driver it really opens up another dimension of that opamp in audio use - especially any hint of veiling is almost if not completely lifted.

 

[raoultrifan]

I was referring to exactly the same scenario of 3-4 amplifiers and not to the rails splitting/virtual ground. I do hope all of us were referring to the same thing. If not, than Objective2's designer has a good article about rails split/virtual ground (eg.: making 2 rails out of 1 x 9V battery).

 

[Ivan TT]

full 3-4 channel active topologies

I recently implemented 4 channel active ground design in my Zishan DSD, tapping in 2.5mm 4 pole output and using OPA1622 configured as buffers providing separate buffered grounds to L and R channel outputs.
I notice better hanging of bass: higher resolution and each note's envelope clearly obvious; and improvement in soundstage due to fuller mids (as opposed to improvement in soundstage due to better/more resolved highs).
As OPA1622 has 5Ohm output impedance SOME BA IEMs don't play well though, when compared to SE output.
Overall buffered ground in 4 channel configuration was definitely an improvement, recommended.

 

[raoultrifan]

I'm sure 32 Ohms cans should play comfortable with the 5 Ohms output impedance.

Not sure this is the right DIY thread for this, but some pics with your design and work might be interesting, if you have the time for...of course.

 

[carlmart]

What options should I consider to swap the D50 output opamps?

The factory ones are OPA1612, which many report as metallic sounding, some people here too.

Any remedies you could find to tame the 1612s or only changing them?

 

[raoultrifan]

I wouldn't touch the OPA1612 opamps, these opamps are very good (proved by measurements).
Are we speaking about Topping D50, right? Have you seen D50's measurements on the Internet? This DAC is simply perfect, you just can't find a better one at this price level.

 

[carlmart]

There have been some complaints that the Topping D50 sound is a bit metallic, which gets a little better with a better power supply.

The idea is to make the DAC sound better than its price level. For that you have to go DIY, and work basically on two areas: the analog output stage and the power supply.

There's no space inside to do anything, so the first thing to provide is a new top cover. Then you can place the regulators for the different power supplies outside the D50, as well as the new alternative analog output stages.

Another option is to place the whole D50 box into another, larger, metallic box. Remove the top, leave it open, and run wires outside, where regulators and analog stages will be.

Many people will find my proposal crazy or stupid, or follow the motto if it ain't broke, don't fix it. Well, I'm a diyer, and I have seen how things can be improved with not too much money and some effort.

One important thing to consider with the D50 is that everything is small and fragile, particularly pcb tracks. That's why you must, IMHO, do everything outside and restrict soldering on the D50 to a minimum.

About the D50 measurements, I don't think they tell the whole story, particularly about the sound qualities. I do think both measurements and listenings should be weighted for evaluation.

 

[LaughMoreDaily]

You could also try 2xAD797 with single-dual adapters, they’re the less noisiest from this group.

The 797 is a dual opamp, yes? What happens when you put two dual opamps on single dual adapter? A single meaning, one connection to the PCB right?

 

[LaughMoreDaily]

I wouldn't touch the OPA1612 opamps, these opamps are very good (proved by measurements).

That sounds a bit weird. You say you wouldn't touch a 1612 and then you say it's a very good opamp.

 

[bunkbail]

There have been some complaints that the Topping D50 sound is a bit metallic, which gets a little better with a better power supply.

The idea is to make the DAC sound better than its price level. For that you have to go DIY, and work basically on two areas: the analog output stage and the power supply.

There's no space inside to do anything, so the first thing to provide is a new top cover. Then you can place the regulators for the different power supplies outside the D50, as well as the new alternative analog output stages.

Another option is to place the whole D50 box into another, larger, metallic box. Remove the top, leave it open, and run wires outside, where regulators and analog stages will be.

Many people will find my proposal crazy or stupid, or follow the motto if it ain't broke, don't fix it. Well, I'm a diyer, and I have seen how things can be improved with not too much money and some effort.

One important thing to consider with the D50 is that everything is small and fragile, particularly pcb tracks. That's why you must, IMHO, do everything outside and restrict soldering on the D50 to a minimum.

About the D50 measurements, I don't think they tell the whole story, particularly about the sound qualities. I do think both measurements and listenings should be weighted for evaluation.

I don't have the D50, but I have a $100 Khadas tone board and it sounds perfect to me, not metallic sounding at all. And it doesn't use a known off-the-shelf opamps, they are custom-made. Measurement wise, it is at least on par or better than the D50 and it speaks volume about the level of engineering put into it as it uses a single ES9038Q2M in stereo mode (D50 has them in dual mono config).

 

[wolkegeist]

I have a question,
Can I bridge 2 cmoy to create a balance setup, with each cmoy for L and R signal? and how can I bridge them if I could?