I just assembled a DCC++ setup using a 3rd party Arduino Uno clone and an official Arduino Motor Shield to operate as a standalone programmer connected to JMRI. I notice that when I put a sound-equipped locomotive on the programming track, its sounds start up. This implies to me that the programming track is not informing the locomotive that it should be in "service mode" as defined in section 9.2.3 of the NMRA standards. While this didn't seem to affect the ability to read or write CVs from the decoder, I don't believe this behavior is "correct". Additionally, in the case of the particular locomotive I was testing with an ESU LokSound Select decoder, the engine was in a loop of playing the prime mover start / prime mover stop sound sequence (which was quite annoying), as if someone was continually toggling on and off function 8.
