|
Post by imstimpy on Mar 4, 2014 15:30:17 GMT
Somehow I managed to find NwAvGuy's blog. I stumbled upon him while attempting to learn about the virtual grounds in the β22 amplifier. If you have never heard of NwAvGuy, well, he reminds me of Solderdude's no bullshit approach to circuit design and blind testing methodology. NwAvGuy set about trying to design a superior headphone amp and DAC through careful evaluation of instrumented tests. Have a look at his blog, if you haven't before, as it is quite interesting. On his blog, he discusses AMB's Mini3 amplifier and why it is incorrectly designed. Quite interesting, but he does not instill a lot of confidence in the β22. He goes on to explain virtual grounds in a separate post. In it, he continues to discredit any merit to virtual grounds and the recommended 3-channel β22 configuration. Most of the information I've found on virtual grounds, as they apply to headphone amplifiers, is either from AMB, and Tangentsoft. If the β22 is supposed to be so good, why aren't many other designs using virtual grounds? Part of my point here was to introduce NwAvGuy's blog, and part was to expand a discussion to virtual grounds (if there is anything to discuss). Please forgive my relative inexperience with circuit analysis, but I am trying to learn
|
|
solderdude
Administrator
measureutternutter
Posts: 4,882
|
Post by solderdude on Mar 4, 2014 19:50:18 GMT
Harold (NwAvGuy) is partly right.
A reason to use a virtual ground is a very simple one... only having a single power supply voltage available for reason x.
There are a few ways to make an amplifier. The more common and easiest way is to use a symmetrical (or dual) power supply. For (mostly practical) reasons one can also design amplifiers running of a single power supply. 1: transformer coupled amplifiers. 2: capacitor coupled amplifiers 3: balanced amplifiers 4: virtual ground amplifiers (which is a form of balanced amplifier)
1: has the disadvantage of the transformer which is limited in power and bandwidth and has electrical issues (distortion) 2: has disadvantages such as needing a 'mute' circuit(ry) during start-up and needing relatively big (in size and capacitance) components that also add a very small amount of distortion. They roll-off in the lower extremes of the frequency range but in a different way than transformers. No roll-off in the highs though. 3: balanced amplifiers can be used with speakers (car radios, class/D/T amps) but need 4 wire and 4-pin plug headphones. No output capacitors or transformers are needed. 4: virtual ground amplifiers have the benefits of being usable for normal headphones (3 pin plug) and lacking coupling caps and transformers. In very rare cases the 3rd channel is used to apply a 'correction' signal or create a negative output resistance.
Harold does have a point that these 3rd channels are inferior to a real ground.
A virtual ground (= 3rd channel) is simply an extra amplifier channel that is only supplying a DC level at the same DC level as the 2 audio channels. Because these DC voltages cancel each other out there is no DC on a driver without having to resort to dual power supplies, transformers or coupling caps. It is an amplifier though which means it has an output resistance, creates distortion, can fail (and cause DC on the drivers), has a limited bandwidth. Another point is that the return signals of BOTH drivers combine at that point. Sometimes the 3rd channel is used as a 'gimmick' to hide the real reason (single power supply) and all kinds of audiophile nonsense is spoken to convince people it 'sounds' better.
It is MUCH easier to use a dual power supply and probably cheaper as well.
Sometimes (as in the case of 9V battery fed devices) a virtual ground is a cheap and easy way to make an amplifier though.
The AMB β22 is a dual power supply design and doesn't need a 3rd channel. Sometimes people like to build symmetrical (balanced) amplifiers with this design. I can see very little being wrong with the β22 it is a very symmetrical and well thought out design. I think the 'recommendation' by AMB to use a virtual ground is just a way of selling more. The 'reason' he gives being the return current NOT flowing into the ground but being directed to the power supplies is nonsense as it will still go into the ground plane via the PS caps only at another point. It is similar as using a return wire that is connected to the PCB at the power supply caps. It will only increase the resistance of the return path. That will be an academical difference though that won't be audible. Strangely enough he didn't do any measurements with a 2 board amplifier in which he could 'prove' his theory, this is a bit suspect.
The M3 can very easily be converted to be used with a dual power supply though (have some instructions for it somewhere)
The mini 3 uses a buffered rail splitter. This has some limits, it would be easier to simply use a rail splitter with a huge amount of capacitance (as I intend to use in the portable) as this increases the maximum output current. Also not ideal but better than the mini-3 solution.
Have build and tested amps with both topologies.
|
|
Rabbit
Administrator
Posts: 7,087
|
Post by Rabbit on Mar 4, 2014 20:10:21 GMT
It would be nice to see him return to his blog and even post on forums but his disappearance was perhaps due to a lot of nonsense from people who disagreed and made things difficult for him unfortunately.
|
|
|
Post by imstimpy on Mar 4, 2014 21:51:31 GMT
Thanks Frans, that is very helpful.
I found NwAvGuy's blog yesterday and have read nearly every entry. It is a shame he has disappeared.
|
|
solderdude
Administrator
measureutternutter
Posts: 4,882
|
Post by solderdude on Mar 5, 2014 9:28:46 GMT
I agree with the biggest part of what he writes but a few things he appears to have misunderstood/not looked into carefull enough. Instead he simply echoed research by others without investigating those results closer. Most people hated his attitude (his firm standpoint/belief) where others liked/loved him for it. Forums like 'Hydrogen' are close to his 'religion' For some digital myths see archimago's blog Boy oh boy, do some 'computer audio experts' hate this guy. Sometimes his conclusions can be a bit 'off' though, like with NwAvGuy as well as my opinions. In the end it all depends on how one feels about technology. Can you measure every aspect, can you measure all the relevant aspects, are ears flawed (in that the brain is easily fooled) or are the ears/brain MUCH more capable than all current technology measurements out there. A choice one makes based on their observations/experience.
|
|
|
Post by imstimpy on Mar 5, 2014 15:45:22 GMT
I do frequent archimago's blog from time to time. There is a lot of good stuff there but I usually get lost in the details of archimago, unfortunately, and find myself skimming.
As with many things in life, I don't feel there is an absolute in audio. In NwAvGuy's defense, yes, you can measure something to perfection. In all those objectivists' defense, perfection may not be what we want to hear. "Too revealing" would be one such statement that throws the wrench into perfection's gears.
I like to to see and attempt to understand the science behind things, then experiment with what I can get my hands on for the sake of experiencing as much as possible. My inquisition for the virtual grounds is so I can build a β22 eventually, but it also lead into a discussion on signal return paths and how they might relate to a dual power supply on my Panda. I've been reading as much as possible on the Sunrise/Horizon projects and fully intend to build one as well. I have Amazon pulled up with a search for Douglas Self's books so I can find some additional reading.
|
|
solderdude
Administrator
measureutternutter
Posts: 4,882
|
Post by solderdude on Mar 5, 2014 16:00:22 GMT
The science behind audio (the electrical part) is just electronics. The voodoo part is something that has never been proven in well performed blind tests. Some voodoo appears to be proven in blind tests but there are blind tests and blind tests.
The returnpath discussion is rather simple. It needs to be as low resistance as possible. This means an un-interrupted ground plane if possible and careful use of it (and how to connect). There is no voodoo there, just knowledge and experience. It is quite easy to f-up a groundplane and also quite easy to f-up a layout with wiring which can make or break a design and its performance. This one cannot see from the schematic and is very hard to simulate with 'standard' programs. If groundplanes are out of the question star ground is an option.
Creating a good PCB and do the wiring to it properly is something one needs to build from experience. So is proper coupling and decoupling.
There are loads of coarses for designing PCB's and certain aspects as decoupling etc. I took a few of those (professionally) which help in designing and building things.
|
|