Indeed - the reason is the present sketch is far too busy to send the data, timed correctly, one character at a time, as the RFM12B demands. The result - if you try - is the PLL loses lock and the results are anyone’s guess.
You need to search out the old, much harder to set up and less accurate “discrete sample” version. I’m struggling at present because my usual computer died, and the old one I’m using predates my involvement with OEM, so I haven’t got access to the old sketches. They are probably still buried somewhere in Github, but I haven’t a clue how to dig them out.
Try this. It is NOT for the emonTx Shield, but the emonTx itself, so you might need to change some I/O pins.
If you’re struggling, ask again because by then, I might have my system back up and running. As it is, I haven’t even got my master crib list of all the I/O for the various versions of hardware.