for real value the offset must test with no music ie before the relay in this case
Why, would you expect a different value? How come?
FWIW, I also did a little experiment. I took a music track, I inverted the waveform in a wave editor and combined the two, resulting in a track of total silence. I played the track, the relays reacted as if it was some real music to be played and, upon measuring the DC offset, I've got exactly the same values as when real music was playing, namely 84mV left and 4.05V right.
There is nothing to be gained by this exercise. If a unit produces a large enough DC offset to be an issue you will hear it. If you don't hear it, it's not a problem. Simple.
Sorry, but I have to disagree. I think it's not so simple. Even if you don't hear a detrimental effect, you could have at least several problems:
1. The dac and the downstream equipment (at least until the signal meets a cap, if there is any to be met) are not working within the intended parameters. It is my understanding that some active components such as transistors will have a worse performance under these conditions - higher distortion and I suppose their thermal behavior might be different too, which I assume could also reduce their life expectancy. As a rather extreme example, a diy tube amp I have heard in my system, that was working perfectly fine with other sources, misbehaved badly when connected to my Gumby - noise, bass distortion, one channel dropped in volume after a few minutes etc. I don't know if it was the DC offset causing this, but I fail to see what else could have been (granted, the preamp might have had a problem as well). This potential for creating trouble was not apparent when I used the dac with other equipment (apart from the already mentioned pops on the right channel).
2. If the DC is amplified in a DC coupled amplified and passed further to the loudspeakers or headphones, the resting position of the cone will be moved from the center of it's excursion range - in other words, it will permanently be "pushed in" or "pulled out" a bit. This will result in uneven forces during the cone movement, as well as not only lower excursion to one side of it's real, correct center position but also higher excursion to the other side - and higher excursion correlates with higher distortion.
3. I've been told that up to 50mV DC offset should be certainly safe for my ex headphones AKG K1000. One can assume therefore that higher than 50mV might not be 100% safe. I had 84mV on the left channel which would be further amplified before reaching a phone such as the K1000, so I would definitely exceed the stated safe zone. My point is that this would happen in spite of the fact that I have not heard anything obviously wrong, such as pops or static, on the left channel and if not for the much higher offset on the right I wouldn't have known something is wrong.
The above sound and reliability issues might be or might be not noticeable at the moment, but might become audible or cause trouble in the future, for example when upgrading partnering equipment (more revealing speakers could make amp distortion more obvious, a DC coupled amp instead of the current AC couple one would start passing the offset to the speakers and so on), if the life of the partnering equipment has been shortened because of operating out of parameters (and you won't even have a clue about it) or when comparing your DC offset afflicted Gumby to a "healthy" one etc. Not to mention a possible drop in resale value.
Given all the above, I think it's reasonable to expect a perfectly functioning unit if I payed perfectly usable money for it, without any "if" or "maybe" regarding sound and reliability, and the same is true for all of you fellow Gumby owners. So, even if you are not convinced or wouldn't bother for your own sake, would it be too much from me to ask you: please, measure and report in order to help me?!
Actually, if this is not an isolated phenomenon, you would also help Schiit, other potential buyers and other current owners too...