@Alcophone, do you notice in any way while listening that this amp in not fully balanced? I believe the H20 is fully balanced, and the Jotunheim that you also have if I recall correctly is known to be fully balanced. The creator of Yggdrasil adds one bit to its resolution for the fact that it's balanced (four 20-bit DAC chips result in 21-bit). I suppose we lose that extra bit of resolution when listening to Yggdrasil through the THX. I suppose it's not audible but I wonder if it's felt in any way.
Before I say anything, I should mention that I'm not an electrical engineer, just an enthusiast like you trying to wrap my head around certain concepts. Happy to get corrected where needed.
Merely using two n-bit DACs per channel does not result in n+1 bits of resolution if you are completely symmetrical, i.e. you feed both DAC chips inverse values, or the same value but with negative reference voltage on one chip (if that's possible from an engineering point of view, I don't know). It just doubles the difference between the effective signal levels, but you're not suddenly doubling the amount of distinct levels you can generate.
What you could do is change the lowest bit in what you send to one of the DACs. That should basically double the number of levels you can achieve as the difference between the levels of the two DAC chips. The resulting slight asymmetry should not matter.
If that is indeed what Schiit does in the Yggdrasil and Gungnir Multibit, then you would only lose that extra bit by completely throwing away one of the two signals per channel coming in via XLR. But even in single ended DACs with XLR input for convenience only there should be something that combines the two signals, converting them to single ended while retaining the extra resolution (much like what Yggy's single ended output should do) and properly rejecting transmission noise.
Anyway,
on Massdrop, Andrew said this:
AndrewMason said:
Errata: The signal path internally is fully differential throughout (per Putzeys), but is NOT truly balanced in a few sections. We preserve the full signal integrity but we avoid increasing the retail price for no measurable benefit.
So that extra bit should not get lost. Even if it were, I don't know that I would be able to hear that - certainly only with very select recordings at high volume.
In another post, Andrew mentioned this article:
https://www.edn.com/design/consumer/4429968/2/The-G-word--How-to-get-your-audio-off-the-ground
The way I understand it is that you can be differential by transmitting one variable signal with the actual audio, and a fixed zero level signal in addition. Distortion along the way affects both roughly equally, so if you subtract one from the other (differential), the error still gets cancelled out.
That makes sense to me when merely sending a signal between components, but less so in an amplifier. If the THX amp does this internally, it would have to first combine the balanced signal coming in to one variable signal and one zero signal, and then carry them through. The supposed benefit is a reduction of expensive components, so this way only the variable signal would need to get amplified, I suppose (instead of having two amplifiers per channel, which you need to be fully balanced). But that would also mean that any transmission interference would get amplified in one of the two signals, almost completely ruining the benefits of the differential approach. But maybe the amp correction magic is applied here. Or a less powerful amplifier can be used for the zero signal since the amplitude of any induced noise should remain low. I would love for Andrew to shed more light on this.
