Just to add a data point to the discussion, I tried to reproduce Mark's observations in my system, because it is similar (though not identical) to his and I never had those problems. (edit: I assumed from the thread title that this is about ALIX/MPD, but upon re-reading the posts above I realize that this is rather not the case.) Result is that my DAC works perfectly fine and I cannot reproduce his findings. Also, power cycling the DAC for like two minutes has always solved any problem (in the rare event that there was any) and I never had to disconnect the USB cable. Bottom line is that your problem may be either related to a different firmware as you suspected, or to your Alix/MPD PC setup. The second hypothesis you could test by repating your experiments from a different source (other mpd server or PC/Mac).
For reference, I documented my findings. Sorry for the long text, but perhaps it is useful for other people to check against my exact procedures.
My Bottlehead DAC is connected via USB (Supra cable) to a Cubox-i4Pro running Voyage MPD version 0.9.5 with real time Kernel optimizations as described on Punky's Website. I use Cantata as my mpd-client. Actually the version of Voyage MPD compiled for Cubox is called "Voyage Mubox", but as I understand it it is basically comparable to Voyage MPD on ALIX (though the differences I observe my be because of that).
DAC is always on "USB" and is and has never been changed to/tested with other source types. Three experiments follow:
1. Switch on Cubox, wait until it has booted, switch on BH-DAC (display briefly shows "44.1", for less than a second, and then shows "384"), put 96kHz file on playlist, hit play (display shows "96"), listen (especially for pitch and speed, which would later reveal if the track is played with the different/wrong sampling rate).
2. Switch of BH-DAC, wait several minutes, clear playlist and put three files on it (first 44kHz track, second 192kHz track, third the 96kHz track from the first experiment), switch DAC back on (display again briefly shows "44.1", for less than a second, and then shows "384"), play 44kHz track (display shows "44"; sound seems OK), play 192kHz track (display shows "192"; sound seems OK), play 96kHz track (display "96", sound OK and seems same pitch and speed as in first experiment).
3. Stop music, switch DAC off briefly (i.e. for one or two seconds), switch DAC on again, play track from experiment no.1 (display "96", sound OK), switch off/on briefly as before, repeat track order from experiment no.2 (all OK, same behavior as in experiment no.2).
There are two BUTs however: 1. Playing around some more, with an additional file from the 48kHz family (actually a 48kHz track), I ONCE could get a problem, which unfortunately I cannot remeber how I produced it and thus cannot reproduce: Switching the DAC off and on briefly, the display showed "352" (and not "384") aftewards and all files from the 44kHz family (44, 96, 192) played slow. Switching the DAC off for a few minutes and then back on cured this problem and it behaved fine afterwards (this is the proper reset procedure for the DAC that John described somewhere here on the forum). This only happened one other time in two years of use and is not a normal error, it has to be forced by playing around a lot. 2. I remember that some time ago occasionally I had problems regarding the correct recognition of the DAC by Voyage MPD, but it always worked and works if I switch on the DAC after the Cubox has booted, which is now my normal routine. If I have to reboot the Cubox while the DAC is on, I usually leave the DAC on and it is usually correctly recognized afterwards. On the very rare occasion that a problem occurs, I switch off the DAC for like 2 minutes and on again, and everything is fine.
