No, it’s not “just you” - the two programs work in very different ways and are suitable for different circumstances.
Dimension 4 is a pretty simple application which just does a periodic clock synchronisation to a time server. It delivers a quick fix based on the assumption that the network time is always better than the local clock. However this makes it very intolerant of a time server that gives the wrong answer, or if the synchronisation is disturbed by long or asymmetric network latency.
The Meinberg program works very differently. It is a port of the original NTP daemon that the NTP protocol was designed to support. It is a long running process which aims to establish a mesh of timeservers which reach a consensus. It aims to transfer not only the time, but to establish error bounds on it. It peers with multiple servers in parallel and there is code to disregard servers that are clearly outliers (so-called falsetickers). When setting the machine’s local clock, it is explicitly designed to avoid sudden jumps - instead it adjusts the rate at which the local clock ticks to bring it gradually into line. It can even carry on doing this in the absence of network connectivity, based on the last known measurement of the native local clock rate.
It follows from this that the Meinberg program is superior for a long running machine. For a server running 24x7 it wins hands down. It works pretty well on a home workstation which may run for several hours at a time.
It is less useful for something like a tablet that frequently goes to sleep and runs very intermittently to minimise battery use. The algorithms rely on the process being scheduled to run regularly. It quite deliberately works rather slowly, so it is not going to solve the problem of taking a tablet out of a bag on a summit and getting on the air straight away with the time spot on.