As background information, you might want to read through this detailed description of similar issues we had on cortex (from 2014). Although the V5 is very different, fundamentally we are dealing with the same issue. We have three wire ports that when the brain is first powered are in a safe un-configured condition with everything set to be a digital input, this can cause some solenoids to activate and some not. Because we never know how a robot may be changed between running different programs, we also reset everything to the digital input state when user programs are not running. This is the default unless you have used the dashboard to manually change them for testing.